agentsclimarketplace
Sign in

Model routing

Skill ArcoCodes/renoise-plugins-official/skills/model-routing

Internal Renoise model router and model-specific prompting guide. Selects the best live image, video, or audio model for a task, then adapts the prompt to that model. Use before estimating or generating Renoise media.From its SKILL.md

Install
npx -y skills add ArcoCodes/renoise-plugins-official --skill model-routing

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.

SKILL.md

19.7 KB, as published. Nobody here has run it

Renoise Model Routing

Select by task, not by habit. Then prompt for the selected model instead of sending every model the same generic brief.

Hard Boundary

Live model capabilities are authoritative for availability, defaults, input roles, combinations, limits, duration, ratio, resolution, and audio controls. This Skill contains only researched routing preferences and prompting styles.

  1. Preserve a model explicitly named by the user.
  2. Filter candidates by the live capabilities required by the task.
  3. Choose the best specialist below.
  4. Use the live isDefault model only when no specialist clearly fits or candidates tie.
  5. Inspect the selected model's live guidance before writing the final prompt.
  6. Let the selected model's profile override generic prompt-craft advice when prompt structure, density, or reference wording conflicts.
  7. Never pass a role or parameter merely because this guide mentions a public model capability; the Renoise deployment may expose a narrower contract.

Routing Questions

Classify the request before selecting:

  • Kind: image, video, or audio.
  • Operation: create, edit, continue, interpolate, or animate a reference.
  • Priority: final quality, strict instruction following, aesthetics, identity consistency, speed, or cost.
  • Structure: exact text/layout, one cinematic shot, multi-shot narrative, dialogue performance, music, or a complete audio scene.
  • References: none, one source image, several identity/style references, source video, endpoint frames, or voice/audio references.

Do not call a model “best” without naming the task it is best for.

Image Routing

Decision Order

TaskPreferWhy
Exact text, UI, diagrams, ads, packaging, compositing, identity-sensitive edits, or peak final qualitygpt-image-2Strong instruction following, photorealism, typography/layout preservation, localization, and complex editing; favor it when fewer retries matter more than latency or cost.
General high-quality photorealistic or commercial imagelive image default, currently seedream-5-0-proStrong realism, dense information design, multilingual typography, multi-image compositing, and controlled local edits at the deployment's normal quality/cost balance.
High-throughput variants, prototyping, interactive generation, or cost-sensitive productionnano-banana-2-liteLow-latency scale tier for clear, shallow workflows; use another tier when the task depends on many references or sequential editing.
Balanced Google workflow, extreme aspect ratios, several references, multilingual localization, or conversational iterationnano-banana-2Google-family workhorse balancing quality, latency, text, reference reasoning, and broad formats.
Maximum Google-family world knowledge, brand consistency, localization, or reasoning-heavy compositionnano-banana-proSpecialist for intricate professional assets and precise spatial relationships.
Best aesthetic exploration, stylized art direction, editorial mood, or concept artmj-v8.2Current Midjourney default and strongest current aesthetic prior.
Faster Midjourney iteration, small-detail retention, or a familiar V7-like lookmj-v8.1Distinct speed/detail specialist when that version is live; V8.2 remains the normal aesthetic choice.
Knowledge/reasoning-heavy visualization, many references, or 2K–4K Seedream workseedream-5-0-liteStrong intent inference and relational editing; Pro has the higher realism, structural-stability, and aesthetic ceiling.
xAI image requestgrok-image for lower-cost iteration; grok-image-quality when higher quality justifies the added costBoth support direct natural-language generation and editing through the live Renoise contract.

Do not auto-select

  • midjourney-v7: keep for explicit requests, validated V7 compatibility, or its known treatment of bodies, objects, and references.
  • Older model versions remain valid specialists; version number alone is not a routing reason.

Image Prompting Styles

GPT Image 2 — production brief

For complex production briefs, use labeled, ordered instructions:

USE: [ad / packaging / UI / infographic / edit]
SCENE/BACKGROUND: ...
SUBJECT: ...
COMPOSITION: ...
LIGHTING/MATERIALS: ...
EXACT TEXT: "..." with placement, hierarchy, font character, and color
PRESERVE: identity, geometry, layout, logo, and all unmentioned elements
CHANGE ONLY: ...
AVOID: ...
  • Put literal in-image text in quotes and specify hierarchy and placement.
  • For edits, say change only X; preserve everything else.
  • Identify every reference by its job: source, identity, product, layout, or style.
  • Prefer one-change iterative edits over rewriting the whole brief.

Nano Banana family — conversational creative direction

[Subject] + [action] + [location] + [composition] + [style]
Camera/viewpoint: ...
Lighting/materials: ...
Exact text: "..."
Keep unchanged: ...
  • Start with a clear operation verb and use positive, concrete natural language.
  • For references, state each source's job, the relationship among them, and the new scenario.
  • For edits, separate the requested change from what remains unchanged; repeat preservation requirements on later turns.
  • Quote exact visible copy and specify its hierarchy, placement, visual character, and localization target.
  • Nano Banana 2: relate multiple references explicitly and iterate conversationally.
  • Lite: keep one clear composition and few dependencies rather than relying on long multi-turn edits.
  • Pro: state brand invariants, localization requirements, and complex spatial relationships precisely.

