CatVTON在衣服类型不匹配时换出来的结果可能不够理想,但Flux对于提示词的理解非常准确。通过使用Ollama对图片进行描述,再利用Flux还原Ollama的描述,可以在一定程度上改善这一问题。最后,通过CatVTON在两个相近衣服之间进行换装,这个comfyui工作流虽然简单,但得到的AI换装效果却非常令人满意。
工作流界面图
作者:dogami6666
模型详情
节点详情
内置节点
CatVTONWrapper
Fast Groups Muter (rgthree)
FluxGuidance
Image Comparer (rgthree)
ModelSamplingFlux
PortraitMasterBaseCharacter
PortraitMasterMakeup
PortraitMasterSkinDetails
PortraitMasterStylePose
Simple String
easy getNode
easy setNode
自定义节点
ComfyUI
– VAELoader
– KSamplerSelect
– RandomNoise
– DualCLIPLoader
– BasicGuider
– VAEDecode
– CLIPTextEncode
– MaskToImage
– SaveImage
– PreviewImage
– UNETLoader
– LoadImage
– SamplerCustomAdvanced
– BasicScheduler
ComfyUI Impact Pack
– SAMLoader
– ImpactSwitch
ComfyUI Layer Style
– LayerUtility: TextJoin
ComfyUI Ollama
– OllamaVision
comfyui-mixlab-nodes
– TextInput_
Mikey Nodes
– Empty Latent Ratio Select SDXL
pythongosssss/ComfyUI-Custom-Scripts
– ShowText|pysssss
segment anything
– GroundingDinoModelLoader (segment anything)
– GroundingDinoSAMSegment (segment anything)
评论0