MOSS Image tested with six labeled cat breeds — full prompt and Krea comparison
Dunc4n1d4h0 · reddit · 2026-10-18
A follow-up test of MOSS Image replaces the usual '1girl' prompt with six cat breeds (Persian, Siamese, Scottish Fold, Tabby, Tuxedo, Ginger) dancing in pairs with Courier-font breed labels, probing multi-subject layout and text rendering. Settings: SFT, 40 steps, cfg 3, euler/sgm. A Krea run with the same prompt (no LoRAs, 10 steps, ersde/beta) serves as comparison. Full prompt included for reproduction.
More from Multimodal
- Style LoRAs broke across all models after reinstalling ComfyUI — PhilosopherSweaty826 · 2026-10-18
- MOSS-Image + LoRAs recreate a Star Wars D-Day landing scene — FortranUA · 2026-10-18
- Luma adds Send to Luma inside Claude Motion, turning animated explainers into real video projects — lmoroney · 2026-10-18
- AI model spontaneously spawns its own procedural audio composer for a numpy-made vaporwave cat — teortaxesTex · 2026-10-18
- A 3D video of Shenzhen's 44 years, made entirely with DeepSeek — zizhpan · 2026-10-18
- AI film 'The Gifted' wins $2.5M grand prize at Future Vision XPRIZE in LA — Glittering-Trick5543 · 2026-10-18