Seedream 5 — design brief

Put the deliverable format first, then subject, layout, light, exact text, and invariants:

[DELIVERABLE FORMAT and ratio/use case]
Subject and setting: ...
Layout and spatial relationships: ...
Lighting, materials, and palette: ...
Exact visible copy: "..."
Keep exactly: ...
  • Pro: specify dense layout, multilingual copy, materials, local edit regions, and everything the edit preserves.
  • Lite: natural language works well; make causal/spatial relationships explicit for complex transformations.
  • Assign each reference a clear role and avoid vague pronouns in multi-reference edits.

Midjourney — aesthetic visual phrase

Use a concise visual description, not a requirements document:

[subject], [medium], [environment], [lighting], [palette], [mood], [composition]
  • Prefer concrete visual nouns and adjectives over keyword spam.
  • Name the lighting and composition that matter instead of relying on “cinematic.”
  • Use Midjourney for look development rather than text-heavy layouts or surgical edits.

Grok Imagine Image — concise scene direction

[subject doing action] in [setting], [camera/framing], [lighting], [material/style]
  • Use standard for lower-cost iteration and Quality when the higher-quality tier is worth the added cost.
  • For edits, attach the source image or images and describe the requested change directly.
  • Keep the prompt concrete enough that the main subject and action remain dominant.

Video Routing

Decision Order

TaskPreferWhy
Explicit Seedance 2.5 request, continuous 16–30 second sequence, rich mixed references, source-video edit, or forward/backward extensionseedance-2.5-byteplusLong-form and reference-heavy specialist with precise edit/extension workflows.
General multimodal video, complex physical motion, recurring references, product/character continuity, or image-to-video with synchronized audioseedance-2.0-byteplusLive generalist with strong image/video/audio role assignment, motion, continuity, and native sound.
Fast Seedance draftseedance-2.0-fast-byteplusOfficial speed/cost balance tier.
Lowest-cost/high-volume Seedance draftseedance-2.0-mini-byteplusCost-performance tier for iteration and selection.
Short 720p generation, text rendering, or source-video edit within its narrow live contractgemini-omni-flashStrong short-form instruction following, multi-shot generation, and conversational editing.
Exact first frame, last frame, frame interpolation, focused video references, 2K/4K delivery, or reference audio paired with visual referenceshailuo-h3Strong endpoint control and structured multimodal audiovisual direction.
Human action, dialogue acting, lip sync, atmospheric commercial work, or concise cinematic multi-shot sceneshappyhorse-1.0Strong motion, physical plausibility, visual depth, and synchronized dialogue/Foley/ambience through the inputs Renoise exposes.
Reusable subjects/products, reference-driven series, identity/voice continuity, or custom multi-shot storieskling-3.0-omniReference-first specialist with native audio, subject consistency, and explicit storyboard control.
xAI text-to-video or one-image animation with native soundgrok-video-1.5Current Renoise 1.5 contract is strongest when zero or one image is sufficient.
xAI video needing several image references or the older Renoise Grok contractgrok-videoMulti-image and lower-duration niche currently exposed by Renoise.

Tier and version notes

  • Seedance Full, Fast, and Mini are respectively the quality, speed/cost-balance, and cost-performance tiers; live estimates decide the actual trade-off.
  • Grok's public upstream capabilities may move faster than Renoise's contract. Route from live roles and limits rather than assuming an upstream feature is connected.
  • Model preference leaderboards are task-, resolution-, and audio-filter-specific; use them as volatile evidence, not a single aggregate ranking.

Video Prompting Styles

Seedance family — multimodal director brief

Assign each reference an explicit job, then write shots:

References:
- @material:<ID>: subject identity
- @material:<ID>: location/style
- @material:<ID>: motion/camera/voice

Shot 1: framing, subject action, camera move, environment, sound/dialogue.
Shot 2: ...
Continuity: traits and objects that must remain unchanged.
Constraints: no subtitles/logo/watermark unless requested.
  • Use a few stable identity features rather than an exhaustive biography.
  • Give each beat a clear camera behavior; avoid simultaneous conflicting moves.
  • Describe body part, speed, force, and physical consequence for actions.
  • Externalize emotion through visible behavior and direct dialogue/audio explicitly.
  • Use the same prompt structure for Fast/Mini while keeping draft requests easy to compare.
  • For Seedance 2.5, begin with the intended result, map every reference to its purpose, then use non-overlapping timestamp ranges or numbered shots with continuity notes.
  • For a source-video edit, name the source, the intended change, its time range when relevant, and what remains unchanged.
  • For extension, state forward or backward direction and describe the visual, motion, and audio continuity across the source boundary.

Gemini Omni Flash — conversational generation/editing

  • For creation: write a direct natural-language brief with subject, action, camera, setting, visible text, and desired audio.
  • Gemini readily creates multi-shot clips; say “single continuous shot,” “single unbroken scene,” or “no scene cuts” when continuity is required.
  • Use natural time phrases or [0–3s] blocks for important beats.
  • For editing: identify the source, say change only X, and list what remains unchanged.
  • Make one edit per turn when possible; use follow-up refinement rather than replacing the entire prompt.

