Updated: Oct 31, 2026
base model
What's the next stage?
About This Model
This is a merge model based on MOSS-Image-Base, combining multiple LoRAs into a single model.
Important Notes
Generation Steps: The
MOSS-Image-SFT-8-steps-dmdLoRA was merged at a reduced strength. Please use 12 steps for generation.Prompt Adherence: Due to the reduced LoRA strength and the nature of this merge, prompt adherence may be weaker than expected. If necessary, use the MOSS-Image Prompt Weight node to adjust prompt strength to your liking.
Original Character Reproduction: At this stage, the model has significant difficulty reproducing original characters consistently. Results may vary, so please keep this limitation in mind.
Recommended Settings: I haven't found the optimal settings for this model yet. Feel free to experiment with different configurations and find what works best for your workflow.
This model is still an experimental prototype, and its results may not meet everyone's expectations. If you're looking for a more polished anime-style model, you may want to wait for the official Anime Version planned by the original developers.
In the meantime, feel free to experiment with different settings and see what works best for you. Have fun creating!
Recommended settings
Sampler: er_sde
CFG: 3
Steps: 12
Quality tags:
masterpiece, best quality, highres, absurdres※The quality tag must be at the beginning
Negative prompt:
worst quality, low quality, blurry, jpeg artifacts, sepia, username, watermark,