How do you actually keep track of prompts that work?

Curious if anyone here has cracked cross-model prompt management, or if you just stay in ComfyUI for everything?

https://redd.it/1teskau
@rStableDiffusion
Character Sheet Worflow or Lora?

I am looking for a way to create high quality character sheets with loras that I trained.
I did face loras in f1 and fluxklein 9b.

There are tutorials with prompting for Nano Banana from one or more Images but it looks not as good I expected.
And some seems to get great results with a single Image on Higgsfield or other saas.

But what would be the ideal process.
It would be ideal to generate one Face close up and one full body shot to any workflow or lora that understands the task!?

Do you guys do that with openpose?
Multiangle with qwen I tried already, however it could have been better.

So if there is any idea how I can process this task, please let me know your ideas.

Currently I make a promptlist and hope to get the results I need.

Thanks a ton!

https://redd.it/1tf19vr
@rStableDiffusion
Aww look at th-AAAHHHH what the F is that
https://redd.it/1tf5f43
@rStableDiffusion
INSTARAW 2.0 FULL WORKFLOWS

[https://civitai.com/articles/29974](https://civitai.com/articles/29974)

[https://huggingface.co/rizzlaa/INSTARAW-2.0-WORKFLOW-CUSTOM](https://huggingface.co/rizzlaa/INSTARAW-2.0-WORKFLOW-CUSTOM)

This pack includes **14 ready-made ComfyUI workflows** designed for:

* Maximum photorealistic image generation
* Face swap
* Detail enhancement and final polishing

Most workflows use the **Grok API** and **Nano Banana**
(**Google Gemini 2.5 Flash/Pro Image**).

If you do not configure the API, some nodes may appear red.


https://redd.it/1tf6v66
@rStableDiffusion