ComfyGallery
A light-weight browser based gallery for ComfyUI
Features
Browse ComfyUI
inputandoutputdirectories.Add custom image directories to browse.
Delete images
Search images by filename or workflow data within PNG images.
Installation
Clone the project to a location of your choosing. You can move comfy_gallery.py to the root of your ComfyUI if you'd like (recommended) , or run it from where it's at.
Exmaple Launch Batch File (from ComfyUI Portable):
.\python_embeded\python.exe -s ComfyUI\comfy_gallery.py
pauseNotes
If you're running ComfyGallery from outside ComfyUI you'll need to provide the ComfyUI root directory to it with the
--comfyui-pathlaunch argument.Slow searching? The more images you have, the slower searching will be, as it's a sequential process.
Launch Arguments
--no-browser: Do not launch system browser when the server launches.--purge-cache: Delete the image gallery cache on startup.--image-paths: A list of additional image paths for ComfyGallery to manage.--comfyui-path: Path to ComfyUI root directory.--comfyui-color-palette: Path to a ComfyUI colorPalette.js file
Description
See github for more information and up-to-date code base.
FAQ
Comments (14)
Waiting for reviews on this, but on paper this is a really cool idea.
It's very bare bones right now. Basic features, but also want to gauge interest in general before going all out. I mean, we all have explorer on Windows, etc.
@WAS You are a legend, been using your nodes for a while now. I was gonna ask questions, but I'll just try it out. A feature that I'm looking for is something that can quickly show Comfy workflow info that I can cut and paste into my setup. I don't always want to load the full workflow just to see what I did before. Cheers!
@MysteryWrecked I was trying for this with the overlay when you via a search result, it'll display the marched content in the workflow via its parent. But tricky mapping it right. Can copy whole workflow though for PNG files that have one.
No github for easy updates?
Click the huge "ComfyGallery" link at the top of the description.
nice, does it show the prompts?
Prompts aren't easily identifiable in ComfyUI. They could be text nodes, primitive nodes, or CLIPTextEncode, all broken up. But you can search by what's in prompts, and it'll show that part of the work flow to copy said prompt. You can also copy the workflow in entirety to paste into ComfyUI to load it.
@WAS true, just I so often want to go through a large number of pics and quickly see the prompt info without opening the entire workflow. I wonder if it's possible to show "clip texts", with worst-case being it might mix up some positive/negatives etc...
@Janet it's possible, but like I mentioned not everyone even uses CLIPTextEncode. A lot of people manipulate text nodes and then just feed the result into a CLIPTextEncode. Or on WAS-NS you can just use the Text to Conditioning node.
@WAS I don't know if it's possible, but would it simply consider any text entering the sampler's conditioning input as the prompt? Maybe it'd be messy with errors, but better than nothing.
@Janet I think the condition is already encoded vectors. But it is a dict and could send prompt with it, but then every node doing so would have to include the prompt.
This is awesome, and pretty much just what I was looking for.. but could you please make the UI viewable over local network also? I run ComfyUI most often through a remote PC on my LAN
You should be able to connect to the port, 8189, or use ngrok ngrok http 8189
It uses 127.0.0.1 as address and port 8189, which you can change at top of py