Anima Edit (Apply Cosmos Reference Latent) conflicts with Apply Anima ControlNet-LLLite. Anima Edit use two frames (T=2) while LLLite only supports one frame (T=1). It seems that before 0.29.0 it did not throw error...
Profil d'exécution
Description de la source
!Known Custom Nodes Conflicts!
Anima Edit (Apply Cosmos Reference Latent) conflicts with Apply Anima ControlNet-LLLite. Anima Edit use two frames (T=2) while LLLite only supports one frame (T=1). It seems that before 0.29.0 it did not throw error but only Anima Edit had effect. Don't use them together.
When Anima Edit Lora is enabled, the Apply Cosmos Reference Latent node will cause error on Tiled Diffusion. Use them seperately in two runs. (Edit, then upscale)
Artist Mix does not work well with certain Lora. Bypass it when needed. (Require loading another model and back to the one you were using.)
================================
v1.2
Commentaire généré par l’IA
Explication générée par l’IA à partir des détails de la source et de la configuration. Les suggestions sont clairement signalées.
Use this Anima image-to-image workflow with an input image to edit the whole image, inpaint masked areas, extend an image outward, or upscale it, producing edited, inpainted, outpainted, or upscaled images.
For whole-image editing, use Image Edit, optional Image Resize, Full Image Latent, Anima Edit Lora, and optional ControlNet.
For masked inpainting or editing, use optional Image Resize and Inpaint Crop and Stitch, followed by Inpaint Latent, Inpaint Lora, optional ControlNet, and Anima Edit Lora.
For outpainting, use Padding for Outpaint, Image Resize, Full Image Latent, and Anima Edit Lora; for upscaling, use a second-stage Full Image Latent with latent, model, or RTX upscaling, plus optional ControlNet, Tiled Diffusion, and Color Fix.
The main input is an image, with optional Image Resize, Inpaint Crop and Stitch, ControlNet, or Padding for Outpaint; the described results are edited, inpainted, outpainted, or upscaled images.
Download version v1.1 from Civitai at https://civitai.com/models/2742680?modelVersionId=3099516.
Some named custom nodes may need manual placement in the custom-nodes folder if Node Manager does not install them.
Suggestion · non vérifié
Listed model files include 4x-AnimeSharp.pth, anima-lllite-any-test-like-v2.safetensors, anima-lllite-inpainting-v2.safetensors, AnimeEditV2.safetensors, and Cosmos-Predict2.5-2B-base-distilled-LoRA.safetensors.
Additional listed model files are Extend Image (Anima Edit) v1.safetensors, JANIMA_v10.safetensors, qwen_3_06b_base.safetensors, and qwen_image_vae.safetensors.
The workflow lists comfyui-image-saver 1.22.0 and comfyui-mxtoolkit 0.9.92; named custom nodes include Anima Sampler, Anima-Artist-Mixer, ComfyUI-Anima-NAG, ComfyUI-Anima-LLLite, ComfyUI-MooshieTiledDiffusion, ComfyUI-KJNodes, and comfyui-upscale-by-model.
Optional controls include Image Resize, Inpaint Crop and Stitch, ControlNet, Tiled Diffusion, and Color Fix; Seed Variance Enhancer and quick bypass access are also included, and Artist Mix can be switched on or bypassed.
Anima Sampler is an alternative to KSampler, and the source says it may or may not produce better results.
Do not combine Anima Edit (Apply Cosmos Reference Latent) with Apply Anima ControlNet-LLLite because their documented frame counts differ: T=2 versus T=1. When Anima Edit Lora is enabled, use separate edit and Tiled Diffusion upscale runs.
Suggestion · non vérifié
Artist Mix may work poorly with some LoRAs; if you bypass it, reload another model and then reload the desired model.
Suggestion · non vérifié
Faut-il modifier cela ?
Connectez-vous pour envoyer une demande de modification.
