metadata

inference: true
language:
  - en
tags:
  - stable-diffusion
  - text-to-image
license: creativeml-openrail-m

@akumetsu971)

Model used to train:

wd-v1-3-full-opt.ckpt (https://huggingface.co./hakurei/waifu-diffusion-v1-3)

Files

3 files available (Best version is V2):

-VCM07_style - 4000 steps (more focused on girl)

-VCM07_style2 - 4000 steps (allowed to create animals)

-Prompt_Blending Script (optional, used for prompt)

Prompt

You need to use DeepDanBooru Tags (https://gigazine.net/gsc_news/en/20221012-automatic1111-stable-diffusion-webui-deep-danbooru/)

Elysium_Anime_V2.ckpt (https://huggingface.co./hesw23168/SD-Elysium-Model)

Prompt_blending script (https://huggingface.co./Akumetsu971/SD_VCM07_Anime_Style/tree/main)

Embedding was trained with images of girls only. Therefore, getting a boy can be difficult. Adjust weight, negative prompt, etc...

Human Example

Positive Prompt:

(VCM07_style2:1.0), (1girl:1.2), looking_at_viewer, (best quality), (masterpiece:1.2), (ultra-detailed),(official art),(an extremely delicate and beautiful), (attractive:1.2), (beautiful detailed eyes), (dynamic colours, vibrant colours), depth of field, god rays, dynamic lighting Negative Prompt:

(mediocre:1.2), (average:1.2), (bad:1.2), (wrong:1.2), (error:1.2), (fault:1.2),( badly_drawn:1.2), (poorly_drawn:1.2), ( low_quality:1.2), no_quality, bad_quality, no_resolution, low_resolution, (lowres:1.2), normal_resolution, (disfigured:1.6), (deformed:1.4), (distortion:1.2), bad_anatomy, (no_detail:1.2), low_detail, normal_detail, (scribble:1.2), (rushed:1.2), (unfinished:1.2), blur, blurry, claws, (misplaced:1.2), (disconnected:1.2), nonsense, random, (noise:1.2), (deformation:1.2), 3d, dull, boring, uninteresting, screencap, (text:1.2), (frame:1.1), (out_of_frame:1.2), (title:1.2), (description:1.3), (sexual:1.2), text, error,(logo:1.3), (watermark:1.3), bad_perspective, bad_proportions, cinematic, jpg_artifacts, jpeg_artifacts, extra_leg, missing_leg, extra_arm, missing_arm, long_hand, bad_hands, (mutated_hand:1.2), (extra_finger:1.2), (missing_finger:1.2), broken_finger, (fused_fingers:1.2), extra_feet, missing_feet, fused_feet, long_feet, missing_limbs, extra_limbs, fused_limbs, claw, (extra_digit:1.2), (fewer_digits:1.2), elves_ears, (naked:1.3), (wet:1.2), uncensored, (long_neck:1.2), (weapon:1.5)

Animals Example

For V2, embedding was trained with dogs, cats, foxes. Therefore, it is easier to get these animals. However, it is possible to get frogs, elephants, tigers, lions, etc... I used a method with blend prompt script then I described the anatomy of the animal: eyes, ears, nose, fur, etc...

Positive Prompt:

(VCM07_style2:1.0), (cat:1.4|dog:0.5|fox:0.5), (cat_nose:1.2), (cat_eyes:1.2), (cat_ears:1.2), (solo:1.2), looking_at_viewer, (best quality), (masterpiece:1.2), (ultra-detailed),(official art),(an extremely delicate and beautiful), (attractive:1.2), (beautiful detailed eyes), (dynamic colours, vibrant colours), depth of field, god rays, dynamic lighting Negative Prompt:

(mediocre:1.2), (average:1.2), (bad:1.2), (wrong:1.2), (error:1.2), (fault:1.2),( badly_drawn:1.2), (poorly_drawn:1.2), ( low_quality:1.2), no_quality, bad_quality, no_resolution, low_resolution, (lowres:1.2), normal_resolution, (disfigured:1.6), (deformed:1.5), (distortion:1.2), bad_anatomy, (no_detail:1.2), low_detail, normal_detail, (scribble:1.2), (rushed:1.2), (unfinished:1.2), blur, blurry, claws, (misplaced:1.2), (disconnected:1.2), nonsense, random, (noise:1.2), (deformation:1.2), 3d, dull, boring, uninteresting, screencap, (text:1.2), (frame:1.1), (out_of_frame:1.2), (title:1.2), (description:1.3), (sexual:1.2), text, error,(logo:1.3), (watermark:1.3), bad_perspective, bad_proportions, cinematic, jpg_artifacts, jpeg_artifacts, extra_leg, missing_leg, extra_arm, missing_arm, long_hand, bad_hands, (mutated_hand:1.2), (extra_finger:1.2), (missing_finger:1.2), broken_finger, (fused_fingers:1.2), extra_feet, missing_feet, fused_feet, long_feet, missing_limbs, extra_limbs, fused_limbs, claw, (extra_digit:1.2), (fewer_digits:1.2), elves_ears, (naked:1.3), (wet:1.2), uncensored, (long_neck:1.2)