V3.2:
Two similar workflows featuring different approaches for the second-pass upscale.
Anima3_T2I_V3.2_SDUpscale: Upscaled via Ultimate SD Upscale. Minor optimizations to nodes and parameters of V3.1.
✔Stable at low denoise values (<0.4), supports upscaling beyond 2x (4MP).
❌the image becomes unstable at high denoise values.
Anima3_T2I_V3.2_HiresUpscale: High-Res Upscale (via second-pass resampling).
✔High-quality output at higher denoise (>0.4). Personally recommended: 1.5x upscale (2.25MP) with 0.65 denoise.
❌Upscaling beyond 2x (>4MP) is very unstable, and low denoise settings may occasionally cause a grainy texture.
V3.1:
This version ensures that Civitai can properly parse the required metadata in saved images, which includes all model info.
V3.0:
Add a bypassable ControlNet (Anima-LLLite). Big thanks to Kohya for creating this ControlNet.
Introduction Page: https://huggingface.co/kohya-ss/Anima-LLLite
recommended CN file: anima-lllite-any-test-like-v2.safetensors https://huggingface.co/kohya-ss/Anima-LLLite/tree/main
custom node: https://github.com/kohya-ss/ComfyUI-Anima-LLLite
V2.0:
Add face detailer and hand detailer.
V1.0:
A simple text-to-image workflow for Anima with Ultimate SD Upscale. (If you notice unnatural transitions when increasing the denoise value in the Upscale node, try increasing the tile_padding, such as 256.)
Thanks to @Boerboer, @sam_tofu and @RavirKun for the suggestions.
Thanks to @UmeAiRT and @CptCharname! I’ve learned so much from studying their workflows.
Resources:
Anima (Base model): https://civarchive.com/models/2458426/anima?modelVersionId=2945208
CLIP: https://huggingface.co/circlestone-labs/Anima/tree/main/split_files/text_encoders
VAE: https://huggingface.co/circlestone-labs/Anima/tree/main/split_files/vae
Anima Style Gallery: https://anima.mooshieblob.com/
Anima Artists and Characters: https://animadex.net/?mode=characters
The following are just my personal preferences. Feel free to swap them out for files that suit you better.
Upscale Model: https://huggingface.co/FacehugmanIII/4x_foolhardy_Remacri/tree/main
face_yolov8n: https://huggingface.co/Bingsu/adetailer/blob/main/face_yolov8n.pt
hand_yolov8s: https://huggingface.co/Bingsu/adetailer/blob/main/hand_yolov8s.pt
PitHandDetailer-v2-Test-v9c: https://huggingface.co/iahhnim/adetailer_collection/blob/main/PitHandDetailer-v2-Test-v9c.pt
Description
Add and modify nodes so that the metadata about models and LoRAs embedded in saved images can be parsed by Civitai.
FAQ
Comments (7)
Very good workflow. After few tweaks I made to feel more comfy with it I really like it
Can you please update this WF and add a Shift setting. I kept getting weird fingers, and then realized there was no Shift at all. For this model, especially when mixing artists, a Shift value of 10 works better. Leaving it at 0 or not using Shift seems to make the result worse
Thanks for the suggestions! I haven't seen anyone using Shift with Anima yet. Which ModelSampling node did you use? I'll run some tests when I'm free.
By the way, does raising the Shift value affect the first-pass generations in any major way?
@SAN0 ModelSamplingAuraFlow.
yes, significant.. if you're using a lot of artist mix, shift 10 is better.. but for general use 4-6 is enough. https://www.reddit.com/r/StableDiffusion/comments/1tj2rcl/stabilizing_mix_of_artist_tags_in_anima/
@RavirKun I did some testing on my end, and so far, increasing the Shift value doesn't seem to make a noticeable difference for hand detailing specifically. Tweaking the denoise and rerunning with different seeds is still the most reliable way to get clean hands.
If you don't mind, could you share the specific prompt where you ran into the issues? I'd love to run some tests with it. Thanks a bunch! ~
@SAN0 it's okay.. it's likely from artist tag that makes the fingers look weird. I already ask chatgpt how to add shift to this workflow, so it's all good now
omg thankyou so much for your work!