i2i — image-to-image. First task: colorizing a B&W manga page with the colorize LoRA trained in anima_lora.
Layout JSON — an Anima-format caption as MOSS's layout-JSON prompt, for LoRAs trained on caption_format=layout_json captions.

i2i

MOSS-Image has no image conditioning. This node adds two paths:

Vision tower. The page goes through Qwen3.5's vision tower ahead of the text in the same user turn. This brings back the page's content, but not its line placement (one token per 32×32 px).
VAE reference. The page also goes in as a clean VAE latent: an extra token block [image | ref | text] on RoPE's reserved r = 1 axis. Its (h, w) positions are stretched onto the target grid. A LoRA trained with these refs keeps the page's line work. A LoRA stamped with ref_timestep (colorize v2: 0) gets the ref rows modulated at that timestep in every block; without it (v1) they share the step's.

Nodes

| Node | In → out | Notes | |---|---|---| | MOSS i2i LoRA Loader | MODEL → MODEL | Plain lora_unet_layers_* LoRA. Reads reference_latents and ref_timestep from the metadata for the Reference node | | MOSS i2i Text Encode (page + text) | CLIP, IMAGE, text → CONDITIONING | <|vision_start|>[page]<|vision_end|>{text}. Empty text = page only. The page is fit to the target size and seen at that pixel count (~1000 tokens at 1 MP) | | MOSS i2i Reference | MODEL, VAE, IMAGE → MODEL, LATENT | Ref at 1/ref_scale per side (2 = 1/4 area). The ref goes into every forward and every CFG branch. Also outputs an empty latent at the target size |

Wiring

MOSS-Image loaders ─ MODEL ─ MOSS i2i LoRA Loader ─ MOSS i2i Reference ─ MODEL ─┐
                   ─ CLIP ─┬ MOSS i2i Text Encode (page, colour tags) ─ positive ─┤
                           └ CLIP Text Encode ("") ─────────────────── negative ─┤ KSampler /
                   ─ VAE ──── MOSS i2i Reference ───────────────────── LATENT ───┘ MOSS Spectrum KSampler
Load Image (page) → Text Encode.image, Reference.image

Use the same resolution on Text Encode and Reference. The negative is a plain text encode of "", which is what the LoRA was evaluated with. For CFG 3 with the uncond block-22 skip, use the MOSS Spectrum KSampler.

Compatibility

Spectrum (MOSS). Cached steps never call the DiT, so they cost no ref rows. On full steps the capture hook sees the image rows only.
DAVE. Its hook reads the leading image rows. The ref rows come after them.
SLG / uncond skip. The patches_replace dispatch is kept, so it still works.

The patch is an APPLY_MODEL wrapper. For the length of each apply_model call, the live diffusion_model runs a ref-aware copy of the moss-image forward.

Layout JSON

| Node | In → out | Notes | |---|---|---| | Anima Caption → MOSS Layout JSON | STRING → STRING | anime_tools.captions.layout_json, byte-identical to what the anima_lora MOSS trainer encodes with caption_format=layout_json. Rating / meta, characters, series and @artists get their own summary keys, the other tags become summary.caption, each position clause a character item, each quoted text line a text item |

Feed its output into CLIP Text Encode with MOSS-Image's text encoder. It needs anime-tools; without it the toolkit loads the i2i nodes only and logs a warning:

pip install 'anime-tools @ git+https://github.com/sorryhyun/anime_tools'

The danbooru tag KB (danbooru_tags_classified.csv) is looked up under $ANIME_TOOLS_HOME / $ANIMA_HOME / the ComfyUI folder (models/) and downloaded there on first use. The node used to ship in the Anima Tagger pack; the node id (AnimaCaptionToMossLayout) is unchanged, so old workflows load as-is.

License

Apache-2.0. reference.py derives from the moss-image node's MOSSImageTransformer (OpenMOSS, Apache-2.0).