Global-AI-Group-Inc/content-generation-skills / skills/mira-video-prompting
mira-video-prompting
VerifiedWhich of Mira's video models to pick (the live list is in references/selection.md) and how each one treats a prompt: Seedance 2.5, Runway Gen-4.5, Kling 4.0, Kling 3.0 (and Turbo), Gemini Omni Flash, MiniMax H3, HappyHorse, Wan 3.0. Use it before any generate_video call on Mira, when the user asks "which model", "how long can it be", "does it have sound", "промпт для видео", or pastes a video prompt to improve. NOT for stills (mira-image-prompting) or for the camera and light banks themselves (mira-craft).
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 5.3 KB |
| references/action.md | markdown | 4.2 KB |
| references/models/capricornus-1.md | markdown | 9.9 KB |
| references/models/happyhorse.md | markdown | 5.8 KB |
| references/models/kling-4.md | markdown | 5.2 KB |
| references/models/kling-motion.md | markdown | 2.2 KB |
| references/models/kling-turbo.md | markdown | 2.0 KB |
| references/models/kling.md | markdown | 8.2 KB |
| references/models/minimax.md | markdown | 7.7 KB |
| references/models/mira-cinematic.md | markdown | 5.3 KB |
| references/models/omni.md | markdown | 5.7 KB |
| references/models/runway.md | markdown | 5.6 KB |
| references/models/seedance-2-5.md | markdown | 9.9 KB |
| references/models/veo-lite.md | markdown | 5.1 KB |
| references/models/veo-quality.md | markdown | 5.1 KB |
| references/models/veo.md | markdown | 5.0 KB |
| references/models/wan3-prime.md | markdown | 7.9 KB |
| references/models/wan3.md | markdown | 7.8 KB |
| references/selection.md | markdown | 6.3 KB |
| references/textures.md | markdown | 10.8 KB |
| references/universal-rules.md | markdown | 7.8 KB |
references/selection.md
View on GitHub<!-- generated from the registry by comfy_study/comfy_video/build_skills.py; edit the registry, not this file -->
# Selecting the model
Facts per model, straight from the registry. The routing rule is in SKILL.md; this is the
table it points at.
| key | model | seconds | output | references | audio | prompt budget |
|---|---|---|---|---|---|---|
| `seedance-2-5` | ByteDance Seedance 2.5 | 4-30s, any whole second | 720p; 1080p with quality=hd; 480p draft, finalized to 1080p | up to 30 native | yes | 6000 |
| `capricornus-1` | Capricornus 1 (Mira) | 4-30s, any whole second | 720p; 1080p with quality=hd; 480p draft, finalized to 1080p | up to 30 native | yes | 6000 |
| `runway` | Runway Gen-4.5 | 5 / 10 | provider default | up to 4, merged into one start frame | no | 1000 |
| `kling` | Kling 3.0 | 3-15s, any whole second | 1080p by default; 720p with quality=sd; 4K with quality=uhd | up to 7 native | yes | 2500 |
| `kling-4` | Kling 4.0 | 3-30s, any whole second | 720p by default; 1080p with quality=hd | up to 7 native | yes | 5000 |
| `kling-turbo` | Kling 3.0 Turbo | 3-15s, any whole second | 1080p by default; 720p with quality=sd | up to 1 native | yes | 2500 |
| `kling-motion` | kling-motion | 5 / 10 / 15 / 30 | 1080p by default; 720p with quality=sd | up to 1 native | yes | 2500 |
| `omni` | Google Gemini Omni 1.2 Flash | 3-15s, any whole second (3-10s with a first and a last frame) | 720p by default; 360p with quality=sd; 1080p with quality=hd; 4K with quality=uhd | up to 6 native | yes | 3000 |
| `minimax` | MiniMax H3 (Hailuo 3.0) | 4-15s, any whole second | 1080p by default; 720p with quality=sd | up to 9 native | yes | 6400 |
| `happyhorse` | HappyHorse 1.1 | 5 / 10 / 15 | 1080p by default; 720p with quality=sd | up to 1 native | yes | 400 |
| `wan3` | Wan 3.0 | 2-30s, any whole second | 1080p by default; 480p with quality=ld; 720p with quality=sd | up to 10 native | yes | 8000 |
| `wan3-prime` | Wan 3.0 Prime | 2-30s, any whole second | 1080p by default; 480p with quality=ld; 720p with quality=sd | up to 10 native | yes | 8000 |
