metadata
pipeline_tag: text-to-image
license: other
license_name: faipl-1.0-sd
license_link: LICENSE
tags:
- text-to-image
- diffusers
prior:
- Disty0/sotediffusion-v2-prior
SoteDiffusion V2
An anime diffusion model finetuned on Würstchen V3.
Release Notes
- This release is sponsored by fal.ai/grants
- Trained on 12M text & image paris including WD tags and natural language captions for a single epoch on 8xH100 80GB SXM5 GPUs.
- Trained with Full FP32 and MAE Loss.
Code Example
pip install diffusers
import torch
import diffusers
device = "cuda"
dtype = torch.float16
model_path = "Disty0/sotediffusion-v2"
def get_timestep_ratio_conditioning(t, alphas_cumprod):
s = torch.tensor([0.008]) # diffusers uses 0.003 while the original is 0.008
clamp_range = [0, 1]
min_var = torch.cos(s / (1 + s) * torch.pi * 0.5) ** 2
var = alphas_cumprod[t]
var = var.clamp(*clamp_range)
s, min_var = s.to(var.device), min_var.to(var.device)
ratio = (((var * min_var) ** 0.5).acos() / (torch.pi * 0.5)) * (1 + s) - s
return ratio
pipe = diffusers.AutoPipelineForText2Image.from_pretrained(model_path, text_encoder=None, torch_dtype=dtype)
# diffusers bugs
pipe.prior_pipe.get_timestep_ratio_conditioning = get_timestep_ratio_conditioning
pipe.prior_pipe.scheduler.config.clip_sample = False
# de-dupe
pipe.decoder_pipe.text_encoder = pipe.text_encoder = None # nothing uses this
del pipe.decoder_pipe.text_encoder
del pipe.prior_prior
del pipe.prior_text_encoder
del pipe.prior_tokenizer
del pipe.prior_scheduler
del pipe.prior_feature_extractor
del pipe.prior_image_encoder
pipe = pipe.to(device, dtype=dtype)
pipe.prior_pipe = pipe.prior_pipe.to(device, dtype=dtype)
prompt = "1girl, solo, looking at viewer, open mouth, blue eyes, medium breasts, blonde hair, gloves, dress, bow, hair between eyes, bare shoulders, upper body, hair bow, indoors, elbow gloves, hand on own chest, bridal gauntlets, candlestand, smile, rim lighting, from side, castle interior, looking side,"
quality_prompt = "very aesthetic, best quality, newest"
negative_prompt = "very displeasing, displeasing, worst quality, bad quality, low quality, realistic, monochrome, comic, sketch, oldest, early, artist name, signature, blurry, simple background, upside down,"
num_images_per_prompt=1
# Encode prompts and quality prompts eperately:
# device, batch_size, num_images_per_prompt, cfg, prompt
prompt_embeds, prompt_embeds_pooled, _, _ = pipe.prior_pipe.encode_prompt(device, 1, num_images_per_prompt, False, prompt=prompt)
quality_prompt_embeds, _, _, _ = pipe.prior_pipe.encode_prompt(device, 1, num_images_per_prompt, False, prompt=quality_prompt)
negative_prompt_embeds, negative_prompt_embeds_pooled, _, _ = pipe.prior_pipe.encode_prompt(device, 1, num_images_per_prompt, False, prompt=negative_prompt)
empty_prompt_embeds, _, _, _ = pipe.prior_pipe.encode_prompt(device, 1, num_images_per_prompt, False, prompt="")
empty_prompt_embeds = torch.nn.functional.normalize(empty_prompt_embeds)
prompt_embeds = torch.cat([prompt_embeds, quality_prompt_embeds], dim=1)
negative_prompt_embeds = torch.cat([negative_prompt_embeds, empty_prompt_embeds], dim=1)
pipe.prior_pipe.maybe_free_model_hooks()
output = pipe(
width=1024,
height=1536,
decoder_guidance_scale=1.0,
prior_guidance_scale=7.0,
prior_num_inference_steps=30,
num_inference_steps=10,
output_type="pil",
prompt=prompt + " " + quality_prompt,
negative_prompt=negative_prompt,
prompt_embeds=prompt_embeds,
prompt_embeds_pooled=prompt_embeds_pooled,
negative_prompt_embeds=negative_prompt_embeds,
negative_prompt_embeds_pooled=negative_prompt_embeds_pooled,
