S016_SybilA
Christina Rene Hendricks (Knoxville, Tennessee), Known as Christina Hendricks, is an american actress, known better for her role as Joan Holloway in the show AMC Mad Men, and as Saffron in the show Fox Firefly.
chr1sth3ndx
23 year old Canadian actress.
S015_NatAlynLind
Trained using the base 1.5 ema only and GrapeLikeDreamFruit
kgerber
A commission of @Zorglub.
Rebecca Romijn is an American actress and former model, known for her roles in the X-Men film series and the Star Trek TV series, among others.
This is a 1200-step TI trained on a dataset of 18 images with my usual settings.
Curious about my work process? I have summarized it here.
Appreciate my work? My TIs are free, but you can always buy me a coffee. :)
rebromijn
follow me on Instagram, Youtube, Patreon and my website.
Suggested Weight: 1.0
848y81d3n in orange jumpsuit, detailed face, sharp focus, prison cell
Tried my best to bake in "methmouth" but it didn't take. Hopefully in an upcoming update, or maybe a separate embedding just for that version.
trained on ~13 images
Learning Rate: 0.005:1000, 0.001:2000, 0.0001:5000, 0.00005
Total Steps: 15,000
Batch Size: 1
Gradient Steps: 1
848y81d3n
Follow me to make sure you see new styles, poses and Nobodys when I post them. Things move fast on this site, it's easy to miss.
Most of the recent, good, training has been using anime and models trained on asian people and it's culture. Nothing wrong with that, it's great that the community and fine-tunning continues to grow. But those models are mixed in with almost everything now, sometimes it might be difficult to get results that don't have asian or anime influence. This embedding aims to assist with that. It can even change anime characters (though that wasn't the intended purpose).
I first created this when trying to make preview images for my South of the Border Style embedding. I was trying to get south american people and culture, but there was a lot of asian culture leaking into the generated images. This embedding fixed that. So it doesn't specifically race-swap asian to white-caucasian, it just removes asian so your prompts can be more effective with whatever other race or culture you're trying to portray.
How to use:
Use the negative prompt primarily, then use the positive prompts only if you need the additional help. There are 3 files to download.
Asian-Less-Neg: Use this one, place in your negative prompt at strength of 1.0
Asian-Less: Only use this in the positive prompt if you need the extra strength. On an asian-specific model for example.
Asian-Less-Toon: Only use this in the positive prompt if you need to remove anime-like features in illustrations.
Again, this isn't meant to race-swap, but to help get other race and/or cultures in your image generation without the influence of the asian image training. But like all models, the intended use and what you do with it are separate matters.
Do you have requests? I've been putting in many more hours lately with this. That's my problem, not yours. But if you'd like to tip me, buy me a beer. Beer encourages me to ignore work and make AI models instead. Tip and make a request. I'll give it a shot if I can. Here at Ko-Fi
Asian-Less
Yuliya is a Russian supermodel, representing Russia at Miss World 2005
This is a textual inversion, for a LoRA of Yulia look HERE
[ SDv1.4, 4e-03, 4500, v/s .201 ]
yuliyaivan
Olga is a Russian model blowing up on Instagram and elsewhere
This is a textual inversion, for a LoRA of Olga look HERE
[ SDv1.4, 4e-03, 5500, v/s .200 ]
olgaabram
Bianca is an Italian actress and burgeoning Instagram celebrity.
This is a textual inversion, for a LoRA of Bianca look HERE
[ SDv1.4, 4e-03, 4500, v/s .200 ]
biancapanconi
S014_SonyaBlaze
See versioning below for 1.5
Summary :
Negative - General // Pruning most of "bad things" in the average standart
Not suitable for natural realistic portrait
Negative - ForAnalog // Heavy pruning of everything related to digital, painting, 3d, artwork and so on
Not suitable if you want painting and vivid colors
Negative - Contrast & scenic // Pruning of contrasts antagonists
not suitable for classical painting, minimalism or other movements of that kind
The pruning of contrast's antagonist may fix weird body / situations behaviour (sometimes) even if it don't include any body face fix at all
All plots (grids) was generated on v2-1_768-nonema-pruned 4711ff4dd2
If you are getting the error "Textual inversion embeddings skipped(3): rz-neg-general", it's because you are on a wrong StableDiffusion version (2-x required)
This negative embedding is suitable for everything, enhanced faces details and coherence, better colours, less blurry generations.unwanted paintings. More realistic overall.
If you wants generations more concept art oriented please emphasis down the negative
Samples :
A photo of a beautiful woman medium shot with long hairs
Negative prompt: rz-neg-general
Steps: 40, Sampler: Euler a, CFG scale: 8, Seed: 2001, Size: 768x768
A photo of an handsome man
Negative prompt: rz-neg-general
Steps: 40, Sampler: Euler a, CFG scale: 8, Seed: 2001, Size: 768x768
A beautiful woman wearing a cyberpunk outfit by Ilya Kuvshinov, krenz cushart, Greg Rutkowski, trending on pixiv, a beautiful highly detailed digital painting
Negative prompt: rz-neg-general
Steps: 25, Sampler: DPM++ SDE Karras, CFG scale: 8, Seed: 2001, Size: 768x768
A sophisticated restaurant room with large window with view to a midcentury cozy kitchen, well maintained, clean, medieval, fantasy genre, natural light, fantasy, natural light, concept art, by greg rutkowski and craig mullins, cozy atmospheric and cinematic lighting, trending on artstation
Negative prompt: rz-neg-general
Steps: 20, Sampler: DPM++ SDE Karras, CFG scale: 8, Seed: 2001, Size: 768x768
A picture of a classy retro car parked in paris, baroque, trending on artstation
Negative prompt: rz-neg-general
Steps: 35, Sampler: Euler a, CFG scale: 7, Seed: 2001, Size: 768x768
There is no body / face modification for the tokenizer, only a careful pruning.
(no vae loaded for thoses generation, raw 2.1 768)
A photo of a beautiful woman in paris, medium shot, detailed face, cinestill800t, analog
Negative prompt: rz-neg-foranalogportrait
Steps: 20, Sampler: DPM++ SDE Karras, CFG scale: 8, Seed: 3002, Size: 768x768
A photo portrait of an handsome man at the beach, medium shot,bokeh, detailed face, cinestill800t, analog
Negative prompt: rz-neg-foranalogportrait
Steps: 20, Sampler: DPM++ SDE Karras, CFG scale: 8, Seed: 3002, Size: 768x768
All plots (grids) was generated on Model hash: cc6cb27103, Model: v1-5-pruned-emaonly
There is no body / face modification for the tokenizer, only a careful pruning. Body and face modification tokens greatly reduce the overall analog feel, so take care if you need to add them and don't want to lose a lot of details.
An analog photo of a beautiful young woman portrait with a detailed face, cinestill800t, cute
Negative prompt: rz-neg-15-foranalog
Steps: 20, Sampler: DPM++ SDE Karras, CFG scale: 7, Seed: 75824, Size: 512x512, Model hash: cc6cb27103, Model: v1-5-pruned-emaonly
romerorz.art - Twitter - DeviantArt - Gumroad
rz-neg-general
While the built in embedding for Emma Stone is not terrible, I was curious on whether I could improve that.
I used 443 sample images, all cropped and tagged manually, mostly chosen from the top 1000 of the posts in her subreddit.
Emma Stone
Tiffani Amber Thiessen is an American actress. Her roles as Kelly Kapowski on NBC's Saved by the Bell (1989–1993) and its spin-off media, and as Valerie Malone on Fox's Beverly Hills, 90210 (1994–98) established her as a teen idol of the 1990s
TiffaniAmberThiessen
A commission of @Springbok707.
Roberta Vasquez is an actress, model and former policewoman. She was the Playmate of the Month in November 1984.
1200-step TI trained on a dataset of 18 images with my usual settings.
Curious about my work process? I have summarized it here.
Appreciate my work? My TIs are free, but you can always buy me a coffee. :)
rovasquez
30 year old pornstar.
If you wanna tip me >> buy me a fruit.
S013_MiaMalkova
follow me on Instagram, Youtube, Patreon and my website.
Suggested Weight: 0.7-1.0
realistic portrait photo of j0hnf3tt3rm4n taken with a hasselblad h6d-400c 1.4, bald
This is not an easy embedding to use, and takes lots of practice, trial and error
it is VERY easy to get a weird "shrink wrapped" look with the skin
Inpainting is almost necessary (every example image you see here was created with inpainting)
All that said, with the right prompts, model and inpainting it's very possible to create lots of variations
trained on ~12 images
Learning Rate: 0.005:1000, 0.001:2000, 0.0001:5000, 0.00005
Total Steps: 15,000
Batch Size: 1
Gradient Steps: 1
j0hnf3tt3rm4n
follow me on Instagram, Youtube, Patreon and my website.
Suggested Weight: 0.7-1.2
photo of young d381m4z4r with a-line bob haircut and makeup as a [something], detailed face, skin texture, sharp focus, light bokeh
most of the training was from recent images, so to get a younger version add "young" to positive prompts, and try adding "old" and "wrinkles" and etc to the negative prompt
Works very well for closeup portraits without inpainting
trained on ~24 images
Learning Rate: 0.005:1000, 0.001:2000, 0.0001:5000, 0.00005
Total Steps: 15,000
Batch Size: 1
Gradient Steps: 1
d381m4z4r
follow me on Instagram, Youtube, Patreon and my website.
Suggested Weight: 0.7-1.0
photo of c1e4:0.8 as batman, detailed face, skin texture
try adding "attractive" and "beautiful" and "woman" to the prompts
try adding "glamour photoshoot" and "celebrity makeup"
LOTS of inpainting needed for anything other than a headshot
trained on ~22 images
Learning Rate: 0.005:1000, 0.001:2000, 0.0001:5000, 0.00005
Total Steps: 8,950
Batch Size: 1
Gradient Steps: 1
c143duv411
Joy is an American model, actress, fashion designer and businesswoman. She is a former Sports Illustrated swimsuit model, Playboy Playmate, and Victoria’s Secret model. She has literally done it all
This is a textual inversion. If you'd rather have a LoRA of Joy please go HERE.
[ SDv1.4, 4e-03, 4700, .199 ]
joycorrigan
Sadie is an American actress and singer. She made her debut as the titular character of the Disney Channel Original Movie Kim Possible.
This is a textual inversion. If you'd rather have a LoRA of Sidney please go HERE.
sadiestanley
Olivia is an American artistic gymnast and social media personality. Geaux Tigers.
This is a textual inversion. If you'd rather have a LoRA of Olivia please go HERE.
oliviadunne
My version of a textual inversion of American film actress Adrianne Palicki. It seemed to lose her signature mole. I may try to get it back in v2.
APal69
27 year old Ukrainian nude model.
You can now buy me a fruit if you like my work.
S012_GloriaSol
Inspired by the work of distinguished Polish photographer Zofia Nasierowska, this model adds high key lighting, diffuse shadows, and dreamy soft focus to portrait photos. It has a tendency toward 1960s/1970s hair and makeup, and for the subjects to be looking off-camera. It likes black-and-white, selective color (red lips especially), or lifted black colour tones.
I'm grateful to BelieveDiffusion and Zovya for their character models that I used to produce the demo images for this model.
A note on the model name. Lodz is a city in Poland; the Polish pronunciation of the city's name is close to "woodge" in English, like adding the first part of the word "wood" to the second part of the word "fridge". The city is home to the legendary Lodz Film School, which has led to the city being nicknamed "Hollylodz".
Hollylodz_Ikon
Lea Seydoux is a french actor mostly known from her appearances in Mission Impossible and James Bond 007 Spectre. Also she appears in the video game Death Stranding.
Trigger word is leaseydoux_ti
v1 was trained with 13 images for 4100 steps
leaseydoux_ti
Keira Strong is a custom character embedding for generating consistent character.
Keira is a tough woman with short fiery red hair, perfect for your barbarian woman or strong woman generations.
This embedding is trained with dataset images that is generated by AI. She's not based on any existing person or character.
k31r4sr0
follow me on Instagram, Youtube, Patreon and my website.
Suggested Weight: 0.6-1.0
photo of c0183rt as a wizard wearing a wizard hat and holding a magic wand, light bokeh, intricate detail
Don't be scared to turn the weight all the way down (even lower than 0.6)
You will need to inpaint the face often
prone to noise/overexposure if you try adding too much detail in prompts
trained on ~24 images
Learning Rate: 0.005:1000, 0.001:2000, 0.0001:5000, 0.00005
Total Steps: 15,000
Batch Size: 1
Gradient Steps: 1
c0183rt
23 year old American pornstar.
S011_BlakeBlossom
A commission of @Alarmed_Leg_5793795.
Léa Elui is a French social media personality, with a huge fanbase in Instagram and Tiktok.
This TI has actually been sitting in my hard drive for quite a while. I created it about a month ago and didn't think it was good enough to upload to CivitAI, but, upon testing it again today, I think at the very least it's worth uploading as a 0.5 version. I may try to upgrade it in the future, though, as I think the TI would benefit from applying my most recent changes in the training process.
This was apparently just a 500-step TI, trained with my usual settings but with just 13 images in the dataset. That's what I'll probably work on during my next attempt.
Curious about my work process? I have summarized it here.
Appreciate my work? My TIs are free, but you can always buy me a coffee. :)
leaelui
Jessica is an Australian model who appeared in the Sports Illustrated Swimsuit Issue a bunch.
