All upscale latent methods in one workflow
5.0
0 reviews
Description
What this workflow does
👉 [Please add here]
How to use this workflow
👉 [Please add here]
Tips about this workflow
👉 [Please add here]
🎥 Video demo link (optional)
👉 [Please add here]
Node Diagram
Discussion
Just tested and a few notes:
Your workflow is broken because all sampler names + schedulers are set to Null (need to update them all)
better to start your emtpy latent image to 1024x1024 to get best quality from SDXL models
all your upscales are blurry due to the Facerestore extra step.
To test a proper upscaling benchmark, it is better to set all seeds to fix with same seed numbers.
The problem with null names + schedulers is because of the openart upload ! i do upload the right file and becomes Null after upload ! weird !!
Node Details
Primitive Nodes (4)
Reroute (4)
Custom Nodes (37)
ComfyUI
- VAEDecode (6)
- CLIPTextEncode (2)
- KSampler (6)
- PreviewImage (5)
- LatentUpscale (5)
- EmptyLatentImage (1)
- CheckpointLoaderSimple (1)
- SaveImage (1)
- FaceRestoreCFWithModel (5)
- FaceRestoreModelLoader (5)
Model Details
Checkpoints (1)
juggernautXL_v7Rundiffusion.safetensors
LoRAs (0)