a1111 works , but comfy doesnt work at all

[Ulf3000]

it alwasy gives me an "undefined node" error.. the comfy nodes are installed , are my models named wrong ?

[FugueSegue]

I've been trying to get ComfyUI to work with this plugin for DAYS. I don't know what I'm doing wrong. I don't have an example of how it's supposed to work. The documentation is very sparse.

All of the native workflows are broken. I loaded them directly into ComfyUI and they have nodes that don't connect to anything.

The slightest clue as to which custom nodes are required would be helpful.

A demonstration of how to create and implement a new workflow is necessary. Converting the SDXL wokflow that generates the fennec fox is not enough. I've been trying to create an SDXL live painting workflow that works with this plugin and it's been 100% SNAFU. I would love to ask for help. But no one knows how to do this.

Please help us. We would gladly teach others.

[MrSorSor]

i have the plugin installed and it works the same way with comfyui as with automatic1111. what i mean is that you dont have to use comfyui workflows you can just hit the regular generation button in the plugin.

did you make sure that in the settings tab in the plugin you chose comfyui? did you make sure that you have the correct comfyui url (the same one that appears in the browser when comyui opens)?

you dont have to use a custom comfyui workflow but if you want to, you have two choices: 1- use the already available workflows in the plugins, for that you have to go to the following location: (this is only if you are getting missing nodes error, if you are not getting missed nodes error then you dont need to do this step)

{photoshop installation location}\Plug-ins\Auto.Photoshop.SD.plugin_v1.4.1 {or the plugin folder whatever you named it}\typescripts\comfyui\native_workflows

then drag a workflow to comfyui then it will tell you which node are missing, try to search online to find out which custom_node project has the missing nodes and install them to comfyui then after restarting comfyui you shouldnt have a problem.

please bear in mind that a lot of the custom_nodes project change the name of their nodes and so sometimes even if you install the correct custom_nodes project the workflows that are native to the plugin will not work. so it is better to try choice 2.

2- (choice 2) make your own workflow in comfyui (or find workflows for comfyui on civitai or reddit etc.) and save it as an api by clicking save (API Format) button in comfyui (red square in the picture) if you dont have that button then click on the little gear button (green square) then the settings panel will appear then make sure Enable Dev mode Options is checked (blue square).

then save the workflow in API format then you can load it in the plugin.

final note: of course make sure you have the Comfy-Photoshop-SD custom_node installed in comfyui to make sure the plugin is working correctly. and finally to help you install the missing nodes you should install comfyui manager in comfyui which makes installing custom_nodes very easy.

i hope i was able to help you and it solved your issues.

[Ulf3000]

ok knowing that i can just use the normal stuff is good , i thought its just the custom workflows ...

but i only get latent garbage back :

[MrSorSor]

i dont know why this is happening, maybe try different model? also try txt2img and see if the problem persists. i've noticed that currently some models dont work well with the plugin, especially inpainting models, as they basically the do nothing. if you go to the inpaint section of the plugin and you try using an inpaint model, at the bottom you have two choices: 1- original: just overstaurates the image even if denoise is at 1 2- latent nothing: which produce latent garbage even at denoise 1

also make sure you have selected the correct vae, in the settings tab of the plugin you can select a vae, however currently i think there is a bug in selecting a vae where even if you select any vae it uses the first one in the list. to get over that as a temp solution just rename the vae in the vae folder to something that makes is appear first like: from "sdxl_vae.safetensor" to "1-sdxl_vae.safetensor" without the quotes.

also make sure that the selection's resolution is within the model's range (between 512512 - 768768 for sd 1.5 and 1024*1024 for sdxl) it should work outside those resolutions but you wont get great results.

[madrooky]

ok knowing that i can just use the normal stuff is good , i thought its just the custom workflows ...

but i only get latent garbage back :

It kinda looks like your not generating that with the comfy module.

The description is really annoying for how easy it actually is to use. In comfyUI you DON'T need an extra node, and I don't even know how it works. I am doing mostly inpainting with a controlnet module, so I just have an IMG input node at the beginning of my workflow, not sure if it is even necessary. So just build your workflow as you would input the image manually and then save it as API workflow. Make sure you set your settings to a good default before saving because they are loaded with the workflow.

On your screenshot it looks like you are generating the image with the default SD module, the comfyUI stuff however is at the bottom. You need to load an API workflow and then you see the nodes and its settings there, you can ignore all the settings for the SD stuff from above.

@AbdullahAlfaraj, in case you read this, i am very pleased with your plugin. However, I hope there is some uncluttering of the plugin UI planned. As you see it can be super confusing and also annoying to scroll up and down. The description on the main page lacks a little Info how the plugin actually gets the image from PS, so people know what they need to have in their workflow. Thank you very much for this amazing stuff.