This is a textual inversion. If you'd rather have a LoRA of Jessicaplease go HERE.
[ SDv1.4, 4e-03, 4100, .201 ]
jessicagomes
Brooke is famous on the internets in suck places and Instagram and the TikTok.
This is a textual inversion. If you'd rather have a LoRA of Brooke please go HERE.
[ SDv1.4, 4e-03, 4300, .200 ]
brookemonk
Berta is a Catalan Spanish actress and model known for her participation as Nuria Vega Valverde in the series Bajo sospecha and Carolina Solozábal in the series El secreto de Puente Viejo
This is a textual inversion. If you'd rather have a LoRA of Berta please go HERE.
[ SDv1.4, 4e-03, 5000, .199 ]
bertacastane
This is a textual inversion trained on pictures of Philomena Cunk, a character played by Diane Morgan.
Still learning and improving the progress of the training, any feedback is welcome.
I don't know why but it seems that she sometimes looks a bit like Aubrey Plaza but that doesn't bother me that much ;
Feel free to leave me a tip if you like what I am doing :)
phil0cunk01
Starting out acting on the teen drama "The V.A.E", Bilphanie got her big break starring in the music video for the Gans 'N Posers song "Movember Pain", where she also met her husband Gavid Paltrose. A modelling career soon followed, and she became bigger than Cindy & Kate combined. After deciding to divorce Gavid, she declared "anyone can do music", and put her money where her mouth was by quitting modelling and forming an all-girl punk band. The subsequent music video for their diss-track "Horsey", featuring a Gavid-lookalike fellating a stallion, cemented her band's reputation as the most dangerous women working in punk-rock.
As a minor side-note ... Bilphanie doesn't exist, and is an AI OC, a character not based on any living person, in personality or looks. Bilphanie may LOOK like somebody you know, but she's a Nobody.
She's happy to be your punk-rock heroine, your successful fashion model, or your (cute) girl next door, when you need a consistent character not based on any one living person. She works in every model I've tested (which is quite a few), likes to keep her hair long, and has been created using a different workflow to my previous two TIs, so should be even more flexible than they are.
Inspired by Zovya's Nobody series. Be sure to check out other contributions to the Nobody tag.
bilphaniem
Anna Hopkins is a Canadian actress from Montreal, best known for her performance in series such as The Expanse, Arrow and Shadowhunters.
The embed has been trained on SD 1.5 and works well on most 1.5 based models. For this embedding I recommend prompting at a slightly reduced weight (for example: Koh_AnnaHopkins:0.95), though this is not necessary for every model.
If you enjoy my creations or have a face you want to commission, consider supporting me on Ko-Fi :)
Koh_AnnaHopkins
27 year old American pornstar.
S010_AlexGrey
A commission of @Zorglub.
Melissa O'Neil is a Canadian actress and singer of Chinese and Irish descent. She's mainly known for her roles in TV shows such as Dark Matter or The Rookie.
This is a 1600-step TI trained on a dataset of 18 images with my usual settings.
Curious about my work process? I have summarized it here.
Appreciate my work? My TIs are free, but you can always buy me a coffee. :)
mel0neil
This is a textual inversion trained on pictures of political commentator Ana Kasparian.
Still learning and improving the progress of the training, any feedback is welcome. Overall I am pretty happy with the way it came out.
I suggest you add something like "top-bun" to your prompts as I think she looks funny with open and flowing hair but that might just be me.
Feel free to leave me a tip if you like what I am doing :)
AnaKa5V1
A request of @shrikeblackthorn764.
Sofia Vergara is a Colombian actress and model of worldwide fame.
1000-step TI trained on a dataset of 25 images with my usual settings.
Curious about my work process? I have summarized it here.
Appreciate my work? My TIs are free, but you can always buy me a coffee. :)
s0f1av3rgara
brecbella-700-1100
A request of @KingPungy.
Paige Woolen is an American social media personality.
800-step TI trained on a dataset of 20 images with my usual settings. I had initially decided to upload step 500, but I think she looked more bimbo-ish than she really is. Not that there's no trace of 'bimboness' by step 800, though, as the example pics evidence... In any case, hope you'll like this and please let me know what you think!
Curious about my work process? I have summarized it here.
Appreciate my work? My TIs are free, but you can always buy me a coffee. :)
pwool3n
Actress Bella Ramsey. Stars as Ellie in the HBO drama series The Last of Us.
bellaramsey
Dutch-Frisian photomodel and ex-Victoria's Secret Angel Doutzen Kroes. Probably not widely known, but I think her face definitely has something to it.
The embed has been trained on SD 1.5 and works well on most 1.5 based models. Some elements are a tad overbaked, but prompting for (Koh_DoutzenKroes:0.99) or playing around with advanced prompting such as [a blonde dutch woman:Koh_DoutzenKroes:0.75] mitigates this.
If you enjoy my creations, consider buying me a coffee on Ko-Fi. As I do consume a lot of coffee while creating embeddings, it'd be much appreciated.
koh_doutzenkroes
this is etzadla a nice Sache
v3 more recent face
v2 less stylized
v1 better for anime etc.
rw1710
Paula Gonu is a Spanish influencer and social media personality.
500-step TI trained on a dataset of 20 images with my usual settings.
I recommend running this embedding at a slightly reduced weight. For instance, type "(p4ul4gonu:0.98)" in your prompt.
Curious about my work process? I have summarized it here.
Appreciate my work? My TIs are free, but you can always buy me a coffee. :)
p4ul4gonu
基于novelAi官网版的高木同学(takagi-san)产出的2000张图片,选取其中220张进行景象翻转训练得到。
训练时的学习率为0.1,只训练了一次,没有精细优化。
内置数据集,希望有人可以优化一下,或者做成lora试试。
【如果需要更多数据集进行训练,可私信或者评论区询问,原版2000图片,以及使用该pt生成的几百张图片,我都可以分享出来】
tag一般有两个。(截止2023年1月时使用)
由于修改过代码的原因,如果出现了{}花括号,请自行改成()圆括号,并将权重改为1.1。【当然,或许不改也没问题,可自行测试】
A正:1girl,solo,takagiza-1400,{{kind smile,}}{best quality}, {{masterpiece}}, {highres}, original, extremely detailed 8K wallpaper, {an extremely delicate and beautiful},black hoodie,
B正:1girl,solo,{{kind smile}},{best quality}, {{masterpiece}}, {higher res}, original, extremely detailed 8K wallpaper, {an extremely delicate and beautiful},black hoodie,takagiza-1400,1girl, solo, forehead, light brown hair , long hair, (((tareme))), brown eyes, (((round eyes))), kind smile, round face, (((ringed eyes))), (((flat chest))), (((flat_chest))), (((facial)))
负:wet clothe,nsfw,lowres,highres,worst quality,low quality,normal quality,artbook,game_cg,duplicate,grossproportions,deformed,out of frame,60s,70s,80s,90s,00s. ugly,morbid,mutation,death,kai juu,mutation,no hunmans,monster girl,arthropod girl,arthropod limbs,tentacles,blood,size difference,sketch,blurry,blurry face,blurry background,blurry foreground,disfigured,extra,extra arms,extra ears,extra breasts,extra legs,extra penises,extra mouth,multiple_arms,multiple legs,mutilated,tranny,trans,trannsexual,out of frame,poorly drawn hands,extra fingers,mutated hands,poorly drawn face,bad anatomy,bad proportions,extra limbs,more than 2 nipples,extra limbs,bad anatomy,malformed limbs,missing arms,miss finglegs,mutated hands,fused fingers,too many fingers,long neck,bad finglegs,cropped,bad feet,bad anatomy disfigured,malformed mutated,missinglimb,malformed hands,lowres,bad anatomy,bad hands, text, error, missing fingers,extra digit, fewer digits, cropped, worstquality, low quality, normal quality,jpegartifacts,signature, watermark, username,blurry,bad feet,sharp fingers,wrong figernails,long hand,double middle finger,index fingers together,interlocked fingers,pieck finger,sharp fingernails,steepled fingers,x fingers,curled fingers,interlocked fingers,fingers different thickness,cross fingers,poor outline,big fingers,finger growth,lowrespolar lowres, bad anatomy, bad face, bad hands, bad body, bad shose, bad feet, bad proportions, {bad leg}, {more legs}, worst quality, low quality, normal quality, gross proportions, blurry, poorly drawn asymmetric eyes, text,error, missing fingers, missing arms, missing legs, short legs, extra digit, long skirt, indoors, 2girls, out of frame, only_upper body, only_lower body, {from above}, {petite}, low background,multiple girls,{{{multiple views}}},clone,{{{cleavage_cutout}}}
takagiza-1400
Trained on a variety of CC and public domain vintage children book illustrations
kidbooks
Metropolis, (1927, directed by Fritz Lang) came into public domain this year.
I’ve loved this movie since I saw it the first time, at some goth industrial club in the late 90’s in Hollywood, the stark-but-somehow-soft brutal retro sci-fi on the screen going perfectly with the music.
Later I watched it for a class, and we pulled apart the meaning of the shape language and symbols and metaphor, and it was very satisfying.
I highly recommend checking it out, in either the giant-stompy-boots or cinematographic-dissection manners.
Let me know if you want a 2.x version, I can do that. 😀
metropolis_e
This has been trained on macro photos of lichens and moss that I once took. My intention was to use it as style modifier. Created with Invoke-AI. It works for me with some models and with some it does not, so there's room to experiment. I had good results with realistic/general models, e.g. Deliberate and it works great with iCoMix, which is comic style biased merge.
Use style token alone and it will generate fantasy macros of alien looking lichens. The fun is in mixing it with prompt which requires finding right weight balance. So this is not a ready-made, but requires a lot of tinkering. Here's my approach...
First of all, 'quality' keywords are not really needed and do not make much sense. Depth of field concept is imprinted in the style, as the photos all had a lot of it, so you may want to counter it with negative prompt. Other that that, I would just leave negative part alone.
I would start with defining weights for style and the rest of the prompt, which would look like that (using invoke-ai syntax, but I'm sure you will figure how does it translate, same with prompts on example images):
(<stardeaf-greenmageddon>)0.8, (sleepy zombie reading a paper, raining)0.8
Then I would adjust weights until I find a sweet spot that blends both in an interesting way. The right ratio depends on model, prompt and image size, so experiment... For me odd, detailed ratios seem to work best, like 0.55:0.67. When you get just lichen vegetation, take weight out of style side. I also keep cfg and steps low (below 5 for cfg, step in range of 9-18.)
iCoMix model is here:
https://civitai.com/models/16164
Taking the experiment further, I made derivatives of the embedding. These were trained on images created by using the original style with iCoMix model, as I wanted to see how will they perform when used in other, realistic models...
So the other versions are not updates but derivatives, each with it's own mood. Again, use token weighting to add twirliness and alien organic feel to your image.
Have fun.
Mia Melano
I literally know nothing about this young lady. She's pretty and seems like a nice person. An actress maybe? Maybe a singer? Mathematician? Philanthropist? I have no idea. But @JernauGurgeh asked me very nicely so here she is. LMK if it's ok
[ SDv1.4, 4e-03, 4200 steps, v/s .2054]
miamelano
In this new textual inversion, I aimed to create a beautiful woman that adheres to the aesthetic standards of the 1980s in China. In the 1980s, girls with this kind of appearance would usually appear as magazine or calendar models. In our setting, Pearl珍珠 came from the southern region of China and went to Shenzhen with the tide of reform and opening up in the 1980s, and established a ready-to-wear clothing store there.
Pearl is not a real person.
If you like my work, feel free to buy me a cup of coffee through ko-fi.com/dalecell,it will inspire me to create better work.
80sgirlpearl
Trigger word is "utpvirt"
Welcome to a future of computer bliss and cloudy blue skies! This textual inversion model was trained on various album covers in the Utopian Virtual genre, with Home and ClearSkies by PrismCorp/Vektroid as a general base.
I'm new to training stable diffusion, and this model is pretty stubborn. If you have any good guides, please let me know. Thank you!
utpvirt
This TI makes stuff more colorful and improves the images in some weird way when used as negative.
It was a attempt at making a certain Oekaki artstyle TI, but it didn't work. Maybe someone finds a use for it.
sd1-failed-oekaki
Alice Delish is a TikTok model, best known for cosplay and her eGirl aesthetics.
4d3lish7
Style of heroic fantasy animal-human mixes. Trained with a dataset of 68 character concept art images.
Save to your embeddings folder in SD installation. Prompt the file with fantasycharacter
Recommended settings
CFG Scale: I tested all images with 7, feel free to experiment with lower values, from 4 to 5.5 to give it more freedom.
Steps: 30 will produce concept art, 50 solid digital artwork, 100-150 for highly detailed renders.
Positive prompt
Formula:
[Medium] of [Subject constants], [Subject Descriptions], [Style tags], [Beautifier tags], [Model Specific Tags]
See a complete list of positive prompts you can use in the version description here, just add your desired medium in front.
