Suggestions

[SilverAndJade]

trafficstars

I haven't checked the newer version but after using this interface for a while, I've found the following possible upgrades;

• Ability to pass the seed value from one sampler to another
• In Interface file browser/image download for remote use (Nothing complicated, just a viewer like the Save_File processor that just gets an update call after save file where I can select and download the previously generated images)
• A Note Item (I'd love to be able to add notes to the interface)

Not required but possible;

• Multi Lora Loader (Being able to add multiple Lora models and switch between them quickly when necessary)
• More detailed queue view (When I've queued multiple items, I'm not 100% sure what the details of the currently processing queue is, just the prompt details on hover would be useful)

[GrimblyGorn]

I had thought some of these might have been handled already by various mods I've happened across and incorporated but maybe not. Many of these would be very useful to have though, so I'm agreeing with this.

Your 1st addition may already be an option as I explained here earlier. It'll explain how to add the custom nodes for samplers with seeds and links to them, etc.

Possibly that's helpful :)

[SilverAndJade]

Thanx, honestly none of the rest are super complicated. I should probably just code them myself if I waste time in the evening generating images for fun. :)

[GrimblyGorn]

Looks like you went and coded some things now too :)

I'm hoping to check them out later possibly. I see you made the note node too, that's handy on its own.

[SilverAndJade]

There's a preview image and a json save to use the nodes in the pull request, I think one last thing I want to do is add a button on the main floating menu to delete all generated images. The seed transfer between samplers seems to work, but I can't be 100% sure if I've done it correctly. Either way, the outputs I get seem to be more along the lines of what I was thinking for an Upscale Fix workflow.

[GrimblyGorn]

Ya one thing that I've found difficult in using this is there seems to be no way of telling what values i'm passing around. In many cases it's clear from having entered the value. But there's a lot of instance where it's not as clear and there's no way to display the value. Something that allowed a universal type input and then displayed its value before possibly passing it to an output would be very handy in those scenarios. A passthrough display node for value checking/watching.

In Grasshopper you can simply hover over the output connector and it'll pop open a little display with the values that connector is receiving or passing. That could be very handy here too. This somewhat fails when there's too much for the little popup window to display. In those cases, a node to display the data would be helpful too.

This could help clarify your seed issue better probably, or any others where a value is being questioned. Might be worth looking into if you cared to or had the time. I might later too if I find some time.

[SilverAndJade]

Well, as I understood Automatic1111's Upscale Fix, it kept the same seed value to reduce the randomization somewhat. The issue I was having was that each sampler gets a random value on "generate start", or a hardcoded value. So I was trying to replicate the Upscale Fix by just having the same seed value on both samplers when passing through an upscale, which is essentially Img2Img, with seed value + denoise to reduce the artifacts in the image. My hopes was to fix the eyes/face + reduce the chance of strange hands or body parts.

I'm honestly not sure if I've done it correctly at all, but shrug it was worth a shot and the pull request at least has some things in that needed fixing, such as the fact that Optional Input values weren't yet added to the UI

[GrimblyGorn]

Oooh you fixed the Optional inputs? That was really bugging me.

I may have to go merge yours to mine now :)

[SilverAndJade]

@comfyanonymous I've finally added my output directory solution and removed the SaveImageList node from my fork. Please have a look and see what you think. https://github.com/Silversith/ComfyUI/blob/master/web/extensions/core/imageFeed.js It's essentially an image drawer that will load all the files in the output dir on browser refresh, and on Image Save trigger, it will add the image to the drawer. The code was partially copied from https://github.com/pythongosssss/ComfyUI-Custom-Scripts and then modified.

[comfyanonymous]

Make a pull request if you want me to properly look at it. It makes it easier for me to keep track of things.

[jahu00]

Ability to pass seeds would be great.

[SilverAndJade]

Ability to pass seeds would be great.

You can already use a primitive and convert the seeds to inputs and pass in the primitive