## What each model is for
- **`seedance-2-5`** — the long-form specialist: up to 30s, heavy referencing (storyboards, keyframe chains, video edit/extend); ~1.6x the price of seedance and pricier than kling-4 or wan3, which also reach 30s - reach for it when length or reference volume decides the clip. ALSO the camera-trick model WITH audio: in live runs (2026-09-06) it executed every phased single-take trick as written - a top-down crash dive, a full 360 orbit returning to the start, a whip pan fusing two locations, a dolly zoom, a camera struck and falling while it records, a locked day-night cycle, a pass through a mirror into a new place - where Runway smeared or split half of them and Kling stopped short of the move; when the MOVE is the concept and the clip needs sound, this is the key
- **`capricornus-1`** — Mira's own model (early access): the house cinematic look - large-sensor optics, soft motivated light, warm highlights over cool shadows, fine grain - written into every shot, so a clip looks graded out of the box; the pick for product films, brand pieces and multi-shot stories. Up to 30 s in one pass with native audio, heavy referencing, timecoded storyboards, extends both ways, and about 15% cheaper per second than the other long-form option. It also executes phased single-take camera tricks as written - crash dive, full 360 orbit, whip pan between two places, dolly zoom, day-night cycle.
- **`runway`** — Gen-4.5, text-to-video AND image-to-video: the sequencing specialist - executes several ordered camera moves and timed beats inside ONE prompt, and takes forceful camera work (sweeping arcs, crash zooms, whip pans) the others smear; weak on fine motor detail; NO audio; 720p; 5/10s
- **`kling`** — best human anatomy and physics (sport, cloth, water), legible in-frame text, native audio with multi-speaker dialogue and accents; 3-15s; 1080p output
- **`kling-4`** — the newest Kling: ONE continuous take of 3-30s, 720p by default and 1080p with quality=hd, native audio always on, 21:9 available; best-in-line anatomy, physics and legible in-frame text; up to 7 photo refs or one reference clip, start and end frames; slightly cheaper than kling; no multi-shot and no 4K - for those use kling
- **`kling-turbo`** — the faster, cheaper Kling 3.0: ONE continuous take from text or a single start frame, native audio always on; 3-15s; 1080p; no reference photos, clips, end frames or multi-shot - for those use kling
- **`kling-motion`** — motion transfer: ONE photo of a character + ONE driving video (3-30 s, one person, single take) - the character in the photo performs the motion of the video; output length = video length; 1080p; keeps the clip's original sound. The only model that copies choreography exactly. Needs both inputs - refuse without a video
- **`omni`** — self-directing storyteller and the only model that can EDIT an existing clip: multi-shots by default, renders on-frame text well, 3-15s at the length you pass (any whole second; 3-10s with a first and a last frame); 720p, plus a cheap 360p tier for quick tries and 1080p / 4K tiers; up to 6 photo refs and first/last frame anchors, but NO clip input at all - its video port is closed on our key
- **`minimax`** — the purpose model for SPEECH plus legible/Cyrillic in-frame text - the only one holding both; millisecond-precise shot timing; 4-15s; NOT a general default
- **`happyhorse`** — dialogue-first with 7-language lip-sync; single-beat clips and exotic vertical ratios; briefs must stay tiny; ONE photo only - the nine-reference route exists but is not paid on our key
- **`wan3`** — the long-take model: 30 seconds in ONE continuous generation, so a whole story keeps one look and one cast - no stitching; shot-numbered storyboard with timecodes, native audio, 10 photo refs; 2-30s; 1080p
- **`wan3-prime`** — the accelerated Wan 3.0: the same long take (30 seconds in ONE generation, one look and one cast), the same shot-numbered storyboard, native audio and 10 photo refs, rendered several times faster; ~1.4x the price of wan3; 2-30s; 1080p
Withdrawn from the lineup (still in the registry for old jobs and pricing): `veo`, `veo-lite`, `veo-quality`, `seedance`, `wan`, `grok`. Do not pick them.
The registry default is `omni`. An empty or unknown model id resolves to it silently.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 323k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 323k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 323k | Updated | Claude CodeCodex | Verified | Install |