Negative prompt:
Because of a few training images, it is recommended to add these:
watermark, logo, artist name, signature, text
The example I used is a mix of a few negative embeddings plus some words I added myself to fine-tune the results:
ng_deepnegative_v1_75t, bad-artist, bad-hands-5, lowres, ((bad anatomy)), blurry, ((mutated hands and fingers)), (poorly drawn face), ((mutation)), ((deformed face)), (ugly), ((bad proportions)), ((extra limbs)), extra face, (double head), (extra head), ((extra feet)), logo, cropped, worst quality, low quality, normal quality, jpeg, humpbacked, long body, long neck, ((jpeg artifacts)), (worst quality:2), (low quality:2), (normal quality:2), lowres, normal quality, ((monochrome)), ((grayscale)), artist name, watermark, logo, signature, text, out of the image, out of frame, pale colors, NSWF, nude
Canvas size
512x768 with a highres fix at 1.5 times and Denosing at 0.5
Model recommendations
I tested this on 10+ models (see examples) and concluded this:
Anime models will do humans with cat ears. Tested 3-4, maybe you get lucky.
Standard quality results with general purpose models like Deliberate and Midjourney family, DreamShaper and Dreamlike Diffusion.
Interesting results with specialized models like RPGv4 (killing on the outfit side), should be interesting with "furry" models (for example FurryMix) or PixarStyle for Disney-like characters.
Amazing results with WestMagic. I expect Asian graphics models to kill the outfit and standing pose.
All the models you can find in the Civitai search.
Beyond the Examples
Feel free to experiment with other models, style strengths (other than 1.0), low CFG (3-5.5) and high steps (100-150) combinations on high-performing models, or anything else that crossed your mind.
Make sure to share the results in the comments.
Follow us for more
Peaksel team is inspired by the willingness of this community to share and is grateful for all the tools and knowledge we got here. This is why we plan to share all the best tools we make. Follow us to stay in the loop.
fantasycharacter
Courteney Cox is an American actress known for the sitcom Friends and the film franchise Scream, amongst many other roles.
My goal with this TI was to capture her looks in the late 80s and early 90s.
I trained the TI for 1000 steps, but it got overtrained pretty soon and I ended up setting for step 400, which might still be very slightly overtrained too, but looks really accurate in my opinion. In any case, I recommend using the embedding at a reduced weight. For instance, type "(c0urt400:0.99)" in your prompt. That should solve any issue in most cases.
Curious about my work process? I have summarized it here.
Appreciate my work? My TIs are free, but you can always buy me a coffee. :)
c0urt400
Geralt of rivia.
Thank you for visiting my post <3
geraltofrivia
A request of @blackpro98.
Sasha Pieterse is a South African/American actress and singer.
1000-step TI trained on a dataset of 20 images with my usual settings.
Curious about my work process? I have summarized it here.
Appreciate my work? My TIs are free, but you can always buy me a coffee. :)
s4sh4p1eterse
Annelese Milton is an Australian Instagrammer famous for posting photos of lingerie and bikinis.
To train this IT I have used a dataset of 29 images . 450 steps
Enjoy :D .
ann3l3s3milton
Actress and mother Alicia Silverstone was in the move Clueless and some music video things back in the 90s I believe.
This is the Textual Inversion counterpart to my LoRA
The results, as they say, speak for themselves
Do you like my work? Buy me a beer!
[ SDv1.4, 4e-03, 5600 steps, v/s .199 ]
(all images used to train this model are of Ms. Silverstone 18-years old and older)
aliciasilverstone
Elise is the fifth in a new set of custom LastName characters, for anyone who wants a consistent character for their SD 1.5 renders. (v2 now available!)
Elise is the stunning Icelandic girl you met in Reykjavik during your gap year after college. She's a trainee nurse by day, and an almost-professional ice skater by night. She looks really, really good in a sparkly dress. She looks really, really good in a bikini. Her favorite colors are blue, gray, and blue-gray. It's far too expensive for you to ever move to Reykjavik, but she's far too beautiful for you to not keep in touch anyway.
To use this embedding with the Automatic1111 web UI, save the downloaded file to /embeddings/eliselastnamesd15.pt
, and then include the trigger word eliselastnamesd15
in your prompt.
(Inspired by Zovya's Nobody series, which is awesome. Go download all of their amazing characters.)
(All of the sample images have full metadata, and are straight out of the sausage machine - no upscaling, inpainting, or img2img-ing.)
eliselastnamesd15
Xena Kai is an Instagram and fitness model, former "Race Queen" promotional model.
zenakayz
Appreciate my work? Can you buy me a coffe please :)
Neringa Križiūtė is a Lithuanian aspiring model, artist, Instagram star, and social media star. She has amassed thousands of fans on Instagram. Also, Ella Neringa has worked and worked with many brands like Fashion Nova. Also, she is a painter.
I still need to detail the technique of how to generate my textual inversion I am still detailing so that they are more flexible but well this already looks a bit woman I hope you like it.
I have only tried with Clarify, an apology for my work, what happens is that at the moment I work from Colab.
Used model: Realistic Vision v1,3
recommended parameters:
Prompt:
RAW photo, a close up portrait photo of 26 y.o woman in wastelander clothes, long haircut, pale skin, slim body, background is city ruins, (high detailed skin:1.2), 8k uhd, dslr, soft lighting, high quality, film grain, Fujifilm XT3
Negative Prompt:
(deformed iris, deformed pupils, semi-realistic, cgi, 3d, render, sketch, cartoon, drawing, anime:1.4), text, close up, cropped, out of frame, worst quality, low quality, jpeg artifacts, ugly, duplicate, morbid, mutilated, extra fingers, mutated hands, poorly drawn hands, poorly drawn face, mutation, deformed, blurry, dehydrated, bad anatomy, bad proportions, extra limbs, cloned face, disfigured, gross proportions, malformed limbs, missing arms, missing legs, extra arms, extra legs, fused fingers, too many fingers, long neck
Euler A or DPM++ 2M Karras with 25 steps
CFG Scale 7
QR to my page buymeacoffee
neringa_t_v10-600
Tricia Helfer (born April 11, 1974) is a Canadian-American actress and former model. She played the enigmatic Cylon model Number Six in the re-imagined Battlestar Galactica series (2004–2009). She also voiced Sarah Kerrigan in StarCraft II and its expansion packs (2010–2015). She played Charlotte Richards/the Goddess of All Creation on the TV series Lucifer (2016–2021).
Filename: sd15-tricia-helfer.pt
Note: You might wish to use sd15-tricia-helfer woman to activate the textual inversion to ensure a woman is rendered instead of a man. If you choose to rename the file, simply change the prompt to include the new filename.
sd15-tricia-helfer
Charlize Theron (born 7 August 1975) is a South African and American actress and producer. One of the world's highest-paid actresses, she is the recipient of various accolades, including an Academy Award and a Golden Globe Award. In 2016, Time named her one of the 100 most influential people in the world.
Filename: sd15-charlize-theron.pt
Note: You might wish to use sd15-charlize-theron woman to activate the textual inversion to ensure a woman is rendered instead of a man. If you choose to rename the file, simply change the prompt to include the new filename.
Note: Charlize does have a mole on her neck which I had never really noticed but which the training did pick up. In some instances, the mole shows up in multiples. Strange, but I don't know a way to fix this other than removing it from all source images.
sd15-charlize-theron
Solomia Maievska, Instagram model from Ukraine, with an unusual appearance.
Be careful while generating pictures, she loves to show off her "talents" so make sure you properly dress her in prompts before you start.
I hope someone likes this TI. I'm still just learning and practicing, so please don't judge me too hard)
maievska
Mirella Papciak @mirella.papciakk
Polish fashion model and Instagram influencer.
Trained with 67 hand-picked Instagram photos, zip attached here.
mirellapapciakk
A request of @derp123.
Mia Sand is a Danish fitness model and social media celebrity.
1000-step TI trained on a dataset of 25 images. This time I modified slightly my usual settings, reducing the learning rate to 0.002 after 500 steps (I had tested the model by then and it looked almost ready, but there were signs of undertraining on her skin features and such).
Given the subject, I've tried to focus mostly on full body shots. I think the approach paid off in the end, as some of the example pics evidence, with her powerful physique on display (without prompting specifically to that effect).
Curious about my work process? I have summarized it here.
Appreciate my work? My TIs are free, but you can always buy me a coffee. :)
missmiafit
Nice little addition to the vocabulary of your SD installation.
Trained on 10 HQ Midjouney images.
Add hobbithouse to your positive prompt. Check the example images for complete tags and settings.
Tested across a variety of models, works well with Cheese Daddy's Landscape Mix Offset
hobbithouse
Nadia Jémez is a Spanish model and social media personality.
I was hesitant about uploading this TI, as Nadia is not a world-famous celebrity, and even some people from my country won't have ever heard of her. She's the daughter of former international footballer Paco Jémez, but I think many of you will agree she's stunningly beautiful and deserves recognition of her own. This is also one of my best embeddings, I'd say, and that's eventually what made me decide to upload it. Hope you'll like it!
500-step TI trained on a dataset of only 12 images with my usual settings.
Curious about my work process? I have summarized it here.
Appreciate my work? My TIs are free, but you can always buy me a coffee. :)
nadiajemez
Madelyn Renee Cline (born December 21, 1997) is an American actress and model. She is known for her roles as Sarah Cameron on the Netflix teen drama series Outer Banks (2020–present) and as Whiskey in Rian Johnson's mystery film Glass Onion: A Knives Out Mystery (2022).
If you like my work, a review is more than welcome! Also, follow me on Twitter or Instagram!
Looking for something... spicy? Here's my Patreon!
Also, I recently started a blog where I will gather resources and tutorials about my work here!
Hello everyone! Today I present to you my textual inversion of Madelyn Cline! I used 50 public images of her with a 7000 steps process. To use it in your works:
download the file and place it into stable-diffusion-webui>embeddings folder;
add madclyne97 to your prompt;
keep the sampling steps around 20/25, and CFG between 7 and 8 to work well. Great results with DPM++ SDE Karras & Edge of Realism;
madclyne97
This is an experiment textual inversion trained on high quality Wakfu artwork. It likes to give characters animal ears - if you don't want that add (((animal ears))) to the negative prompt (that's because a lot of characters in the original artwork have either horns, animal ears or funny hats).
wakfu
This is a Stable Diffusion textual inversion about a former female wrestler. If you wanna test it, you can put this word on the prompt: stcyKblr .
PS: The test images was made with 'vainilla' SD 1.5 model. I don't know if works on SD 2.1 .
If you wanna support me, you can buy me a coffee: https://www.buymeacoffee.com/elrivx
stcykblr
Keri Lynn Russell (born March 23, 1976) is an American actress. She portrayed the titular character on the drama series Felicity (1998–2002), which won her a Golden Globe Award, and Elizabeth Jennings on the FX spy thriller series The Americans (2013–2018), which earned her nominations for several Primetime Emmy and Golden Globe Awards.
Filename: sd15-keri-russell.pt
Note: You might wish to use sd15-keri-russell woman to activate the textual inversion to ensure a woman is rendered instead of a man. If you choose to rename the file, simply change the prompt to include the new filename.
Note: Keri does have a lazy eye which I had never really noticed but which the training did pick up.
sd15-keri-russell
Lucia Javorcekova is a Slovakian model, former professional cyclist and social media celebrity.
1000-step TI trained on a dataset of 25 images with my usual settings.
I recommend running this embedding at a slightly reduced weight. For instance, type "(1uciajavorcekova:0.99)" or "(1uciajavorcekova:0.98)",
Curious about my work process? I have summarized it here.
Appreciate my work? My TIs are free, but you can always buy me a coffee. :)
1uciajavorcekova
This style has been trained on 50 Midjourney head and shoulders portraits of old women in steampunk style. It can produce young women, but men very hard.
Recommended settings:
Canvas: 512x512 with high res fix at 1.5 or 2.0 upscale
Models: realistic (Deliberate, AnalogMadness) or Midjourney family (Dreamshaper, Dreamlike Diffusion, Protogen family etc). Anime models perform well.
Simple concept, feel free to experiment with simple and detailed prompts and share your results in the comments.
steampunkgranny
575 step TI of instagram model, performs better from testing on lower realism models like Duchenai, babes, stably etc.
Feedback is appreciated.
this is my first TI created from a dataset of amanda lee, an instagram model, following the directions laid out by JernauGurgeh.
amandal33
One of the most popular supermodels of the last decade, Lusy shared her prodigious...uh...talents with the world and we were all happier for it.
Do you like my work? Buy me a beer!
[ SD1.4, 4e-03, 5600, v/s .210]
lucypinder
Appreciate my work? Can you buy me a coffe please :)
Venessa Bohorquez : The Venezuelan model Vanessa Bohórquez began creating content at the end of 2016. Her career in social networks began as a fitness model and influencer, then she made the leap to creating exclusive content loaded with sensuality. In 2017, she saw a business opportunity on the Onlyfans platform, she became one of the pioneers of the subscription service and is now established as the most successful Venezuelan, above recognized figures.
Used model: Realistic Vision v1,3
recommended parameters
RAW photo, a close up portrait photo of 26 y.o woman in wastelander clothes, long haircut, pale skin, slim body, background is city ruins, (high detailed skin:1.2), 8k uhd, dslr, soft lighting, high quality, film grain, Fujifilm XT3
Negative Prompt:
(deformed iris, deformed pupils, semi-realistic, cgi, 3d, render, sketch, cartoon, drawing, anime:1.4), text, close up, cropped, out of frame, worst quality, low quality, jpeg artifacts, ugly, duplicate, morbid, mutilated, extra fingers, mutated hands, poorly drawn hands, poorly drawn face, mutation, deformed, blurry, dehydrated, bad anatomy, bad proportions, extra limbs, cloned face, disfigured, gross proportions, malformed limbs, missing arms, missing legs, extra arms, extra legs, fused fingers, too many fingers, long neck
Euler A or DPM++ 2M Karras with 25 steps
CFG Scale 7
QR to my page buymeacoffee
vanessab_t_v2-600
Chloe Jones
My trip through the 90s ends here, lads. Chloe Jones was another one of those supermodels that seemed to be everywhere, which was a good thing.