Sources
1 sourceExtraits de sources
3 extraitsSource context: 269 downloads · Type Workflows · Base model Anima
Fixed Anima ControlNet-LLLite node for ComfyUI v0.29.0
Added NegPip
Changes to layouts
v1.1
Added Outpaint feature
Changed Normalized Attention Guidance to Anima Normalized Attention Guidance. Separated it from subgraph. Now you can bypass it independently. It could be useful even when not using Turbo lora
v1.0
Fixed missing connections
================================
An Anima i2i Workflow to Cover Your Basic Needs
Features:
One workflow to do inpaint and edit lora.
Outpaint Anima edit lora
Quick bypass access
No messy connections. Clean and tidy. Easy to understand and customize.
Anima Sampler, an alternative to KSampler. May or may not produce better results.
ControlNet
Seed Variance Enhancer, a second dice to roll for desired outcome.
Artist Mix: Hot switch to enable/disable. Bypass the node to avoid compatibility issues. (Bypassing/Unbypassing the node requires loading another model then loading your desired model again to clean cache. I hope this will be fixed in the future)
Tiled Diffusion for high upscale factors.
Use cases: (Options to enable)
Whole image editing using Anima Edit Lora
1st - Image Edit
Image Resize (optional)
Full Image Latent
Anima Edit Lora
ControlNet (optional)
Masked inpaint:
1st - Image Edit
Image Resize (optional)
Inpaint Crop and Stitch (optional)
Inpaint Latent
Inpaint Lora
ControlNet (optional)
Upscaling:
2nd - Upscale
Full Image Latent
Upscaled by Latent/Model/RTX
ControlNet (optional)
Tiled Diffusion (optional, recommended for whenever the output is having artifacts)
Color Fix (optional)
Masked image editing using Anima Edit Lora:
1st - Image Edit
Image Resize (optional)
Inpaint Crop and Stitch (optional)
Inpaint Latent
Anima Edit Lora
Outpaint with Lora:
1st - Image Edit
Padding for Outpaint
Image Resize
Full Image Latent
Anima Edit Lora (Enable Use Outpaint Lora in the subgraph node)
Custom nodes used (not shown in "In Workflow" inside the Node Manager):
Anima Sampler: https://github.com/KeithZ117/Comfyui-anima-sampler
Anima-Artist-Mixer: https://github.com/An1X3R/Anima-Artist-Mixer
ComfyUI-Anima-NAG: https://github.com/hybskgks28275/ComfyUI-Anima-NAG
ComfyUI-Anima-LLLite: https://github.com/kohya-ss/ComfyUI-Anima-LLLite
ComfyUI-MooshieTiledDiffusion: https://github.com/Mooshieblob1/ComfyUI-MooshieTiledDiffusion
ComfyUI-KJNodes : https://github.com/kijai/ComfyUI-KJNodes
comfyui-upscale-by-model: https://github.com/cdzcj/comfyui-upscale-by-model/ (Node Manager links to wrong repository.)
Some of these may not install properly through the Node Manager, so you might need to download and place them manually into the custom nodes folder.
Estimation des besoins VRAM
Estimation indisponible
4,19 GB sur 3 de 9 fichiers de modèle. Total des fichiers du modèle + 25 % de surcharge de chargement + 2 Go de tampon d'exécution, arrondi à l'unité supérieure.
Exigences
Exigences 11Upscaler · 63.9 MB · Hugging Face · Kim2091/AnimeSharp
anima-lllite-any-test-like-v2.safetensors
Non résoluControlNet · Unknown
anima-lllite-inpainting-v2.safetensors
Fichier non vérifiéControlNet · SAFETENSORS · Unknown
AnimeEditV2.safetensors
Non résoluLoRA · Unknown
Cosmos-Predict2.5-2B-base-distilled-LoRA.safetensors
Non résoluLoRA · Unknown
Extend Image (Anima Edit) v1.safetensors
Non résoluLoRA · Unknown
Checkpoint · 3.90 GB · SAFETENSORS · bf16 · pruned · Civitai · 2642932
Text encoder · Civitai · 2695493
Sources possibles
VAE · 242 MB · Civitai · 1842123
Sources possibles
Pack de nœud · Registry
Pack de nœud · Registry
Source context: 127 downloads · Type Workflows · Base model Anima
Source context: 105 downloads · Type Workflows · Base model Anima