sorryhyun/MOSS-I2I-LORAs

MOSS-Image i2i LoRAs

Image-to-image LoRAs for MOSS-Image-Base, trained with anima_lora. The input image enters the DiT as a clean VAE reference latent. One folder per task.

Folder	Task
colorize/	colorize a B&W manga page or line-art illustration
sanitize/	remove text, SFX, speech-bubble text and watermarks from an illustration
decensor/	remove bar and mosaic censors

ComfyUI

Needs ComfyUI-MOSS-Toolkit 0.6.1 or later, the moss-image node, and the Spectrum KSampler pack for the MOSS Spectrum KSampler. MOSS Caption Helper registers only when anime-tools is installed in ComfyUI's environment (pip install 'anime-tools @ git+https://github.com/sorryhyun/anime_tools').

colorize

1. Put colorize/moss_colorize_3k_v3.safetensors in models/loras/.
2. Load colorize/moss-i2i-colorize.json and pick your page in Load Image.

How the workflow is set up:

• MOSS i2i Reference: the LoRA, task colorize. The LoRA's metadata turns on ref_isolate and ref_timestep 0; task colorize reads them and puts the page in at full size.
• Positive: MOSS Caption Helper with task colorize into a plain CLIP Text Encode. The helper takes the page too: it OCRs the lettering and adds each line to the layout caption, which is how the LoRA redraws the lettering. Type colour tags in its caption field (touhou, purple eyes, purple hair, blue shirt), or leave comic for a manga page.
• Negative: worst quality, greyscale, blurry, jpeg artifact.
• Sampler: MOSS Spectrum KSampler, 50 steps, CFG 3, er_sde, uncond skip on block 22.

Trained on 3,334 colour / B&W pairs, half illustrations and half manga pages (rank 32, 8,000 steps).

sanitize

1. Put sanitize/moss_i2i_sanitize.safetensors in models/loras/.
2. Load sanitize/moss-i2i-sanitize.json and pick your image in Load Image.

How the workflow is set up:

• MOSS i2i Reference: the LoRA, task custom, ref_scale 1 (full-size reference). The LoRA's metadata turns on ref_isolate and ref_timestep 0.
• Positive: MOSS Caption Helper with task sanitize into a plain CLIP Text Encode. It outputs no sfx, no dialogue, no watermark, the one caption the LoRA was trained on, so don't type your own prompt.
• Negative: CLIP Text Encode with an empty string.
• Sampler: MOSS Spectrum KSampler, 30 steps, CFG 4, uncond skip on block 22.

The output keeps the input's aspect ratio, with the area capped at resolution² (1024). Trained on 49 texted / clean pairs (rank 32, 588 steps).

decensor

1. Put decensor/moss_i2i_decensor.safetensors in models/loras/.
2. Load decensor/moss-i2i-decensor.json and pick your image in Load Image.

The sanitize setup, with MOSS Caption Helper on task decensor: it outputs no bar censor, no mosaic censor, uncensored, the LoRA's one training caption. Same sampler (30 steps, CFG 4) and output size. Trained on 48 censored / uncensored pairs (rank 32, 576 steps).