I will upload more examples
Do you like my work? Buy me a beer!
[ SD1.4, 4e-03, 5200, vs .199]
chloejones
Mads Dittmann Mikkelsen (22 November 1965) is a Danish actor. Originally a gymnast and dancer, he rose to fame in Denmark as an actor for his roles such as Tonny in the first two films of the Pusher film trilogy (1996, 2004). Mikkelsen achieved worldwide recognition for playing the main antagonist Le Chiffre in the twenty-first James Bond film, Casino Royale (2006).
If you like my work, a review is more than welcome! Also, follow me on Twitter or Instagram!
Looking for something... spicy? Here's my Patreon!
Also, I recently started a blog where I will gather resources and tutorials about my work here!
Hello everyone! Today I present to you my textual inversion of Mads Mikkelsen! I used 40 public images of him with a 9000 steps process. To use it in your works:
download the file and place it into stable-diffusion-webui>embeddings folder;
add mads65 to your prompt;
keep the sampling steps around 20/25, and CFG between 6 and 8 to work well. Great results with DPM++ SDE Karras & Realistic Vision;
mads65
A request of @Stories9.
Julia Rose is an American model and social media celebrity.
601-step TI trained on a dataset of 20 images with my usual settings.
Curious about my work process? I have summarized it here.
Appreciate my work? My TIs are free, but you can always buy me a coffee. :)
jul14r0s3
Appreciate my work? Can you buy me a coffe please :)
This textual inversion is inspired by the style of the artist Oscar Bazaldua (Jan), a famous cartoonist for magazines with an erotic and sensual touch called 'Sexacional' such as 'Las chambeadoras, Bellas de Noche, Almas Perversas', which are just some of the titles of the comics that abounded in Mexican magazine outlets and that have been generically called sexational during the 1980s and 1990s.
This is my first textual inversion, I haven't mastered the technique yet, but I already got this first version, which I thought was a good training session and very similar to the artist's style.
I haven't tried it with other models, so far I've only tried it with YuzuGinger, I suppose it can be used with models that have a good training of Anime images in the background, I don't know if it works with realistic style models.
QR to my page buymeacoffee
bazaldua_v2-10000
Emma Kotos is an American model and social media influencer.
850-step TI trained on a dataset of 20 images with my usual settings.
Curious about my work process? I have summarized it here.
Appreciate my work? My TIs are free, but you can always buy me a coffee. :)
k0t0s850
Textual inversion model based on Françoise Hardy, French singer, actress and songwriter.
françoise hardy
Kristina Milivojcevic, @chefk1ki is a Serbian girl living in Chicago. An economist by education, she became famous for her pancakes (Kiki sweet crepes) she was making in a restaurant called "Moscow" in Chicago.
She is a perfect model for your kitchen designs.
How to use
Copy the file to your embeddings folder and prompt chefk1ki
chefk1ki
Kathleen Marie Ireland (born March 20, 1963) is an American author, entrepreneur, fashion designer, philanthropist, and former fashion model. Ireland was a supermodel in the 1980s and 1990s, initially known for appearing in 13 consecutive Sports Illustrated swimsuit issues, including 3 covers.
https://en.wikipedia.org/wiki/Kathy_Ireland
(* No NSFW imagery was used for this.)
This is a TI I made for Artoria Lancer Alter. Tends to give quite revealing clothes, with cutouts, cleavage, etc.
cooler_lalter-9000
The Stable Diffusion 1.5 have been trained in some images of Emilia Clarke so If you use the actress name this can boost her accuracy, but if doesn't work just remove, also I recommend use game of thrones in the positive prompt.
I also recomend use wide-angle in negative prompt if you getting too skinny faces.
Triggers
daenerys in AUTOMATIC1111
<daenerysv100sd15> in InvokeAI
Recommended model: https://civitai.com/models/3855/epic-diffusion
Recomended CFG scale at 9 or 10. But all depends on wich engine, sampler and model you are using.
Don't forget to do your due prompt work.
daenerysv100sd15
Another supermodel from the 90's, Neriah is as smart as she is stunning, having earned a PhD from Stanford in Molecular Chemistry.
I will upload more examples later
Do you like my work? Buy me a beer!
[ SD1.4, 4e-03, 5800, vs .199]
neriah
Now for something more recent, filling a gaping hole in out selection of supermodels, Keeley was in a few episodes of Ted Lasso I believe.
I will upload more examples later
Do you like my work? Buy me a beer!
[ SDv1.4, 4e-03, 5600, v/s .202 ]
keeleyhazell
I always had a thing for Crissy, one of the more underrated supermodels of the late 90s - early 00s.
I will upload more examples later
Do you like my work? Buy me a beer!
[ SDv1.4, 4e-03. 5400, v/s .210 ]
crissymoran
Hi guys!
This is an embedding that inspired by a vintage Pokemon trainer style.
If you wanna test it, you can put this word on the prompt: art by knSgmrTrnr
And I suggest this negative prompts: Deformed, blurry, bad anatomy, disfigured, poorly drawn face, mutation, mutated, extra limb, ugly, poorly drawn hands, missing limb, blurry, floating limbs, disconnected limbs, malformed hands, blur, out of focus, long neck, long body, ((((mutated hands and fingers)))), (((out of frame)))
This model only works on SD 1.5 .
if you wanna support me, you can buy me a coffee: https://www.buymeacoffee.com/elrivx
knsgmrtrnr
Appreciate my work? Can you buy me a coffe please :)
Erin Eevee is an Instagram Star from Australia. She is well known for her hot looks and cute smile. She is famous on various social media platforms including Instagram. Erin Eevee has earned over 313K followers on Instagram.
Used model: Realistic Vision v1,3
recommended parameters
RAW photo, a close up portrait photo of 26 y.o woman in wastelander clothes, long haircut, pale skin, slim body, background is city ruins, (high detailed skin:1.2), 8k uhd, dslr, soft lighting, high quality, film grain, Fujifilm XT3
Negative Prompt:
(deformed iris, deformed pupils, semi-realistic, cgi, 3d, render, sketch, cartoon, drawing, anime:1.4), text, close up, cropped, out of frame, worst quality, low quality, jpeg artifacts, ugly, duplicate, morbid, mutilated, extra fingers, mutated hands, poorly drawn hands, poorly drawn face, mutation, deformed, blurry, dehydrated, bad anatomy, bad proportions, extra limbs, cloned face, disfigured, gross proportions, malformed limbs, missing arms, missing legs, extra arms, extra legs, fused fingers, too many fingers, long neck
Euler A or DPM++ 2M Karras with 25 steps
CFG Scale 7
QR to my page buymeacoffee
evee_t_v5-550
Julia Garner as Ruth Langmore from Ozark on Netflix. Trained from 42 images with filewords. I made this a while back, but I want to say I trained at 0.001 ¯\_(ツ)_/¯
Buy me a slice of Pizza if you're feeling the vibes. More rare embeddings/loras to come. https://www.buymeacoffee.com/crept
This was my first attempt, so go easy on the quality. I'm learning and my video card isn't the best. :D
j-garns
Follow me to make sure you see new poses and Nobodys when I post them. Things move fast on this site, it's easy to miss.
Meet Sasha Nobody, this person doesn't exist. You may think she looks like someone you know, but she isn't. Just the imprecise nature of diffusion. Images were generated using the legendary A to Zovya RPG Artist's Tools model. Some poses helped by zPoses
If you need a consistent character but don't want a known celebrity, Sasha is your gal. Sasha prefers to have long straight hair and body freckles/moles unless you prompt that out.
Follow me for more of the Nobody family coming soon for 1.5 and 2.1.
Click the zNobody tag for more, or the Nobody tag to see other's contributions.
Do you have requests? I've been putting in many more hours lately with this. That's my problem, not yours. But if you'd like to tip me, buy me a beer. Beer encourages me to ignore work and make AI models instead. Tip and make a request. I'll give it a shot if I can. Here at Ko-Fi
sashanobodysd15
ellenripleyt
Ashley Resch is a Canadian model and social media celebrity.
This TI has been updated! Please check the "About this version" section for more info about training.
Curious about my work process? I have summarized it here.
Appreciate my work? My TIs are free, but you can always buy me a coffee. :)
jg_ashleyresch
Originally posted to Huggingface.
The embeddings in this repository were trained for the 768px Stable Diffusion v2.1 model. The embeddings should work on any model that uses SD v2.1 as a base.
Usage
To use the embeddings, download and then rename the files to whatever trigger word you want to use or just keep the original name.
Example Prompt:
```
a cute skinny cat wearing a winter toque and scarf in the snow, realistic, at artstation and behance, art by wcat8-v2-2000, cinematic lighting, night, heavy snowfall, snowstorm
```
Negative Prompt:
```
blurry, photo, smooth, cartoon animal, vector art, 2d, illustration, two deer wearing suits, comic book, 1970 film photography, very sexy woman with black hair, red on black, lace dress, deformed, tokyo japan, pen drawing, fat, chubby, concept art
```
wcat8-v1-2000
A TI of model and TV presenter Kelly Brook. I wasn't happy with the default in any SD model, so I made this.
Most of the training images are from her mid-20s (none were -18).
Of course, a large proportion were lingerie, so it does tend to produce lacy bras! None were NSFW though, so accidental nudity doesn't seem to be an issue.
K3LLY8R00K2
Originally posted to HuggingFace.
The embeddings in this repository were trained for the 512px Stable Diffusion v1.5 model. The embeddings should work on any model that uses SD v1.5 as a base.
For Stable Diffusion v2.x, see my KnollingCase embeddings for SD 2.0.
Knollingcase v1
The v1 embeddings were trained for 4000 iterations with a batch size of 4 and a text dropout of 10%, using Automatic1111's WebUI. A total of 147 training images with high quality captions were used.
Embeddings are provided for steps 1000, 2000, 3000, & 4000.
Usage
To use the embeddings, download and then rename the files to whatever trigger word you want to use. They were trained with kc16, but any trigger word should work.
The knollingcase style is considered to be a concept inside a sleek (sometimes scifi) display case with transparent walls, and a minimalistic background.
The images of people standing inside the case were made with using the embeddings on the F222 model.
kc16-4000-sd1-5
Riae (born September 24, 1984) is an Italian model who is best recognized for her alternative modeling style and for having earned representation from the Suicide Girls modeling agency. She shares additional modeling photos through her Instagram as well for her over 4.5 million followers.
If you like my work, a review is more than welcome! Also, follow me on Twitter or Instagram!
Looking for something... spicy? Here's my Patreon!
Also, I recently started a blog where I will gather resources and tutorials about my work here!
Hello, nice people of Civitai! Today I present to you my textual inversion of Riae! I used 50 images of her (30 close portraits, 20 half-lenght), with a 9000 steps process. To use it in your works:
download the file and place it into stable-diffusion-webui>embeddings folder;
add riae84 to your prompt;
keep the sampling steps around 30, and CFG between 6 and 8 to work well. Great results with DPM++ SDE Karras & Realistic Vision. Face restore could help sometimes, to make some softer lines when necessary;
riae84
Heather Carolin
One in a billion. The most magnificent redhead ever.
Do you like my work? Buy me a beer!
[ SDv1.4, 4e-03, 5500, v/s .201 ]
hcarolin
This Textual Inversion has trained publicly available images of Tropical Island with a Stable Diffusion 1.5 with 12 images at a scale of 512x512.
Works best with DPM++ 2M Karras.
Tested on SD 1.5 and hasdx_hasedsdx (NSFW)
The latest version is trained at 5400 steps.
So use tropical-island or tropical-island-5400 in a prompt to trigger this textual inversion
(As the sun sets on this remote tropical island, the sky is painted with a fiery glow, as if the nearby volcano has erupted and spilled its lava into the air. The dramatic scenery is only heightened by the crystal-clear water that reflects the colors of the sky, creating an otherworldly ambiance that is truly unforgettable),tropicalparadise, remoteisland, untouchednature, purewater, lushgreenery, secludedparadise, naturalwonderland, hiddenbeauty, crystalclearwater, densejungle, vibrantfoliage, highqualityimages, photorealism, camerasettings, daylightlighting, realpictures, 8Kresolution, islandescape, breathtakingviews, exoticwildlife, naturalhabitat, unspoiledparadise, serenitynow, peacefulretreat, wanderlustdestination, offthebeatenpath, escapethecity, ecoadventure, sustainabletourism, conservationefforts, tropical-island-5400
tropical-island
Don't underestimate emb by it's only over a dozen kb. Although this metaphor is not appropriate, early computers were as big as houses, but how big are chips today?
In a way embedding is better than lora with face loyalty restoration, hope more bro try to train your emb
liuyanalexzuov1
This TI is based on Crowley from the show Good Omens, as played by David Tennant.