MiniMax H3 — mode-specific audiovisual plan

Choose one live mode and prompt accordingly:

  • First frame: describe only the motion, camera path, action development, and sound after the supplied opening state.
  • Last frame: describe the plausible path that converges on the supplied ending.
  • First + last: describe the continuous transition between states; avoid re-describing the two stills. Prefer one coherent shot unless a cut is essential.
  • Reference mode: assign each image, video, and audio an explicit identity, motion, style, voice, action, or sound job.

For complex reference work use:

References: each @material:<ID> image/video/audio and its job.
Task summary: target result and what each source contributes.
Preserve: identity, motion, voice, style, or content that carries over.
[Shot 1] Visual action, camera motion, dialogue, and diegetic sound.
[Shot 2 at time] Cut and continuation.
Overall soundscape: ambience, Foley, and non-verbal sound.
Non-diegetic music: instrumentation, tempo, dynamics — or none.

Specify camera type + amplitude + speed only when they matter. Keep dialogue exact, label speakers consistently, and separate in-scene sound from background score.

HappyHorse 1.0 — concise performance direction

[subject] [specific action] in [setting], [lighting], [one camera cue].
Dialogue in [language]: "..."
Sound: foreground action, midground Foley, background ambience; no music if unwanted.
  • Start with the entity, scene, and motion; add framing/lens, lighting, camera movement, style, and synchronized sound when they matter.
  • For image-to-video, emphasize motion and camera development rather than re-describing the still.
  • Use explicit shot structure for multiple beats and detailed cinematic direction when the scene benefits from it.

Kling 3.0 Omni — explicit storyboard

Shot 1 (duration): framing, named subject, action, camera, light, sound.
Shot 2 (duration): reaction or continuation, camera, exact dialogue and speaker.
  • Define each reference's subject, product, motion, scene, or voice job before the shot plan.
  • Once appearance or voice is bound by a reference, focus the prompt on action, interaction, camera, and story progression.
  • Keep each shot to one readable action and one camera behavior.
  • Write dialogue after the associated action; use “then” or “immediately” for timing.

Grok Imagine Video — animate the change

For image-to-video, do not re-describe the static source:

[subject motion/change], [one camera move], [atmosphere/lighting change].
Sound: specific dialogue/SFX/ambience; no music if unwanted.
  • Use one primary action or a short causal sequence and front-load the important motion.
  • For text-to-video, include the subject and setting; for image-to-video, describe the change from the starting frame.
  • Assign every live reference a clear visual or motion job.
  • Request dialogue, effects, ambience, or music explicitly when native audio matters.

Audio Routing

TaskPreferWhy
Short song excerpt, instrumental, loop, preview, score cue, or music bedlyria-clipDedicated 30-second music model for rapid iteration, social assets, and background cues.
Dialogue, expressive narration, podcast, dubbing/re-voicing, SFX/ambience, or a complete sound sceneseed-audio-1.0Unified speech, dialogue, effects, and ambience generation; live default audio model.

Neither broadly replaces the other: Lyria composes short music clips; Seed Audio handles speech-led and complete sound scenes.

Lyria 3 Clip — composer brief

Genre/style: ...
Mood: ...
Instrumentation: ...
Tempo/rhythm: ... BPM
Vocals: instrumental OR voice type, delivery, and language
Lyrics/theme: exact lyrics or subject
Structure over ~30 seconds: intro → development → ending
Production: mix character, era, space, dynamics
  • Explicitly say instrumental when vocals are unwanted.
  • Name instruments, tempo, vocal range/texture, backing vocals, and language.
  • Keep custom lyrics concise for 30 seconds; Lyrics:, [Verse], and [Chorus] can clarify structure.
  • Use timestamps only for meaningful structural changes; musical alignment follows bars rather than sample-accurate timing.
  • Describe genre, era, instrumentation, and production traits instead of requesting a named artist imitation.
  • Use an optional guide image only when the live material roles expose it.

Seed Audio 1.0 — sound-scene script

Setting/acoustics: ...
Continuous sound bed: ...
Speaker A (age, accent, timbre, emotion, pace): "Exact line."
Action-tied SFX: ...
Speaker B (...): "Exact line."
Music cue and ending: ...
  • For full-scene work, direct it as a scene; for speech-only work, specify voice identity, performance, pacing, and acoustics.
  • Label speakers and exact lines; match prompt language to dialogue language.
  • Use clean authorized voice references, or an authorized character image for inferred voice, according to the mutually exclusive live reference modes.
  • Use concrete Foley and ambience; onomatopoeia can clarify transient sounds.
  • Use precise timing for dialogue; describe SFX, ambience, and music timing approximately unless live guidance says otherwise.

Research Basis

Reviewed 2026-08-11. Rankings are directional and decay quickly; live capabilities and task-specific tests override them.

Primary guidance:

Keep looking

Skills are one crate of 616,272. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.