Flux Controlnet to Manga Style
5.0
0 reviewsDescription
资源合集:闲话AI官网 talk2ai.cn
本工作流旨在Flux模型及LoRA下将真人照片转换为独特的漫画风格。这是一个实验性工作流,因为Flux模型下的Controlnet效果目前尚不完全稳定。
注意事项:
• Flux模型下的Controlnet可能不稳定,即使同样的设置,在不同照片下呈现的结果可能会有较大的差异
• Prompt对结果影响非常大,所以使用了条件合并节点 - Conditioning(Combine),你可以根据需要手动调节Prompt
• 为了稳定黑白漫画的效果,使用了颜色校正节点 - Color Correct,表现力不佳的时候也可以尝试提高一些明亮度 - brightness
• LoRA模型可在Civitai下载:https://civitai.com/models/801701/zanshou-kin-flux-ueno-manga-style
欢迎尝试并分享您的结果!
This workflow aims to transform real photos into a unique manga style using the Flux model and LoRA. It is an experimental workflow, as the Controlnet effect under the Flux model is not yet fully stable.
Notes:
• Controlnet under the Flux model may be unstable, and results may vary significantly with different photos even with the same settings
• Prompts have a significant impact to the results, so a Conditioning (Combine) node is used - you can manually adjust the prompt as needed
• To stabilize the black and white manga effect, a Color Correct node is used - If less expressive, you can also try increasing the brightness
• The LoRA model can be downloaded from Civitai: https://civitai.com/models/801701/zanshou-kin-flux-ueno-manga-style
Feel free to try and share your results!
Discussion
(No comments yet)
Loading...
Reviews
No reviews yet
Versions (1)
- latest (a year ago)
Node Details
Primitive Nodes (9)
ControlNetApplySD3 (1)
EmptySD3LatentImage (1)
FluxGuidance (1)
Image Comparer (rgthree) (1)
ModelSamplingFlux (1)
PrimitiveNode (2)
SetUnionControlNetType (1)
UnetLoaderGGUF (1)
Custom Nodes (22)
ComfyUI
- CLIPTextEncode (3)
- DualCLIPLoader (1)
- ControlNetLoader (1)
- ConditioningCombine (1)
- KSampler (2)
- VAELoader (1)
- LoraLoaderModelOnly (1)
- VAEDecode (2)
- PreviewImage (1)
- LoadImage (1)
- SaveImage (1)
- TilePreprocessor (1)
- ColorCorrect (1)
- LLava Loader Simple (1)
- LLavaSamplerSimple (1)
- ViewText (1)
- SimpleText (1)
- LlavaClipLoader (1)
Model Details
Checkpoints (0)
LoRAs (1)
ueno.safetensors