ComfyUI - IMG2IMG and SD Prompt Reader

[@RuinDweller]

Thank you for the video!! Can you please tell me how you accomplished the meters on your manager? In the meantime I may be able to help you in return. 1) When you click "save" in your workflow menu, if you just click "ok" when given the chance rather than entering a filename, you will be given a chance to save it wherever you'd like, as whatever name you'd prefer. 2) When you load a new workflow you can just hit "install missing custom nodes", and ComfyUI will automatically show only those nodes featured in your workflow which you have not yet loaded. Cheers!

[@joeconway85]

Hi, thanks for your comment. To set your connectors to be straight, select the cogwheel in the ComfyUI menu (not manager), then change the Link Render Mode to Straight. This is a global setting so it will affect all your Workflows, not just the current one. The resource meters are part of the Crystools Nodes suite. Have a look at my ComfyUI - Crystools Resource Monitor & Nodes video at: kzbin.info/www/bejne/j32pkIVuhLOke9E Hope this helps.

[@grahamastor4194]

Unfortunately when you re-activated the Img2Img group you also activated the loras; did they have any effect on the output?

[@joeconway85]

Good spot - my bad! On this occasion it had no impact because although I had (mistakenly) brought the Load Lora Nodes back online, there were no trigger words for the Loras in the positive comment box so they were not activated. My next video will also be on Loras and I will try to get it right this time! :-)

[@dronearon3085]

where to download the workflow

[@joeconway85]

Hi, my Workflow for this video is available from my DropBox account, just follow the link: www.dropbox.com/scl/fi/zxrhxy1ihyxkteogknpkt/IMG2IMG-SD-Prompt-Reader.json?rlkey=818l5ra8jgbn5kicbo8nzrel6&dl=0

[@dronearon3085]

@@joeconway85ok tq will try it

[@RuinDweller]

This worked the first time I used it, but now the prompt reader will not work. :( I get the error: "Error occurred when executing SDPromptReader: invalid literal for int() with base 10: 'None'" Okay, it still works, but ONLY for images generated with that workflow, why is that? I thought this was supposed to be able to detect prompts for images even if they were generated by other workflows...

[@joeconway85]

Hi, I have used the SD Prompt Reader many times on different Workflows and calling images generated on different Workflows without any problems thus far, so I don't know what this might be. I have had a couple of images (out of hundreds) I generated that did not work in the Prompt Reader, but only a couple and I don't know why. I usually deactivate all other Nodes when I use it (to save undue processing), which is why I put my Nodes into groups to make that easier to do. If you haven't already tried it, try deactivating the rest of your Nodes and see if that helps. Might be also worth checking images from other Models to see if maybe the Prompt Reader doesn't like a particular Model you are using. The error message is code speak so over my head I'm afraid - take a look at this if you understand code: stackoverflow.com/questions/1841565/valueerror-invalid-literal-for-int-with-base-10 I don't do online forums, I find them too abrasive, but there may be a ComfyUI forum out there that can help if the issue persists. Sorry I couldn't be of more help.

[@RuinDweller]

@@joeconway85Please don't worry about it, I'm only glad for your help. I've been doing some looking, and it appears that there is an issue when using certain file formats. Images generated by ComfyUI are only viewable in PNG format. ...But I just looked, and every single one of my images is in that format, so I'm not sure what's happening. When I click on the "image name" field I'm presented with quite a long list of images. These will ALL load just fine, but so far no image that is NOT on that list will load. :( I even downloaded a standalone app to read prompts from an image (by the same people who made the CUI node), and even it can't read the prompts from most of my images, unless they are on that list I mentioned earlier... Sigh... I would love to be able to report the prompt selected instantly, for wildcard prompts.

[@RuinDweller]

@@joeconway85One further thing, I'm noticing that even when I'm able to pull the metadata from images without using the prompt reader (or its standalone desktop counterpart), it only shows the FULL prompt, with all possible choices, which is of no use to me, rather than showing only the selections which were chosen. The only time I'm seeing those is when I generate right from your workflow. I have no idea why that's happening, but I'm working on it! Wish me luck, I suppose...