num_images_per_prompt=num_images_per_prompt,
).images[0]
display(output)
Training:
GPU used: 7x Nvidia H100 80GB SXM5
Stage C
| parameter | value |
|---|---|
| amp | no |
| weights | fp32 |
| save weights | fp32 |
| resolution | 1024x1024 |
| effective batch size | 84 |
| unet learning rate | 2e-6 |
| te learning rate | 1e-7 |
| optimizer | AdamW 8bit |
| images | 6M * 2 captions per image |
| epochs | 1 |
GPU used: 1x Nvidia H100 80GB SXM5
Stage B
| parameter | value |
|---|---|
| amp | no |
| weights | fp32 |
| save weights | fp32 |
| resolution | 1024x1024 |
| effective batch size | 8 |
| unet learning rate | 8e-6 |
| te learning rate | none |
| optimizer | AdamW |
| images | 120K |
| epochs | 6 |
WD Tags:
Model is trained with this tag order:
aesthetic tags, quality tags, date tags, custom tags, rating tags, character, series, rest of the tags
Date:
| tag | date |
|---|---|
| newest | 2022 to 2024 |
| recent | 2019 to 2021 |
| mid | 2015 to 2018 |
| early | 2011 to 2014 |
| oldest | 2005 to 2010 |
Aesthetic Tags:
Model used: shadowlilac/aesthetic-shadow-v2
| score greater than | tag | count |
|---|---|---|
| 0.90 | extremely aesthetic | 125.451 |
| 0.80 | very aesthetic | 887.382 |
| 0.70 | aesthetic | 1.049.857 |
| 0.50 | slightly aesthetic | 1.643.091 |
| 0.40 | not displeasing | 569.543 |
| 0.30 | not aesthetic | 445.188 |
| 0.20 | slightly displeasing | 341.424 |
| 0.10 | displeasing | 237.660 |
| rest of them | very displeasing | 328.712 |
Quality Tags:
Model used: https://huggingface.co/hakurei/waifu-diffusion-v1-4/blob/main/models/aes-B32-v0.pth
| score greater than | tag | count |
|---|---|---|
| 0.980 | best quality | 1.270.447 |
| 0.900 | high quality | 498.244 |
| 0.750 | great quality | 351.006 |
| 0.500 | medium quality | 366.448 |
| 0.250 | normal quality | 368.380 |
| 0.125 | bad quality | 279.050 |
| 0.025 | low quality | 538.958 |
| rest of them | worst quality | 1.955.966 |
Rating Tags:
| tag | count |
|---|---|
| general | 1.416.451 |
| sensitive | 3.447.664 |
| nsfw | 427.459 |
| explicit nsfw | 336.925 |
Custom Tags:
| dataset name | custom tag |
|---|---|
| image boards | date, |
| text | The text says "text", |
| characters | character, series |
| pixiv | art by Display_Name, |
| visual novel cg | Full_VN_Name (short_3_letter_name), visual novel cg, |
| anime wallpaper | date, anime wallpaper, |
Limitations and Bias
Bias
- This model is intended for anime illustrations.
Realistic capabilites are not tested at all.
Limitations
- Can fall back to realistic.
Add "realistic" tag to the negatives when this happens. - Far shot eyes and hands can be bad.
- Still has a lot of room for more train.
- Diffusers outputs aren't as good as ComfyUI outputs.
License
SoteDiffusion models falls under Fair AI Public License 1.0-SD license, which is compatible with Stable Diffusion models’ license. Key points:
- Modification Sharing: If you modify SoteDiffusion models, you must share both your changes and the original license.
- Source Code Accessibility: If your modified version is network-accessible, provide a way (like a download link) for others to get the source code. This applies to derived models too.
- Distribution Terms: Any distribution must be under this license or another with similar rules.
- Compliance: Non-compliance must be fixed within 30 days to avoid license termination, emphasizing transparency and adherence to open-source values.
Notes: Anything not covered by Fair AI license is inherited from Stability AI Non-Commercial license which is named as LICENSE_INHERIT.