H3 Prompt Writing
Content Design Prerequisite
This skill compiles an already coherent visual and audiovisual design into MiniMax H3 syntax. When the task creates new generated content or substantively changes camera, composition, pose, location, furniture, background, lighting, subject count, reference roles, action, or shot order, first use the shared generative-prompt-design skill. Bring its completed Design Contract into this workflow; do not use H3 field structure to invent or repair the underlying situation.
Typo fixes, exact dialogue transcription, provider-schema corrections, reference-label formatting, and timestamp edits that leave the generated scene unchanged may proceed directly. A failed generation that requires semantic change is not a formatting-only edit.
Workflow
- Identify the input mode: T2VA, I2VA, FL2VA, L2VA, or full-reference Ref2VA.
- For base text/keyframe modes, read
references/base-en.txt and follow its final prompt structure.
- For full-reference mode, read
references/ref-en.txt and follow its six-section rewrite format.
- Preserve the exact field names, section order, labels, and timing notation from the selected guide.
- Before handoff, verify the final prompt fits the target interface. The local MiniMax H3 manifest loader accepts at most 8,000 characters; shorten by removing duplication after the Design Contract is complete, never by dropping a required causal or reference relationship.
Base Modes
- T2VA: build the full audiovisual timeline from text.
- I2VA: start from the first frame and develop forward from it.
- FL2VA: describe the continuous path between the first and last frames.
- L2VA: infer a plausible opening and converge to the supplied last frame.
Use integrated_multimodal_description, overall_soundscape, and non_diegetic_music in the order shown in references/base-en.txt.
Full-Reference Mode
Ref2VA rewrites use subject_definitions, summary, retention_analysis, detailed_description, overall_soundscape, and non_diegetic_music in that order. Reference labels stay consistent across all sections.
Read references/ref-en.txt for label rules, retention analysis, and complete examples.
Output Rules
- Write rewrite sections in English; preserve dialogue, lyrics, and visible scene text in their original language.
- Describe each shot by composition, subjects, environment, actions, camera, sound, and the exact point where referenced content appears.
- Avoid plot summaries, unresolved reference labels, and timing that does not match the requested duration.
1---2name: h3-prompt-writing3description: Write MiniMax H3 video generation prompts for T2VA, I2VA, FL2VA, L2VA, and Ref2VA. Use when rewriting multimodal requests into H3 prompt structures, composing integrated_multimodal_description, overall_soundscape, and non_diegetic_music, aligning keyframes, or defining reference labels for images, videos, and audio.4---56# H3 Prompt Writing78## Content Design Prerequisite910This skill compiles an already coherent visual and audiovisual design into MiniMax H3 syntax. When the task creates new generated content or substantively changes camera, composition, pose, location, furniture, background, lighting, subject count, reference roles, action, or shot order, first use the shared [generative-prompt-design](../../generative-prompt-design/SKILL.md) skill. Bring its completed Design Contract into this workflow; do not use H3 field structure to invent or repair the underlying situation.1112Typo fixes, exact dialogue transcription, provider-schema corrections, reference-label formatting, and timestamp edits that leave the generated scene unchanged may proceed directly. A failed generation that requires semantic change is not a formatting-only edit.1314## Workflow15161. Identify the input mode: T2VA, I2VA, FL2VA, L2VA, or full-reference Ref2VA.172. For base text/keyframe modes, read `references/base-en.txt` and follow its final prompt structure.183. For full-reference mode, read `references/ref-en.txt` and follow its six-section rewrite format.194. Preserve the exact field names, section order, labels, and timing notation from the selected guide.205. Before handoff, verify the final prompt fits the target interface. The local MiniMax H3 manifest loader accepts at most 8,000 characters; shorten by removing duplication after the Design Contract is complete, never by dropping a required causal or reference relationship.2122## Base Modes2324- T2VA: build the full audiovisual timeline from text.25- I2VA: start from the first frame and develop forward from it.26- FL2VA: describe the continuous path between the first and last frames.27- L2VA: infer a plausible opening and converge to the supplied last frame.2829Use `integrated_multimodal_description`, `overall_soundscape`, and `non_diegetic_music` in the order shown in `references/base-en.txt`.3031## Full-Reference Mode3233Ref2VA rewrites use `subject_definitions`, `summary`, `retention_analysis`, `detailed_description`, `overall_soundscape`, and `non_diegetic_music` in that order. Reference labels stay consistent across all sections.3435Read `references/ref-en.txt` for label rules, retention analysis, and complete examples.3637## Output Rules3839- Write rewrite sections in English; preserve dialogue, lyrics, and visible scene text in their original language.40- Describe each shot by composition, subjects, environment, actions, camera, sound, and the exact point where referenced content appears.41- Avoid plot summaries, unresolved reference labels, and timing that does not match the requested duration.
ryryo