Sign In

Photoshop Power User- Nunchaku Kontext/Flux/SDXL/SD1.5

Download

1 variant available

Archive Other

28.29 KB

Verified:

Type

Workflows

Stats

187

Reviews

Published

Jul 23, 2025

Base Model

Flux.1 Kontext

Hash

AutoV2
B15E74FD71

The FLUX.1 [dev] Model is licensed by Black Forest Labs. Inc. under the FLUX.1 [dev] Non-Commercial License. Copyright Black Forest Labs. Inc.

IN NO EVENT SHALL BLACK FOREST LABS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH USE OF THIS MODEL.

Photoshop Power Users' Workflow

Flux / Kontext on Nunchaku, plus SDXL & SD1.5 for other stuff

Features:

  • Easily swap between model types from Photoshop (you still have to go to the comfyui window to change checkpoints and loras

  • Easily swap between usage modes

  • Works on 8GB VRAM (turn off hardware acceleration in Photoshop as a precaution)

  • Great for inpainting or turning your sketches into something, compositing stuff and forcing it together

Cons:

  • You gotta deal with Photoshop!

  • Not good for batching

  • MUST downgrade ComfyUI to 0.3.44 to work! (as of 7/23/25) and rgthree module needs to be 1.0.0 as well.

  • When you're inpainting, you'll end up doing a lot of mask-editing to get things looking just right. Having a Wacom Cintiq or similar device really helps.

INSTALLATION:

  1. Make sure your Nunchaku install works, doesn't matter how you do this. Make sure you're getting the right Kontext and Flux models to use with Nunchaku.

  2. Install Nima Nizrai's ComfyUI Photoshop and follow all the directions!

  3. Load this workflow and add any missing custom modules. When you add rgthree, make sure you downgrade or only install version 1.0.0! Later versions break the menu in the Photoshop plugin. (as of 7/23/25).

Again, if it runs, and you get to the last step and you get an error when it tries to send the image to Photoshop, you can try the code changes in that thread, or just downgrade your Comfy version to 0.3.44 where it will work. In 0.3.45 something changed with how it handles async nodes, which I don't understand so downgrading is my solution, for now. (as of 7/23/25)

Note: If you have a lot of VRAM, feel free to readapt this to a non-Nunchaku version, but I won't do it because I only have a 3070Ti so I couldn't test it myself.