Illustrious Workflow (with AI and upscale) I quite often use griptape and an AI service to re-imagine or tweak my prompts. This is the workflow that I use, took a while to get it in a state that someone other than I knew how to use it lol I also had to upload a custom node I had created to Comfyui Manager. Essentially the flow is this: you give it a prompt, it uses a LLM (its set up for featherless, but griptape is one of the most documented node sets i have ever seen with youtube videos on everything, making it work with local llm or another service is childs play) the LLM creates a list of 10 prompts, a custom node splits those prompts and they are then run through the workflow as 10 images. The 10 images are then upscaled using SD Upscale Script (the best upscaler I have found) I have added extensive notes to the workflow. so if your stuck there is likely a note for that lol. This is created for Illustrious but should work for pony or sdxl.. they are all the same beast to a degree. I advise tweaking the LLM rules to your liking, think of this as a starting point for you to make your own thing.... or use as is. Note on SD Upscale script: At the minute it does not download from comfyui manager like it should. To fix:navigate to: ComfyUI\custom_nodesDelete ComfyUI_UltimateSDUpscaleclick on the address bar of that folder type CMD<pre><code>type git clone https://github.com/ssitu/ComfyUI_UltimateSDUpscale --recursive</code></pre>this will clone the repo into your folder in the correct way and work. If you want to check what i am saying is legit (no offense taken) click manager, click custom nodes manager, locate the sd upscale custom node in the list and click its name. that will take you to the github page for it. scroll down and you will see the command i just gave you under installation (I hope it gets fixed on comfyui manager soon)if you get a git error...install git! for AI stuff you really should have it anyway <img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/24761daa-1f6b-493f-9821-cd7e3f4c2eba/width=525/24761daa-1f6b-493f-9821-cd7e3f4c2eba.jpeg" /> IMPORTANT: We are using open source projects, and free models, loras etc. To sell workflows or to lock them behind paywalls, to profiteer off the work of others, is a slap in the face of that ideal. Do not buy workflows, do not sell workflows. I was disgusted to see workflows behind paywalls (including ones that use non-commercial licensed nodes) when mentioning it I was called a freeloader and attention seeker. Those that know me from other places know just how wrong that statement is, those that do not know me from other places... well all I can say is that its almost a joke to be called a freeloader. If you need a workflow for a task, ask in the comments... I make no promises, I will give no time frame, but i will do my best to make it happen. The exception to this is the "one workflow fits all" (the ones that do video gen, image gen, train lora's, walk the cat, feed the fish and is good for sd3, flux, pony, sd1, etc etc), i get the appeal but its not really how comfyui is set up, and means a lot can go wrong. I wont make those. TL:DR. Do not buy or sell workflows. Final note: This is the first workflow I have submitted so I am hoping I have done it right. Also yeah I may end up creating more lora's again. I need to re-learn how to create them for modern models.

Files with SHA256: 3a9657e51222f84d06cea6a0351d435ac10f0cbb08a44f864fa0accfb261df9d