Tips: I've found this to work best with the Realbiter model, but Dreamshaper can do some wonderful stuff as well. It looks overcooked in most other models. Lowering CFG may help. His yellow eyes are trained in but only show up with models that are trained for unusual colors. Prompt "sunglasses" to summon his glasses. It is not trained on his outfit. Alas, the face tattoo is too complicated for Stable Diffusion to pick up.
gocrow
This TI is based on the character of Aziraphale from Good Omens, as played by Michael Sheen. Sheen's dimples, upturned nose, and wholesome look also lend themselves well to a unique female character.
Tips: I've found this to work best with the Realbiter model, but Dreamshaper can do some wonderful stuff as well. Other models tend to distort and feminize the face. This TI has memorized his outfit and will use it as a default. If results look too young, try aging words in your prompt like "jowls", "wrinkles" and "chubby." For the fan favorite chubby look, try "thick thighs".
goaz
Sometimes, you just build a billionaire's enterprise on a leaked adult tape.
There's some slight redundancy to having a embedding of her, since most models are familiar with her due to her prevalence online. This should provide more consistency however, as SD1.5 amalgamates everything, fanart, old photos, bad pictures, photoshops etc.
The embed has been trained on SD 1.5 and works well on most 1.5 based models. When rendering full body images I recommend using inpainting. It also works well to let the AI use known information from the model by prompting for [Kim Kardashian|Koh_KimKardashian] or [Kim Kardashian:Koh_KimKardashian:0.8]. This allows the AI more creativity and it will get more varied outputs, it's worth trying.
If you enjoy my creations, consider supporting me on Ko-Fi, coffee does work like fuel ^^
koh_kimkardashian
This is a favor for work friend (Hey Sammy!) who asked me a million questions about SD today, and threw in that Minji is his current crush.
Minji is a member of the K-Pop group NewJeans. There's 5 girls in the group, Minji is only one of 2 who are 18, so there won't be a complete set for at least 3 years.
I got roasted for my Tzuyu embedding, so I'm sure another one is coming.
Do you like my work? Buy me a beer!
[ SDv1.4, 5e-03, 4400, .207 ]
minji-nj
Former competitive skier from Sweden, she took a wrong turn and ended up at the Miss Sweden beauty pageant, where she finished second just by showing up. She was the spokesmodel for just about everything in the mid 90s.
Do you like my work? Buy me a beer!
The examples are literally 10 of the first 20 gens I made of her, so create with confidence!
[ SDv1.4, 5e-03, 3800, .204 ]
victsilv
Eva Angelina
Another shocking omission from our treasure trove of adult embeddings. Eva hit the scene HOT I tell you. And those glasses? Hoo-boy.
You'll have to prompt the glasses in, most of my source pics did not have glasses.
Do you like my work? Buy me a beer!
The vector strength is higher on this that my usual embeddings, earlier files didn't quite pass the "at first look" test (as every good embedding and LoRA should)
[ SDv1.4, 5e-03, 2800, v/s .214 ]
evaangelina
Shay Mitchell is a Canadian actress and model.
300-step TI trained on a dataset of 15 images with my usual settings.
Curious about my work process? I have summarized it here.
Appreciate my work? My TIs are free, but you can always buy me a coffee. :)
sh4ym1tch
Someone asked if I'd make an embedding of adult film star Emily Willis and having never heard of her, didn't. Curiosity got the better of me so I checked her out and WOW have I been missing out. Thank you kind sir for bringing this goddess to my attention.
Do you like my work? Buy me a beer!
[ SDv1.4, 5e-03, 5400, v/s .203 ]
emwillis
She's and a german-brazilian actress.
I mixed some older photos of her to help increase her characteristics.
Triggers
sophie in AUTOMATIC1111
<sophiev100sd15> in InvokeAI
Recommended model: https://civitai.com/models/3855/epic-diffusion
Recomended CFG scale at 11 or 14 except if your sampler starting tearing the face then use something like 9 or 10 but always higher.
Don't forget to do your due prompt work. I recomend the prompts of this person: https://civitai.com/user/helicalink
sophiev100sd15
Gorgeous, buxom, and slender brunette knockout Patricia Ford was born in Honolulu, Hawaii. She's of German, Irish, Chinese, Hawaiian, and Portuguese descent. She achieved her greatest enduring popularity as a model in numerous Playboy newsstand special editions, voted Playboy's Lingerie Supermodel of the Year in 1996.Ford stopped modeling while still in her early thirties.
Do you like my work? Buy me a beer!
[ SDv1.4, 5e-03, 3600 steps, .202 vector strength ]
patriciaford
For a long time, I've had this idea that Prince performing "Purple Rain" in the middle of a purple rainstorm would be an incredibly powerful and awe-inspiring scene. And now, thanks to AI technology, I've finally been able to bring this idea to life.
If you like this model, feel free to buy me a cup of coffee.
prince
Gwyneth Kate Paltrow (born September 27, 1972) is an American actress and businesswoman. She is the recipient of various accolades, including an Academy Award, a Golden Globe Award, and a Primetime Emmy Award.
If you like my work, a review is more than welcome! Also, follow me on Twitter or Instagram!
Looking for something... spicy? Here's my Patreon!
Also, I recently started a blog where I will gather resources and tutorials about my work here!
Hello there! Today I present you my textual inversion of Gwyneth Paltrow. I used 60 public images of her (30 close portraits, 30 half-lenght), with an 8200 steps process. Hope you'll like it!
To use it in your works:
download the file and place it into stable-diffusion-webui>embeddings folder;
add gwypaltrow72 to your prompt;
keep the sampling steps around 30, and CFG between 6 and 8 to work well. Great results with DPM++ SDE Karras & Realistic Vision. Face restore could help sometimes, especially when the image is not a close-up/portrait;
gwypaltrow72
Embedding that generates Shadow of the Tomb Raider like Landscapes.
Trained on video game images, which represents the mood of the Tomb Raider.
For good with high details results it needs to be upscaled.
style_tombraider
更新人物小传,具体可看评论
I hope the model will inspire artists and enthusiasts to create unique and beautiful artwork in the New Dragon Inn style. If you are satisfied with this model, please consider supporting my work by leaving a positive review or buying me a cup of coffee.
I am excited to share my latest AI drawing model, inspired by the iconic film New Dragon Inn. This model is designed to enable artists and enthusiasts to create artwork in the distinctive style of the film.
The New Dragon Inn model is based on SD1.5. The model can be used to generate high-quality images that capture the mood and atmosphere of the film, with its bold use of colors, dramatic lighting, and intricate details.
The image size for this model is 540x300, but it can also be used to generate images in other standard sizes.
Moreover, the model includes several control parameters that allow users to adjust the generated image's atmosphere. By adjusting parameters such as coat, sand, fire, and others, users can create images that convey a variety of moods, from peaceful and serene to intense and dramatic. These control parameters provide even greater flexibility for artists and enthusiasts to experiment and create unique artworks in the New Dragon Inn style.
dragon inn-18500
Embedding of Lara Croft
laracroft
This is a textual inversion trained on selected wallpapers of legendary artists VladStudio. The style will produce soft gradient colors, black outlines around the character, and a soft brush effect of watercolor painting.
Random cute wallpapers or images for puzzle games or just add it as a style to landscapes or scenes. It doesn't work well on characters as it will try to doodle them into the style.
Honestly, the style is unpredictable but you will be able to pick 20-30% of images and some will be very interesting.
To use the style on the landscape, follow the workflow explained here.
Settings
Recommended style strength 0.5-0.8 for creative results, 1.0 for full Vlad Studio characters. Example: (vladstudio:0.8).
CFG 7-9, depending on the model. Steps 30-50 (simple graphics, so enough). Canvas size 640x360 with high-res fix of x1.8 or 2 times, your fav upscaler, and denoising strength of 0.2-0.4. When happy, upscale more for your desired HD resolution.
Positive prompt formula
8k Unity Wallpaper, (vladstudio:0.8), [X]
[X] holds the values from the X field in the StylePile extension. Feel free to add your additional style descriptor at the end ([X], volumetric light, surreal etc) if you wish, but may affect the style.
[X] Values (subject only, no descriptors):
cute wallpaper
1boy and 1girl
lonely tree, centered
multiple hearts, love, musical notes, red theme
starry night, dark blue theme
Negative prompt
use a couple of negative embeddings and negative tags from example images of your model.
For deliberate, I used:
ng_deepnegative_v1_75t, ((((ugly)))), (((duplicate))), ((morbid)), ((mutilated)),(((cross eyed))), [out of frame], extra fingers, mutated hands, extra ears, ((poorly drawn hands)), ((poorly drawn face)), (((mutation))), (((deformed))), blurry, ((bad anatomy)), (((bad proportions))), ((extra limbs)), cloned face, (((disfigured))). out of frame, ugly, extra limbs, (bad anatomy), gross proportions, (malformed limbs), ((missing arms)), ((missing legs)), (((extra arms))), (((extra legs))), mutated hands, (fused fingers), (too many fingers), (((long neck))), (((text))),(((signatures))), face cut, head cut, cropped head, jpeg artifacts, low quality, lowres, text, watermark, logo, text, letters, signature, username, words, blurry, cropped, cartoon, worst quality, split images, out of frame, duplicate, watermark, signature, frame, text
Please check the review examples below, some amazing results.
This style delivers totally unpredictable and non-repeatable results, so please share your creations with us!
vladstudio
I noticed that the embedding tends to make her forehead unreasonably large, so it works best when adding something like "large forehead, big forehead" in negative prompt.
emb_beautx
This Chloe Grace Moretz embed was trained in SD 1.5, for 5000 Steps using Textual Inversion at a learning rate of 0.005:2000,0.00005:5000. With a dataset of 30 images at 3 vectors.
Type the name of the embed in prompt to use in your gens, its "ChlMtz2" by default but you can rename it to whatever you please.
PLEASE USE WITH THE "vae-ft-mse-840000-ema-pruned" VAE!!!
and DO NOT OVERLOAD with negative prompts, sometimes they affect the embed and you will LOSE likeness.
I'd really appreciate it if you could take the time to leave a rating/comment after you try out the embed! Helps me a lot and also tells me if I should be tweaking or changing anything for future uploads or versions!
Thank you in advance <3
Check out my DevArt for artworks and photoshoots created using my embeds
https://www.deviantart.com/seedspiller
I also have a Ko-fi for tips and commissions
https://ko-fi.com/seedspiller
chlmtz2
Embedding of Adam Sandler
adamsandler
Embedding of Bruce Willis
brucewillis
Embedding of Sophia Thomalla, girlfriend of Alexander (Sasha) Zverev
sophiathomalla
Embedding of Vanessa Mai
vanessamai
A textual inversion trained on a collection of human anatomy illustrations
anatomical-illustration
This is a Stable Diffusion textual inversion about . If you wanna test it, you can put this word on the prompt: kthrineLngFrdWhy .
PS: The test images was made with 'vainilla' SD 2.1 model.
If you wanna support me, you can buy me a coffee: https://www.buymeacoffee.com/elrivx
kthrinelngfrdwhy
My first TI, any tips or suggestions would be greatly appreciated!
4dd1sonrae-950
测试完成了。接下来我会弄2.0版本的,全新的训练集处理和新的参数
The test is complete. Next, I will work on version 2.0, new training set processing, and new parameters
更正一下,trigger name 是我乱填的 a woman
真正的tirgger name 是embedding本身的名字 我已经更正了。为这个失误抱歉
Correct it, the trigger name is a woman that I filled in incorrectly
The real tiger name is the name of embedding itself. I have corrected it. Sorry for this mistake
Don't underestimate emb by it's only over a dozen kb. Although this metaphor is not appropriate, early computers were as big as houses, but how big are chips today?
In a way embedding is better than lora with face loyalty restoration, hope more bro try to train your emb.
tongliyaalexzuov1.0
Anastasiya Kvitko is a Russian glamour model and entrepreneur. Kvitko was born and raised in western Russia in the Kaliningrad Oblast and moved to the United States in her late teen years. She first moved to Miami to pursue modeling.
Trained with images from her Instagram account.
To reproduce all the examples, you will need: Deliberate,
Style of Hopa Games, Slot Video Games Style, Ori Art Style Lora, Deep Negative, Negative Facelift
The embedding turned out pretty well, enjoy!
anastasiyakvitko
Keanu Reeves is a Canadian actor.
I think no introduction is necessary for this great great actor. This IT is made with photographs from 2022-23 where it already seems that his pact with immortality has been breaking little by little hahahaha.
To train this IT I have used a dataset of 40 images. 400 steps .
"You’re breathtaking" - Keanu
k34nur33v3
A request of @M17reardon.
Olivia 'Livvy' Dunne is an American artistic gymnast and social media celebrity.
750-step TI trained on a dataset of 14 images with my usual settings. Some of the pics were perhaps a bit lower quality than usual, so there's a little bit of saturation. Remember that you can always reduce the embedding weight - for instance, type something like "(l1vvydunne:0.99)".
Curious about my work process? I have summarized it here.
Appreciate my work? My TIs are free, but you can always buy me a coffee. :)
l1vvydunne
Eden Fines is an Israeli model and social media celebrity.
Update: Version upgraded. Step 850 was actually much better and I should have realised, but I judged too quickly after a couple of very good results with the previous version. Really sorry for the quick replacement, but I hope you'll like this one better. :)
Curious about my work process? I have summarized it here.
