[Important] What is now and what's to come
Added 2026-01-05 01:27:54 +0000 UTCTLDR: I'm pivoting from full AI to 3D, or a mix of the two for now.
I feel like the gen AI has plateaued. The anime/drawing branch at least. New checkpoints are being released, new LoRas, but no amount of the restraints can ever give you the exact result. Especially when it comes to very specific setting or composition. You want some character in a generic pose, in a generic setting? - it'll do just fine. But not what I want.
Inpainting is great, I can draw a base sketch, generate and refine individual parts. However, there's only so much I can control or give. I am still relying on those concepts to be in the model's dataset, which is becoming an issue.
So, then next step for me is to switch back to 3D. It will take longer to make. It will look worse (for now). And I will lose some characters (drawn, that don't have a 3D model). But it IS, what I want it to be. Every setting, every pose, every detail.
Which I guess to most is an obvious conclusion. I did bet on generative AI specifically because of how unique I wanted to setting to be. And now I'm back to the drawing board, quite literally.
I will continue to use AI, at least for now, for drawn/anime characters, until I figure out Blender's new compositor to mimic 2D drawing.
Comments
Yes! AI art is becoming increasingly homogenized. It's worth trying something different!
doralice
2026-01-05 18:22:34 +0000 UTCMakes sense to try new things. Interesting to see how new stuff will look.
Kazeman
2026-01-05 11:04:40 +0000 UTCSounds like a good plan to me. AI models were always going to be limited by their datasets, so if you're running into roadblocks, changing up the workflow is worth exploring.
Chris
2026-01-05 02:22:04 +0000 UTCWhat are your thoughts?
NordSlate
2026-01-05 01:28:55 +0000 UTC
