abdullahalfaraj / comfy-photoshop-sd

Download this extension via the ComfyUI manager to establish a connection between ComfyUI and the Auto-Photoshop-SD plugin in Photoshop. https://github.com/AbdullahAlfaraj/Auto-Photoshop-StableDiffusion-Plugin

Comfy-Photoshop-SD

Download this extension via the ComfyUI Manager to establish a connection between ComfyUI and the Auto-Photoshop-SD plugin in Photoshop.

How to Install:

1. Install Auto-Photoshop-SD v1.4.0 or a later version -> using either installation Method 1 (.ccx) or Method 2 (.zip) file. Important: skip any instruction related to Automatic1111

2. Install ComfyUI-Manager inside of ComfyUI

   • a) Navigate to ComfyUI Folder
   • b) Copy the following command:

   git clone https://github.com/ltdrdata/ComfyUI-Manager.git
   

   Follow the steps here:

   install.comfyui-manager.mp4

4. Install this project (Comfy-Photoshop-SD) from ComfUI-Manager

how.to.install.ComfyUI-Manager.mp4

5. Similarly install (comfyui_controlnet_aux)[https://github.com/Fannovel16/comfyui_controlnet_aux] from ComfyUI-Manager to use controlnet

Load Custom ComfyUI Workflows in Photoshop

1. Load Workflow: Open your workflow in ComfyUI. Make sure it's error-free.

2. Convert to API: Save the workflow as an API in JSON format. Use 'Save (API Format)', not 'Save'. See an example.

export.api.from.workflow.mp4

3. Move API File: Transfer the API JSON file to a folder of your choice, only include json files that are made using this guide. Here's an image for reference.

4. Load Workflow in Photoshop: Load your custom workflow within Photoshop. Here's an example.

load.custom.workflow.inside.photoshop.mp4

How to Use:

Txt2Img with Hires fix:

txt2img_hires_fix.mp4

Img2Img with Hires fix:

img2img_hires_fix.mp4

Pure Img2Img :

img2img_only.mp4

Pure Inpainting:

pure_inpainting.mp4

Outpainting with Controlnet:

outpaint_with_ctrlnet.mp4

comfy-photoshop-sd's Issues

The workflow can run in the browser, but when clicking Generate in the Custom Comfyui Workflow section of PS, an error is reported

got prompt
Prompt executed in 0.04 seconds
ERROR:aiohttp.server:Error handling request
Traceback (most recent call last):
File "E:\SD\ComfyUI_windows_portable\ComfyUI\venv\lib\site-packages\aiohttp\web_protocol.py", line 332, in data_received
messages, upgraded, tail = self._request_parser.feed_data(data)
File "aiohttp_http_parser.pyx", line 557, in aiohttp._http_parser.HttpParser.feed_data
aiohttp.http_exceptions.InvalidURLError: 400, message:
Invalid char in url query:

bytearray(b'GET /view?subfolder=&type=output&filename=\xe7\xac\xac\xe4\xb8\x80\xe6\xac\xa1KSampler_00046_.png HTTP/1.1')
                                                       ^

ERROR:aiohttp.server:Error handling request
Traceback (most recent call last):
File "E:\SD\ComfyUI_windows_portable\ComfyUI\venv\lib\site-packages\aiohttp\web_protocol.py", line 332, in data_received
messages, upgraded, tail = self._request_parser.feed_data(data)
File "aiohttp_http_parser.pyx", line 557, in aiohttp._http_parser.HttpParser.feed_data
aiohttp.http_exceptions.InvalidURLError: 400, message:
Invalid char in url query:

bytearray(b'GET /view?subfolder=&type=output&filename=\xe6\x89\xa9\xe5\x9b\xbe\xe5\x90\x8eXLUpscale_00021_.png HTTP/1.1')

I think it's too complicated to analyze the workflow of comfyui

Send to Auto1111 webui| alternative

Hey I am not finding any extensions in webui to do the same this extension is doing for Comfyui. Is there one out there? I was hoping this was the case.

All i can find are extensions to use SD in Photoshop, but I want to save my image in Photoshop and send automatically to webui.

Anychance this can be applied the same way as for ComfyUi?

Windows Only?

It seems that this only runs on Windows (getting ModuleNotFoundError for 'winreg' and a FileNotFoundError for 'python.exe'). If this is true, consider making the requirement clear on your YouTube and GitHub pages to save people from the hassle of trying to install it.

Can anyone help me with this problem?

Can anyone help me with this problem? I do everything in the description, but I can’t solve this problem,
None of these workflows work except the one in the example, why doesn’t it find it? Has anyone encountered this problem?

I didnt find the "save(api format)" button

ive installed pulgin by comfyui manager,but i didnt find the "save(api format)" button

btw, i installed pyssss's custom-sripts,and tried to disable it ,still didnt see the button

Layer Sync in Custom Workflow not working

Thanks for all the work you've put into this!

To start: I'm on windows, talking to the API via port 8188. I'm using a remote computer with an A100 GPU and running comfyui there. That comfyui is forwarded to my local machine via VS Code. I installed using the .ccx file on January 7 2024. I have used A1111 and comfyui a decent amount, and have also used a Krita extension with comfyui.

I'm trying to run a basic img2img workflow test_miku_img2img_api.json in the custom workflows tab and I have a node that looks like this:

I can upload from the canvas manually to get something into that node, but it is the only way for me to input into that node. When I click the "Sync Current Layer" button it marks the current layer as red:

I can unmark the layer with right click, then I can click the sync current layer button again to mark the layer as red. Point being that the button press is definitely doing something. The Sync from Canvas button does not work. This is true if I have something selected via the Marquee tool and also if I have nothing selected. Either way the input image that gets passed in appears to just be a white square:

Note that this is still true even if I don't have a white background (note the background layer is not visible and is also not marked red)

I can confirm that the workflow works because if I turn the denoising strength up to 1 then I can get an image:

The "Load Image from Canvas" button will work if I have a selection:

And I can use that as the input for my workflow. That being said, that button doesn't help if I want to do live img2img for instance.

I tried poking around in the logs (mostly because of other crashes and timeouts related to both regular and custom workflow bugs), but there are a lot of messages spread between multiple log files, so I'm not sure what would be helpful for me to attach for you.

Here's a screencap of my settings page:

Last note is that the regular txt2img setting works, although sometimes it does not actually send anything to comfyui (confirmed via the comfyui.log file) and hangs while the loading bar goes upwards of 200%. So I do know the plugin is able to talk to my comfyui server at least sometimes. I haven't had a hang/infinite progress bar issue yet with a custom workflow.

---

Let me know if there's any testing you would like me to do or any logs files that might help.

Reading images from the Photoshop selection box is very slow, but importing directly from the folder is fast

Reading images from the Photoshop selection box is very slow, but importing directly from the folder is fast

Image sampler issue

May I ask if there is a scheduler selection in the plugin? When using "dpmpp3m_sde_gpu", the image may have strange problems. I have tried various cfg scale values but it is not useful.

Using euler_ancestral images without collapse, I am not sure what the reason is

please add real-time mode

like krita ai plugin that

[Feature Request]color mark

In comfyui, necessary nodes can be marked with colors, and in ps, it can be clearly determined which node needs to modify parameters

Add A List of Nodes to Install

Please add a list of dependency nodes to install.

Just awesome, great work :)

Just awesome, great work :)

Is there a way to generate images in the comfyui browser that can be seen directly in Photoshop

as title

Add "Enable Developer Mode" to instructions

It seems to be missing. That needs to be checked under settings to reveal the Save API button.

In comfyui, my node color is brown. Why is the green border displayed in Photoshop? This should be a bug