Which Model to Choose for Character LoRA
Baddmaan0 · reddit · 2026-07-19
The poster found that while training a character LoRA, they achieved stable results on Z-Image Turbo, but the results were underwhelming after switching to Z-Image Base, suspecting an issue with training configs or parameter settings.
They also mentioned trying Krea 2 and wanted to know if anyone has successfully trained a better character LoRA on the Base version, and what configurations were used. Essentially, the post is a call for experiences and reproducible practices for training LoRAs across different image generation models.
More from Multimodal
- Dense instance segmentation via Muse Wonder + SAM 3.1 on Meta's Model API — nikhilaravi · 2026-10-26
- LichtFeld Studio v0.5.4.3 ships community-built SteadySplats, POPSpa and Voxelization — janusch_patas · 2026-10-26
- One prompt to a 2-3 min narrated short: hands-on test of Antigravity video generation — AI_Andrew · 2026-10-26
- Creator makes 10,564-frame animated short entirely with Gemini toolchain — AI_Andrew · 2026-10-26
- FLUX.3 Klein step-distilled to 4-8 steps with CFG baked in — Fit-Structure-6489 · 2026-10-26
- ostris pre-training "Primal VAE": 1-channel latents match modern 8x VAEs via patching — ostrisai · 2026-10-26