Appreciate my work? My TIs are free, but you can always buy me a coffee. :)
ed3nf1
I managed to create a nel TI that i'm happy with and that delivers consistently good results. This was trained on NAI with 4 different data sets on multiple different weights over 35,000 steps with over 400 images.
I know that by now there's an amazing LORA by Rakosz but i prefer TI's and really wanted to make it work.
This is a textual inversion embedding for Nelliel Tu Odelschwanck, or short Nel, from Bleach. Use vslinx_nel
to activate.
Recommended is to use Clip Skip 2 since the embedding was trained on it!
Feedback is very welcomed and i'd love to see your creations with it if you'd like to attach them in a review.
The images were created with different models:
vslinx_nel
If you like this then please buy me a coffee sometime: https://www.buymeacoffee.com/Darkseal
Trained on some PuffPuffPuff
Show some love, review!
Support me on gumroad - https://darkseal.gumroad.com/
See more of my ai art on DeviantArt deviantart.com/darksealstudios
more 3d stuff - https://www.cgtrader.com/darkseal
more pron 3d for poser/daz for making your controlnet - https://www.renderotica.com/content-artists/Darkseal/201
pffpffpff
Tried that embedding, but dosn't turn out as good as i wanted, maybe its to the lack of creating males with SD... :D
But wanted to release just for the fun of it
the_trump
If you like this then please buy me a coffee sometime: https://www.buymeacoffee.com/Darkseal
V3RH03V3N was trained on some Maarten Verhoeven styling. They might hate ai but I love their work. May it live on in any form! I have not really focused on making renders "for" this embedding... I use this more as a "feeling" for lighting and style. I'm sure it has WAY more potential than what I've used it for here in the sample images.
Support me on gumroad - https://darkseal.gumroad.com/
See more of my ai art on DeviantArt deviantart.com/darksealstudios
more 3d stuff - https://www.cgtrader.com/darkseal
more pron 3d for poser/daz for making your controlnet - https://www.renderotica.com/content-artists/Darkseal/201
v3rh03v3n
更正一下,trigger name 是我乱填的 a woman
真正的tirgger name 是embedding本身的名字 我已经更正了。为这个失误抱歉
Correct it, the trigger name is a woman that I filled in incorrectly
The real tiger name is the name of embedding itself. I have corrected it. Sorry for this mistake
炼气期冲击筑基期的一次试水,看后续视情况追加步数。希望有一天可以冲击大乘期境界。
Don't underestimate emb by it's only over a dozen kb. Although this metaphor is not appropriate, early computers were as big as houses, but how big are chips today?
In a way embedding is better than lora with face loyalty restoration, hope more bro try to train your emb.
Special thanks JernauGurgeh
We exchanged a lot of training experiences.
jiajingwenalexv1
Caroline Podlewski is a French artistic director and model. She lives in Paris, France. She has a vast professional experience in different sectors. Her studies and work experiences make her a true modern entrepreneur: Caroline is so attentive to her career that her professional reputation far exceeds that of a model.
If you like my work, a review is more than welcome! Also, follow me on Twitter or Instagram!
Looking for something... spicy? Here's my Patreon!
Also, I recently started a blog where I will gather resources and tutorials about my work here!
Hey everyone! This is my textual inversion of Caroline Podlewski. I used 50 images of her, with an 8700 steps process. Hope you'll like it! To use it in your works:
download the file and place it into stable-diffusion-webui>embeddings folder;
add carpodlewski to your prompt;
keep the sampling steps around 20, and CFG between 7 and 9 to work well. Great results with DPM++ 2M Karras & Edge of Realism. Face restore could help sometimes, especially when the image is not a close-up/portrait;
carpodlewski
The trigger word is: Luxuanxuan
No effect, try using the name of the file as the trigger word
没效果请尝试使用文件的命名作为触发词
我的另一款LORA模型↓
luxuanxuan
shinobukochou
Norman Mark Reedus (born January 6, 1969) is an American actor best known for starring as Daryl Dixon in the AMC horror drama series The Walking Dead. He also starred as Murphy MacManus in the film The Boondock Saints and its sequel The Boondock Saints II: All Saints Day, Scud in Marvel's Blade II, Marco in Deuces Wild, and himself in his AMC TV show Ride with Norman Reedus.
If you like my work, a review is more than welcome! Also, follow me on Twitter or Instagram!
Looking for something... spicy? Here's my Patreon!
Also, I recently started a blog where I will gather resources and tutorials about my work here!
So happy about this one :D Today I present my textual inversion of Norman Reedus. I used 60 public images of him, with a 4700 steps process. Hope you'll like it! To use it in your works:
download the file and place it into stable-diffusion-webui>embeddings folder;
add norreedus69 to your prompt;
keep the sampling steps around 20, and CFG around 7 seems to work well. Great results with DPM++ SDE Karras & Realistic Vision. Face restore could help sometimes, especially when the image is not a close-up/portrait. Check the gallery for some inspiration!
norreedus69
Embedding of Monster Doodles
Usage: monster description, [optional black and white ink], doodle_monsters style
e.g.: a set of cartoon monsters with different expressions, black and white drawing, doodle_monsters
doodle_monsters
Text Inversion of Eliza Scanlen
Australian Actress
Not trained with NSFW content, but seems to be able to generate NSFW images if used with a checkpoint that does it.
3liza2
Jason Voorhees from the Friday the 13th series based his appearance in Part 3. I used 20 images, mostly from the film Friday the 13th Part 3. I also threw in a few images from The Game and of a deluxe figure as well. I was hoping to get more of a consistent look on the hockey mask design, but after multiple repeated attempts I can't make the design perfect. This should work for a general Jason Voorhees look if that's all you're looking for. It might not be painted correctly but you will still get some pretty cool similar designs. Please feel free to tell me why you think it sucks though. I will probably be making a Friday the 13th Part 3 LORA instead, that hopefully will be more accurate. I might even refine this with a newer version at some point, most likely if the LORA doesn't work out. If you actually do enjoy this please help me out by rating it, I can't seem to get any ratings even after hundreds of downloads.
pt3jasonvoorhees
Embedding of Nora Tschirner (german actress)
noratschirner
This TI it's not the best but you can tell who she is, better versions will come as I find the the perfect learning rate.
The trained images is from her 18+.
Use the word celebrity before the trigger
i.e.
celebrity sabrina in AUTOMATIC1111
celebrity <sabrinav1sd15> in InvokeAI
Recommended model: https://civitai.com/models/3855/epic-diffusion
Recomended CFG scale at 11 or 14 except if your sampler starting tearing the face then use something like 9 or 10 but always higher.
Don't forget to do your due prompt work.
sabrinav1sd15
Embedding of Helene Fischer
helenefischer
ALESSANDRA GILIOLI is Fashion model and Instagram influencer from Milan, Italy.
I admit it, redheads lose me and this girl is a real beauty.
To train this IT I have used a dataset of 45 images. 300 steps
Enjoy :D .
(p.d For no reason today his instagram has stopped working...)
al3ssagigioli
Character: Kagari (篝) from 2011 game (or 2016 anime) Rewrite
kagari
This is a style trained on backgrounds and characters from slot video games. You can use it on characters and scenes. Start (or end) your prompt with (slotgames:1.0) style or style by (slotgames:1.0)
Work with this style in the same way as described here.
I recreated some of the popular models' covers so you can see the difference.
If you copy prompts from the comparison images in the examples here, add the style tag manually. Single image examples have it, just copy-paste it all.
Enjoy!
slotgames
If you like this then please buy me a coffee sometime: https://www.buymeacoffee.com/Darkseal
trained on Gantz outfits from multiple source and styles.
Support me on gumroad - https://darkseal.gumroad.com/
See more of my ai art on DeviantArt deviantart.com/darksealstudios
more 3d stuff - https://www.cgtrader.com/darkseal
more pron 3d for poser/daz for making your controlnet - https://www.renderotica.com/content-artists/Darkseal/201
gntz
This is my second model related to Wes Anderson, following the success of my previous Grand Budapest Hotel style model. With this Moonrise Kingdom model, I hope to provide artists and enthusiasts with a new tool to create beautiful and unique artwork in the distinctive style of Wes Anderson's films.
This AI drawing model is based on movie screenshots from Moonrise Kingdom and can be used to create images with a similar style. The model was created using the SD1.5 model and can be used with other checkpoints based on SD1.5. The input image size for this model is 600x320, but standard sizes can also be selected when generating the image. Additionally, it is recommended to add Lora or other embeddings when generating the image to add character representation, as the model does not produce clear depictions of people.
By the way, If you like my work, feel free to buy me a cup of coffee. Thank you very much for your support!
moonrise kingdom
Morena Baccarin is an American/Brazilian actress, known for her roles in Firefly, Deadpool, Homeland and Gotham.
975-step TI trained on a dataset of 15 images with my usual settings.
Remember that, in case you encounter saturation, you can reduce the embedding weight. For instance, type "(morb4cc4r1n:0.98)" in your prompt.
Curious about my work process? I have summarized it here.
Appreciate my work? My TIs are free, but you can always buy me a coffee. :)
morb4cc4r1n
Embedding of Josefine Preuß (german actress)
josefinepreuss
Mila Azul (born January 12th, 1997) is a Ukrainian adult model. Azul started her modeling career in 2016 at age 19 and has done fitness, lingerie, fashion, and nude art modeling, she came into the Adult Entertainment Industry in 2017 at the age of 20 and did her first shot for Film Studio ‘The Score Group’. Till now she has been cast in many scenes and Azul now has more than 50 adult films behind her.
If you like my work, a review is more than welcome! Also, follow me on Twitter or Instagram!
Looking for something... spicy? Here's my Patreon!
Also, I recently started a blog where I will gather resources and tutorials about my work here!
Hello there! Today I am happy to present my textual inversion of Mila Azul. I used 60 images of her, with a 7100 steps process. I wanted to experiment a bit this time because I used only one of her sets. Hope you'll like it! To use it in your works:
like Eva Elfie's textual inversion, this one is also particularly horny 😂 check the examples for some inspiration about the prompt;
download the file and place it into stable-diffusion-webui>embeddings folder;
add milazul97 to your prompt;
keep the sampling steps around 20, and CFG around 7 seems to work well. Great results with DPM++ SDE Karras & Realistic Vision. Face restore could help sometimes, especially when the image is not a close-up/portrait. Check the gallery for some inspiration!
milazul97
This is a Stable Diffusion textual inversion about a great actress. If you wanna test it, you can put this word on the prompt: kthrineLngFrdWhy .
PS: The test images was made with 'vainilla' SD 1.5 model. I don't know if works on SD 2.1 .
If you wanna support me, you can buy me a coffee: https://www.buymeacoffee.com/elrivx
kthrinelngfrdwhy
sablynv1
Embedding of Palina Rojinski (german celebrity and TV Host)
palinaroj
Embedding of Lena Meyer-Landrut (german singer and celebrity)
lenaml
Nicola Cavanis is a German model and social media influencer with more than 1M followers on Instagram.
This TI has been updated! Please check the "About this version" section for more info.
Remember that, in case you encounter saturation, you can reduce the embedding weight. For instance, type "(nc4v4n1s:0.99)" in your prompt.
Curious about my work process? I have summarized it here.
Appreciate my work? My TIs are free, but you can always buy me a coffee. :)
nc4v4n1s
WiP Concept of CCTV like images
Will probably fine-tune, but as for now its a Concept
style_cctv
The characters in the second version will have more artistic conception.
i hope you will like it.
If the character is very strange, you can try using my prompts.
oldstyle_v1:
When creating a character, there will be a black and white photo effect. If not, you can add "greyscale" to increase the weight.
old_v1_6500:
Some new photos of cats have been added to make the scene more natural. Smoking characters would look better.
cigarettes:
cat:
TIPS: This embedding is suitable for changing the scene style to make the image more vintage. If specific poses or characters are required, they need to be used in conjunction with Lora.
oldstyle_v2
When activating this five vector negative embedding in your negative prompt it should help you attain an anime style more easily when using the checkpoint DreamShaper 4. The activation word by default is "anime-style-negative-embedding", but I recommend you change the file name to whatever word you want, as the filename is the activation word.
The embedding is probably fine to use on a variety of checkpoints, but the training images were generated on the DreamShaper 4 checkpoint and the embedding was also trained on it, so that is where to expect the best results.
The training images were generated with the use of the bad-artist and bad-artist-anime negative embeddings- they are good negative embeddings. They are at this hugging face repository if you want to download them.
anime-style-negative-embedding
This is a model of the character kroos in arknights that I created a few months ago and uploaded for everyone to use.(这是我在几个月前所制作的明日方舟中的角色克洛斯的模型,上传上来供大家使用。)
注意事项:在过多提示词的情况下可能会无法正确显示(比如缺少兔耳,发色不对等)。
kroos
This embedding will invoke the works of an architect, who has clearly lost all of his marbles. Nightmarish city scapes, grim and dark, with buildings glued together in chaotic manner, defying laws of physics and common sense.
Use it spice up your prompt, for best effect you will need to balance the weight of the token, check prompts in example images (these use Invoke-AI syntax). I particularly liked mixing it with city of Venice.
The embedding will introduce these properties: chaotic 'glued cubicles' architecture, odd angles, darkness, wires, ink drawing style and for some reason old style lanterns. For me it worked nicely with realistic models, particularly these two:
https://civitai.com/models/21616/neurogen-v10
https://civitai.com/models/21813/edge-of-realism
It works in high resolution, without the fix (it's chaos anyway). Example images are not up-scaled.
