MOSS-Image Character Lab
Community Hugging Face ZeroGPU app for MOSS-Image text-to-image generation with Base/SFT selection and four character LoRAs.
What happened lately
MOSS-Image Character Lab timeline
MOSS-Image Character Lab is tachiyumi’s community browser app for text-to-image generation on Hugging Face ZeroGPU. It offers MOSS-Image Base and SFT checkpoints alongside Alya, Emilia, Frieren and Marin character adapters, adjustable strength and a session contact sheet.
The English documentation says the rank-32 adapters were trained against Base with ai-toolkit. Applying them to SFT is described as approximate. The app reserves GPU time for one run at a time and checks estimated cost against a configurable ceiling before generation. The documentation’s approximately 35 GB resident-memory figure is an author estimate, not an independently reproduced measurement.
This is not an official OpenMOSS release. Model, inference-code and adapter rights need separate review; the adapter documentation says the source character frames remain the property of their rights holders. AITrending inspected the English documentation and browser controls but did not run inference.