ComfyUI Node
M3 Song Planner (Local VLM)
Turns a creative idea into a production-ready MiniMax Music 3 caption + lyrics pair using a LOCAL VLM (Qwen3-VL / Gemma-3-Vision). Two-stage pipeline: caption generation → lyrics generation. No API keys needed.
M3 Song Planner (Local VLM)
- clip
- caption
- lyrics
- debug
◄text_encoder▾►
◄idea►
◄genre_hint►
◄vocal_configauto (decide from idea)►
◄languageEnglish►
◄duration_seconds120►
◄seed0►
◄temperature0.80►
◄top_p0.95►
◄top_k64►
◄max_tokens2048►
◄keep_model_loadedtrue►
CategoryM3/SongPlanner
Inputs (13)
| Name | Type | Default | Description |
|---|---|---|---|
| text_encoder | COMBO | VLM checkpoint from models/text_encoders/ (e.g. Qwen3-VL or Gemma-3-Vision instruct repack). | |
| idea | STRING | — | |
| genre_hint | STRING | — | |
| vocal_config | COMBO | auto (decide from idea) | 5 options: female vocals, male vocals, duet, instrumental, auto (decide from idea) |
| language | COMBO | English | 5 options: English, Chinese (Mandarin), Korean, Japanese, auto |
| duration_seconds | FLOAT | 12030–300 | Target song length. Affects structure and token budget. |
| seed | INT | 00–4294967295 | — |
| temperature | FLOAT | 0.800–2 | — |
| top_p | FLOAT | 0.950–1 | — |
| top_k | INT | 640–500 | — |
| max_tokens | INT | 2048512–8192 | Max tokens per generation (caption and lyrics each). |
| keep_model_loaded | BOOLEAN | true | Cache the text-encoder CLIP between runs. |
| clipopt | CLIP | Connect CLIP from CLIPLoader. Overrides the dropdown. |
Outputs (3)
| Name | Type | Description |
|---|---|---|
| caption | STRING | Wire to MiniMaxMusic3TextEncode 'caption' input. |
| lyrics | STRING | Wire to MiniMaxMusic3TextEncode 'lyrics' input. |
| debug | STRING | JSON with timings, clip source, model family, warnings. |