Have fun.
Character: Nakamura Yuri (仲村 ゆり) from anime Angel Beats!
yuri
This is a TI for actress Phoebe Cates.
2500K steps with 43 images
Trigger: pho3b3cat3s
pho3b3cat3s
khorhaiya
TANYA MAVRINA is a Russian instagram model.
The truth is that she is a fairly unknown model, but with an unparalleled "Russian doll" beauty.
To train this IT I have used a dataset of 21 images. 250 steps. My goal this time was to reduce the number of images in the data set to almost half of the ones I've used so far and keep a decent result since a lot of time is spent preprocessing the images and optimizing a bit.
Enjoy it.
t4ny4m4vr1n4
A request of @gw2herb in my buymeacoffee website.
Anllela Sagra is a Colombian fitness model and social media celebrity.
Really sorry for the quick update, guys. Should have checked step 500 before I uploaded the previous one! Please check the "About this version" section for more info.
Remember that, in case you encounter saturation, you can reduce the embedding weight. For instance, type "(anllesagra:0.98)" in your prompt.
Curious about my work process? I have summarized it here.
Appreciate my work? My TIs are free, but you can always buy me a coffee. :)
anllelas4gr4
Honey Gold
Magnificent performer Honey Gold is an absolute goddess. No level of artificial intelligence could possibly do her justice, so in the meantime we use stable diffusion and dream of a better tomorrow.
Do you like my work? Buy me a beer!
[ SDv1.4, 5e-03, 3400 ]
ALSO! If anyone comes up with a good prompt for her short curly hair (short curly hair isn't sufficient apparently) please let me know!
honeygold
Rina Ellis (born April 15, 1993), also known as Little Miss Rina, was born in Seoul, South Korea. Making her adult entertainment debut in 2016 around the age of 23, she quickly found work for studios and websites including Brazzers, Digital Playground, Babes.com, and Reality Kings.
If you like my work, a review is more than welcome! Also, follow me on Twitter or Instagram!
Looking for something... spicy? Here's my Patreon!
Also, I recently started a blog where I will gather resources and tutorials about my work here!
Hey nice people! This is my textual inversion of Rina Ellis. I used 60 public images of her, with a 5200 steps training process. She's one of my favorites ever... Hope you'll like it! To use it in your works:
like Eva Elfie's textual inversion, this one is also particularly horny 😂 check the examples for some inspiration about the prompt;
download the file and place it into stable-diffusion-webui>embeddings folder;
add rinellis93 to your prompt;
keep the sampling steps around 20, and CFG around 7 seems to work well. No face restore and great results with DPM++ SDE Karras & Realistic Vision. Check the gallery for some inspiration!
rinellis93
Adriana Chechik
She left us too soon...having fractured her back in 2 places she's likely no longer in the business. She's successfully made the transition to streaming, however, so here's to a full recovery!
Do you like my work? Buy me a beer!
[ SDv1.4, 5e-03, 4000 ]
achechik
This textual inversion is not focused on Joker as a subject, but rather on visual style, colors, and lighting. There are two ways to use it:
Generate images of men (it picked up actors from the movie) or interiors and city landscapes. Examples above.
Add it as a secondary style to your main style to add a variety of colors and lighting to the image. Better use, a few examples shared.
To use it, download it to your embeddings folder in Web UI installation and start the prompt with jokermovie style or style of jokermovie.
Enjoy and share the results in comments!
jokermovie
Enbedding of Sarah Engels
If to strong, adjust the weighting: (sarah3ngels:0.9)
sarah3ngels
This Textual Inversion is trained publicly available images of dense forest with Stable Diffusion 1.5 with 6 images at scale of 512x512.
Works best with DPM++ 2M Karras.
Tested on SD 1.5 and hasdx_hasedsdx (NSFW)
Take a deep breath and immerse yourself in the rich, verdant atmosphere of a lush meadow, where the air is thick with the scent of blooming flowers and grass,dense-forest-5000
As you hike along the verdant forest trail, take a moment to appreciate the canopy of leaves overhead, filtering the sunlight into a kaleidoscope of green,dense-forest-5000
Stand at the edge of a vast, emerald-green lake and take in the tranquil beauty of the surrounding forest, where trees tower high and vibrant foliage abounds, Tranquility, Serenity, Biodiversity, Wildlife, Foliage, Canopy, Undergrowth, Ecosystem, Flora, Fauna, dense-forest-5000
Picture yourself relaxing in a hammock, swaying gently under the leafy canopy of a sprawling tree, surrounded by verdant greenery and the sounds of nature, Tranquility, Serenity, Biodiversity, Wildlife, Foliage, Canopy, Undergrowth, Ecosystem, Flora, Fauna, dense-forest-5000
In the midst of a bustling city, find refuge in a verdant urban oasis, where greenery abounds and the rustling of leaves provides a soothing soundtrack to your day, Tranquility, Serenity, Biodiversity, Wildlife, Foliage, Canopy, Undergrowth, Ecosystem, Flora, Fauna, dense-forest-5000
Take a deep breath and inhale the crisp, fresh air of a mountain meadow, where the vibrant green grass stretches as far as the eye can see and the sky is a deep, endless blue, Tranquility, Serenity, Biodiversity, Wildlife, Foliage, Canopy, Undergrowth, Ecosystem, Flora, Fauna, dense-forest-5000
Picture yourself wandering through a winding, verdant labyrinth of hedgerows and shrubs, where every turn leads to a new discovery in the world of greenery, Tranquility, Serenity, Biodiversity, Wildlife, Foliage, Canopy, Undergrowth, Ecosystem, Flora, Fauna, dense-forest-5000
Stand at the edge of a steep cliff, looking out over a vast, rolling landscape of green fields, where the tall grasses sway gently in the breeze,Tranquility, Serenity, Biodiversity, Wildlife, Foliage, Canopy, Undergrowth, Ecosystem, Flora, Fauna, dense-forest-5000
Find yourself lost in the lush, green wilderness of a tropical rainforest, where towering trees, exotic foliage, and an abundance of wildlife await, Tranquility, Serenity, Biodiversity, Wildlife, Foliage, Canopy, Undergrowth, Ecosystem, Flora, Fauna, dense-forest-5000
Train Further or Fork on GitHub https://github.com/Ygohel18/dense-forest
Follow Me On Social Media
dense-forest
Emily O'Hara Ratajkowski (born June 7, 1991) is an American model, actress, and author. Born in London to American parents and raised in Encinitas, California, her modeling debut was on the cover of the March 2012 issue of the erotic magazine treats! which led to her appearing in Maroon 5's music video "Love Somebody".
If you like my work, a review is more than welcome! Also, follow me on Twitter or Instagram!
Looking for something... spicy? Here's my Patreon!
This is a textual inversion of the amazing Emily Ratajkowski. I used 40 public images of her, with a 5300 steps training process. Hope you'll like it! To use it in your works:
download the file and place it into stable-diffusion-webui>embeddings folder;
add emrata91 to your prompt;
keep the sampling steps around 30, CFG between 7 and 8 seems to work well. Very good results with DPM++ SDE Karras & Realistic Vision. Check the gallery for some inspiration!
emrata91
Again the SD version of her isnt all that bad but I wanted to preserve her likeness from Into the Blue, Fantastic 4 and Sin City. When she was one of the hottest girls in business. So this TI lets her stay forever 24.
Trigger word is jessicaalba_ti
v1 was trained with 10 images for 3000 steps
jessicaalba_ti
Embedding of Alexis Bledel
If to strong, adjust the weighting: (alexisbl3del:0.9)
alexisbl3del
Character: Nakatsu Shizuru (中津 静流) from 2011 game (or 2016 anime) Rewrite
shizuru
Character: Kamio Misuzu (神尾 観鈴) from the game (or the anime) Air
misuzu
Another fix for a celebrity I really care about, SD really hates her nose for some reason. I trained on 10 images from mostly 2005 - 2007 so right when Aeon Flux came out. But there are also some older pics from Atomic Blonde present, so sometimes you get the young Charlize and sometimes you get the older one. Maybe I will split these into 2 distinct versions in the future.
Trigger word is charlizetheron_ti
Lowering the weight to 0.98 seems to help in some models.
v1 was trained with 10 images for 2500 steps.
charlizetheron_ti
Embedding of Faith from the video game Mirror's Edge
me_faith
Character: Tachibana Kanade (立華 かなで) from anime Angel Beats!
kanade
Kaia is an American supermodel with a substantive Instagram following. I can't help but think she looks an awful lot like Cindy Crawford, no wonder she's so popular.
Do you like my work? Buy me a beer!
[ SDv1.4, 5e-03, 3500 steps ]
kaiagerber
Impossibly beautiful goddess Jourdan Dunn. If you haven't heard of or seen her before, you can send me money below, you're very welcome. I've never created an embedding where every single image I generated for the examples were perfect. Just, wow.
Do you like my work? Buy me a beer!
[ SDv1.4, 5e-03, 4000 steps ]
jourdandunn
keyword: 0xwpap
Prompts to try:
0xwpap, portrait of a man, shades, swaggy
0xwpap, portrait of a fantasy fox, funny
0xwpap, portrait of a cute fantasy cat, cute, infectious
0xwpap, portrait of a beautiful woman, princess, leader
0xwpap
Instagram superstar and bikini model Sierra Skye. I mean that last one, having seen thousands of pictures and she's never been in anything but a bikini. No complaints, however. Did you know she's Dakota's younger sister?
Do you like my work? Buy me a beer!
[ SDv1.4, 5e-03, 3500 steps ]
sierraskye
veryBadImageNegative_v1.3使用了新的训练图集,在AOM3和viewer-mix_v1.7中使用时表现较好。
veryBadImageNegative_v1.3 uses a new training atlas and performs well when used in AOM3 and viewer-mix_v1.7.
veryBadImageNegative是使用viewer-mix_v1.3生成的特殊图集训练而来的负面嵌入。
所以veryBadImageNegative是viewer-mix_v1.3的专用负面嵌入。
或许在其他的扩散模型中也有不错的效果,但缺乏验证。
veryBadImageNegative is a negative embedding trained from the special atlas generated by viewer-mix_v1.3.
So veryBadImageNegative is the dedicated negative embedding of viewer-mix_v1.3.
It may also have a good effect in other diffusion models, but it lacks verification.
关于v1.2:
增强图像的质量,削弱了风格。
在使用v1.2版本时,可以降低一些权重以获得更好的灵活性。
例如:(veryBadImageNegative_v1.2-6400:0.9)
About v1.2:
Enhance image quality and weaken style.
When using v1.2, you can reduce some weights to obtain better flexibility.
For example: (veryBadImageNegative_v1.2-6400:0.9)
关于v1.0与v1.1的差异:
几乎没有差别,仅用于训练的图集在生成参数有一些微妙的差别。
关于1600~6400的差异:
数字代表了训练步数,理论上步数越多效果越好。
About the difference between v1.0 and v1.1:
There is almost no difference. The atlas used for training only has some subtle differences in the generation parameters.
About the difference between 1600 ~ 6400:
The number represents the number of training steps. In theory, the more steps, the better the effect.
verybadimagenegative_v1.3
Eva Elfie (May 27, 2000) is a Russian adult model and glamour model, best known for her work with Playboy Plus. She has been featured in Playboy editorial spreads and has appeared on the cover of the magazine. She is also active on her personal Instagram account, where she shares photos of her modeling work and her personal life.
If you like my work, a review is more than welcome! Also, follow me on Twitter or Instagram!
Looking for something... spicy? Here's my Patreon!
Also, I recently started a blog where I will gather resources and tutorials about my work here!
Another one for today! A textual inversion of Eva Elfie. I used 70 public images of her, with a 6300 steps training process. Hope you'll like it! To use it in your works:
this one is particularly horny 😂 check the examples for some inspiration about the prompt;
download the file and place it into stable-diffusion-webui>embeddings folder;
add evaelfie00 to your prompt;
keep the sampling steps around 30, CFG between 7 and 8 seems to work well. Great results with DPM++ SDE Karras & Realistic Vision. Check the gallery for some inspiration!
Have fun!
evaelfie00
TERRY CREWS its a Nort American actor, bodybuilder, comedian and voice actor.
Probably one of my favorite actors, especially in the series "Brooklyn Nine-Nine" as Sergeant Terry Jeffords.
You can't always live by seeing sexy girls and here we have this great guy who in my opinion is great.
To train this IT I have used a dataset of 45 images. 350 steps were enough for an optimal result.
Terry 4 ever :D
tcr3ws
Ivana Knöll is a Croatian model and social media celebrity, who became known as the "girlfriend of the Qatar World Cup" in 2022.
400-step TI trained on a dataset of 19 images. Fixed learning rate of 0.004. Images were not upscaled this time either, which seems to be helping somehow to avoid early saturation (I'll probably be proved wrong in my next TI, though xD).
Curious about my work process? I have summarized it here.
Appreciate my work? My TIs are free, but you can always buy me a coffee. :)
1vanakn0ll
Do you remember Megan Fox from 2007 when Transformers came out? I certainly do and I dont know what she did to her face but this TI is trained only on images from 2007-2009, so pre-surgery.
