Lazycaster: I haven’t been following the text-to-image space for a little while—looking for some recommendations on native fine-tuning, both on which base models take to fine-tuning well and also frameworks/scripts/recipes for native fine-tuning.

For those who need a little more architecture info/context, I’ll be doing the fine-tuning in Google CoLab Pro, so I won’t need hacks to get around desktop GPU VRAM limitations.

✵ Create ✵ Share ✵ Remix ✵

Experimental 2D/3D artist. Data science Pythonista. He/They. Nerdy++; (Nearly) all work CC-BY/CC0/FOSS

https://mxvoid.com

Even more context:

Currently using some SDXL native fine-tunes trained off my own art that I’m very happy with, but I’m curious to see what the latest crop of models and tools are.

i still havent tried it but flux came out with a new finetuning api that looked pretty good

flux on comfy ui might be what your looking for or pretty fun on krea with use of your and others models - google imagen also pretty good, I'm using on nim.video and believe you can get there thro ggogle labs.

On the app that will not be named AiAT still meets every morning for the most part 8-9 ish ET AM to whenever it rugs