ToonCrafter AI去噪动画ComfyUI 工作流节点

简介

本工作流仅需要一张初始图片，使用DifferentialDiffusion及KSampler Gradually Adding More Denoise，产生keyframe，透过Denoise的强度变化，Mask范围会逐渐被侵蚀，最后转换为目标图片。工作流可产生八张图片，将Denoise的画面分成前中后三个区段，每个区段为三格，使用Switch选择需要的部分输入Toon。

工作流节点截图

附件文件一览

模型
tooncrafter_model.ckpt👉前往下载

内置节点
Anything Everywhere3
Fast Groups Bypasser (rgthree)
Note
ToonCrafterInterpolation

自定义节点
ComfyUI
– VAEDecode
– VAELoader
– CLIPTextEncode
– UpscaleModelLoader
– ImageUpscaleWithModel
– LoadImage
– FreeU_V2
– DifferentialDiffusion
– InpaintModelConditioning
– ControlNetLoader
– LatentFromBatch
– PreviewImage
ComfyUI Essentials
– ImageCASharpening+
ComfyUI Frame Interpolation
– RIFE VFI
– KSampler Gradually Adding More Denoise (efficient)
ComfyUI Impact Pack
– ImpactMakeImageBatch
– ImpactSwitch
ComfyUI Inspire Pack
– ImageBatchSplitter //Inspire
ComfyUI Layer Style
– LayerMask: SegmentAnythingUltra
ComfyUI Nodes for Inference.Core
– MiDaS-DepthMapPreprocessor
ComfyUI-Advanced-ControlNet
– ACN_AdvancedControlNetApply
– ScaledSoftControlNetWeights
ComfyUI-VideoHelperSuite
– VHS_VideoCombine
Efficiency Nodes for ComfyUI Version 2.0+
– Efficient Loader
Save Image with Generation Metadata
– String Literal
Various ComfyUI Nodes by Type
– JWImageResizeByLongerSide