So as you can imagine the resources are limited and its really hard to find images without her signature duckface. So putting "open mouth" or "open lips" into negatives can help here and there but its in the training data so hard to get rid of.
HOWEVER I think its a significant improvement over the Megan Fox that normal SD spits out.
Trigger word is meganfox_ti
v1 was trained with 7 images for 2500 steps
meganfox_ti
Textual inversion model based on Juan Ignacio Delgado Alemany aka Ignatius Farray, Spanish comedian, writer, stand-up comedian, actor, screenwriter and singer.
ignatius farray
Hand-drawn video game style that will add strong, black outlines to the objects in front and blur objects in the background to give depth to the image.
You can work with this style in the same fashion as described in this tutorial.
The style is so strong when used with 1.0 strength that the choice of the model is only a matter of composition and responsiveness to prompting.
The best results from the 8 models I tested with come from RealisticVision, Deliberate, CheeseDaddy'sLandscapeMixOffset.
Use a few of your favorite negative embeddings and general negative tags.
positive prompt
(Award-winning photograph), (Japanese garden), (Autumn foliage), (Vibrant colors), (Fallen leaves), (Nature's beauty), (Red and gold hues), (Maple trees), (Reflections on water), (Peaceful tranquility), (Traditional wooden architecture), (Rustic bridges), (Soothing natural sounds), (Seasonal delight), (Serene garden views), (Low-angle perspective), (natural lighting), Nikon F2 Camera, hidden objects games, video game concept art, (8K Unity wallpaper), fine details, award-winning image, highly detailed, 16k, cinematic perspective, ((video game environment concept art style)), pretty colors, cinematic environment, (handmade2d:1.0) style
negative prompt
ng_deepnegative_v1_75t, bad-artist, bad-hands-5, (worst quality, low quality:1.4), multiple views, multiple panels, blurry, watermark, letterbox, text, character, human, people, 1girl, 1boy, NSFW, nude, signature, medium quality, deleted, lowres, comic, cleavage, frame, watermark, signature,watermark,signature
settings
CFG 7 and 50 steps on any model or the ones listed above
handmade2d
William Bradley Pitt (born December 18, 1963) is an American actor and film producer. He is the recipient of various accolades, including two Academy Awards, two British Academy Film Awards, two Golden Globe Awards, and a Primetime Emmy Award. As a public figure, Pitt has been cited as one of the most powerful and influential people in the American entertainment industry.
If you like my work, a review is more than welcome! Also, follow me on Twitter or Instagram!
Looking for something... spicy? Here's my Patreon!
Also, I recently started a blog where I will gather resources and tutorials about my work here!
Today, I am happy to present a textual inversion of Brad Pitt. He really needs no introductions, right? :) I used 70 public images of him, with an 8000 steps training process. I am pretty satisfied with the results. Hope you'll like it! To use it in your works:
download the file and place it into stable-diffusion-webui>embeddings folder;
add brapit63 to your prompt;
keep the sampling steps around 30, CFG at 7 seems to work well. Great results with DPM++ SDE Karras & Realistic Vision. Check the gallery for some inspiration!
Enjoy!
brapit63
keyword: 0xisometric
Prompts to try:
0xisometric, bedroom, yellow, bright room
0xisometric, gaming room, red, LED room
0xisometric, glowing temple
0xisometric, staircase to the end portal
follow me here for new models! ^.^
https://twitter.com/0xaiwhisperer
0xisometric
Keep in mind that the generation could vary signicantly depending on the model and the sampler you are using. Some models simple don't work at all while others work perfecly. Also higly recomended to use CFG scale at around 11.
Use the words woman tulsi as trigger in AUTOMATIC1111
Use woman <tulsiv1sd15> in InvokeAI
Recomended model: https://civitai.com/models/3855/epic-diffusion
Training details are in the version notes.
Keep your comments clean.
Feedback is very welcome.
tulsiv1sd15
kagua_wrise
keyword: 0xaisticker
Prompts to try:
0xaisticker, fantasy fox, funny face, cute, charming, captivating
0xaisticker, cute fantasy monster, glowing yellow eyes
0xaisticker, cryptocurrency logo
0xsticker, funny cat face
follow me here for new models! ^.^
https://twitter.com/0xaiwhisperer
0xaisticker
roryg
Yet another goddess that I'm surprised isn't represented here yet. Arguably the most beautiful person to have ever existed ever in the world of all time.
Do you like my work? Buy me a beer!
[ SDv1.4, 5e-03, 3500 steps, 0.197 vector str. ]
adrianalima
Jessica Bartlett is an American model and social media celebrity.
TI updated. Please check the "About this version" section for more info on training.
Important note: This TI works better with a reduced weight! Type "(jernau_bartlett:0.98)" in your prompt, it'll get rid of the saturation in most cases.
Curious about my work process? I have summarized it here.
Appreciate my work? My TIs are free, but you can always buy me a coffee. :)
jernau_bartlett
American actress Yara Shahidi, strikingly gorgeous, I was introduced to her recently and felt compelled to create this so I could share my admiration with you animals.
Do you like my work? Buy me a beer!
[ SDv1.4, 3000 steps, 5e-03]
yarashahidi
Raven Lyn Corneil (born January 9, 1995) is a well-known professional model, managed by Women 360 Management New York, Freedom Los Angeles, and Storm Management London. She is simply referred to as "Raven Lyn" in the industry. Her commercial work includes G-Star Raw, Adidas, and Gap.
If you like my work, a review is more than welcome! Also, follow me on Twitter or Instagram!
Looking for something... spicy? Here's my Patreon!
Also, I recently started a blog where I will gather resources and tutorials about my work here!
Hey, people of CivitAI! This is a textual inversion of Raven Lyn Corneil. She's one of my favorite models, I could not exclude her from my work! I went back to the "basics" this time: 3000 steps and 40 public images were enough. As usual: do not use the face restore.
To use it:
download the file and place it into stable-diffusion-webui>embeddings folder;
add ravlyn95 to your prompt;
keep the sampling steps around 25, CFG at 7 seems to work well. Best results with DPM++ 2S a Karras + Realistic Vision. Also got some good results with Deliberate (last pic). Check the gallery for some inspiration!
ravlyn95
I've been working on this for a long while and it finally dawned on me to remove from the source images the ones where she's not smiling. This has virtually eliminated all of the derpiness from her look and I think it's finally where I want it to be.
I ran 5,000 steps and have settled on the 3200-step version.
I also built her on v1.4 and I can't complain - as usual she looks best on all of the nicer Models available here.
Important Notes: Sometimes she'll gen cross-eyed, no idea why, she ain't cross-eyed in any of the source photos. Second, if you want a smile you gotta prompt it.
violetsummers
This TI was trained with the most recent photos of adult 18+ Greta.
Place the activist word befor her trigger
activist <gretav2sd15> in InvokeAI
activist greta in AUTOMATIC1111
Recomended model: https://civitai.com/models/3855/epic-diffusion
Keep in mind that the generation could vary signicantly depending on the model and the sampler you are using. Some models simple don't work at all while others work perfecly. Also higly recomended to use CFG Scale at around 11.
gretav2sd15
Hobbit house-inspired interiors, with large windows, nice views, luxury furniture, and dark wood and leather material.
Positive prompt
(Professional 3D rendering:1.3) of (Ultrarealistic:1.3) magicalinterior style, bathroom, bathing room, plant, scenery, window, tree, nature, bathtub, forest, mirror, bath, faucet, water, house interior, indoors, cottage interior, treehouse interior, dark, gothic, Goth, luxury, fantasy, magic, Harry Potter, Slytherin, perfect composition, award-winning,CGSociety,ArtStation
Negative prompt
ng_deepnegative_v1_75t, bad-artist, (worst quality, low quality:1.4), multiple views, multiple panels, blurry, watermark, letterbox, text, character, human, people, 1girl, 1boy,((Wireframe)),Polygons,Screenshot,Character design,Software,UI,watermark,signature
Example Settings
768x512 px canvas size, CFG scale 7 and 70 steps on Deliberate.
[your medium], magicalinterior style, [Subject], [Descriptors]
Medium
You can use "professional 3d rendering" or set Image Type to Random in StylePile (highly recommended extension).
Subject
Use one of these subjects or copy-paste them to the [X] field of Style Pile and use [X] in your prompt formula instead of [Subject]. You can also generate your original subjects in ChatGTP, using the workflow described here.
kitchen, kitchen decor, kitchen design, window, scenery, plant, table, bottle, sunlight, chair
bathroom, bathing room, plant, scenery, window, tree, nature, bathtub, forest, mirror, bath, faucet, water
living room, livingroom, plant, candle, window, scenery, couch, table, candlestand, cup, potted plant, teapot, sleeping, chair, stained glass
potions lair, candle, city, scenery, window, fantasy, witchcraft, alchemy, alchemist lab, candlestand, bottle, chandelier, table, shelf
library, no scenery, books, bookshelf, window, plant, tree, lamp, table, candle, flower, chair
Descriptors
The last part of the positive prompt is descriptive tags. Use one of these or make your own:
house interior, indoors, cottage interior, treehouse interior, dark, gothic, Goth, luxury, fantasy, magic, Harry Potter, Slytherin, Lord of the rings, LOTR
house interior, indoors, cottage interior, treehouse interior, light, bright, luxury, fantasy, magic, Harry Potter, Slytherin, perfect composition, award-winning
house interior, indoors, cottage interior, treehouse interior, dark, gothic, Goth, luxury, fantasy, magic, Harry Potter, Slytherin, perfect composition, award-winning, Dirk Crabeth, unreal engine 5 highly rendered, a detailed matte painting, gothic art
Negative prompt
Use two of your favorite negative embeddings (in my case Deep Negative and Bad Artist) plus negative words from the model you are using. My case:
ng_deepnegative_v1_75t, bad-artist, (worst quality, low quality:1.4), multiple views, multiple panels, blurry, watermark, letterbox, text, character, human, people, 1girl, 1boy
Recommended models
Good results on Deliberate, The Allys Mix III, and Surreal Steampunk AI (without trigger words). Style fails to deliver with Anime models, such as Grapefruit.
Upscaling
Examples are not upscaled but you can do it with a denoising of 0.2 and your favorite upscaler in IMG2IMG.
magicalinterior
Lindsay used to be a child actress way back in the day, now acting as an adult, too. Which is where all of the source images came from, as an adult. Over 18 for those in the States. All portrayals of Miss Lohan are of a legal age likeness. Can't be more clear than that.
Anyway, I tries something a bit different and ran this for 20,000 steps at a rate of 2e-03 so I could keep an eye on the progress, stop, and restart if needed. This embedding is the 15,500 step version but don't worry, vector strength is .198
lindsaylohan
KIARA MOON its a Nort American OnlyFans and Instagram model.
To train this IT I have used a dataset of 49 images. 300 steps were enough for an optimal result. I would say it is the best IT I have created so far, very happy with it.
I hope you enjoy this tattooed angel with asian features. :D
kiaram00n
Another old project from my archive. I'm aware there's already another Daisy Ridley Embedding by Balbrig, but I figured the results are different enough due to differences in dataset and training settings that it warrants an upload. If you haven't tried Balbrig's embedding yet, be sure to get it as well!
The embed has been trained on SD 1.5 and works well on most 1.5 based models. When rendering full body images I recommend using inpainting. It also works well to let the AI use known information from the model by prompting for [Daisy Ridley|Koh_DaisyRidley] or [Daisy Ridley:Koh_DaisyRidley:0.8]. This allows the AI more creativity and it will get more varied outputs, it's worth trying.
If you enjoy my creations, consider buying me a coffee on Ko-Fi. As I do consume a lot of coffee while creating embeddings, it'd be much appreciated.
koh_daisyridley
I used a bunch of pics from old 1980's anime or renditions of them. A lot of it was sci-fi and gundam so expect there to be mech and sci-fi elements in the output.
80sanime
I trained this embedding on images from Deadpool 2 and some concept art and couple comic style shots of the X-men character, Cable. It gives some results I'm satisfied with. Enjoy!
brolincable
A request of @463259395282.
Alissa Violet is an American model, actress and YouTuber.
This TI has been updated. Please check the "About this version" section below for more info on its training.
Curious about my work process? I have summarized it here.
Appreciate my work? My TIs are free, but you can always buy me a coffee. :)
alissav10l3t
Embedding of Jennifer Lawrence
jflw
Salma Hayek Pinault (September 2, 1966) is a Mexican and American actress and film producer. She soon established herself in Hollywood with appearances in films such as Desperado (1995), From Dusk till Dawn (1996), Wild Wild West (1999), and Dogma (1999)... and a lot more.
If you like my work, a review is more than welcome! Also, follow me on Twitter or Instagram!
Looking for something... spicy? Here's my Patreon!
Also, I recently started a blog where I will gather resources and tutorials about my work here!
Hey everyone! Today I present you a textual inversion of Salma Hayek. I wanted to experiment with a specific time window: 7900 steps and 90 public images, but only in the 2010-2019 years. Suggestion: do not use the face restore.
To use it:
download the file and place it into stable-diffusion-webui>embeddings folder;
add salhayek66 to your prompt;
Keep the Samples around 25, CFG between 5 and 7 seems to work good. As usual, feel free to experiment;
salhayek66
This is a pretty straight forward model of the actor Jeffrey Dean Morgan in more recent years. Pretty much all of the pictures has facial hair so that is likely on most of the results.
jeffreydm