Dr.Lt.Data c5034e737d fix: resolution issue
- ensure (min length > 128)
- pad for multiply of 64 and remove pad after upscale
improve: set default model as 'last.ckpt' in model combo

new feature: add nodes `LDSRModelLoader`, `LDSRUpscale`
- separate out model loading part (prevent loading model everytime)
- refactor LDSR class for this
2024-01-06 00:26:52 +09:00
2024-01-06 00:26:52 +09:00
2024-01-06 00:26:52 +09:00
2024-01-03 13:10:48 +02:00
2024-01-03 13:10:48 +02:00
2024-01-03 13:10:48 +02:00
2024-01-03 13:10:48 +02:00
2024-01-03 13:10:48 +02:00
2024-01-03 13:10:48 +02:00
2024-01-03 13:18:30 +02:00
2024-01-03 13:10:48 +02:00
2024-01-03 13:18:30 +02:00

ComfyUI-Flowty-LDSR

This is a custom node that lets you take advantage of Latent Diffusion Super Resolution (LDSR) models inside ComfyUI.

LDSR models have been known to produce significantly better results then other upscalers, but they tend to be much slower and require more sampling steps. Results may also vary based on the input image.

example example

I've created this node for experimentation, feel free to submit PRs for performance improvements etc.

Installation:

  • Install ComfyUI
  • Clone this repo into custom_nodes:
    $ cd ComfyUI/custom_nodes
    $ git clone https://github.com/flowtyone/ComfyUI-Flowty-LDSR.git
    
  • Install dependencies:
    $ cd ComfyUI-Flowty-LDSR
    $ pip install -r requirements.txt
    
  • Download LDSR and place it in ComfyUI/models/upscale_models
  • Start ComfyUI (or restart)

Workflow:

example

This is a community project from flowt.ai. If you like it, check us out!

flowt.ai logo
S
Description
No description provided
Readme GPL-3.0
17 MiB
Languages
Python 100%