Luma Modify Video (Ray3.2)
Video-to-video editing mode of Ray3.2. It re-renders existing footage from a natural-language prompt while conditioning on the source, with nine strength presets across three bands (Adhere, Flex, Reimagine) that trade structural fidelity against creative reinterpretation.
- Up to
- 18 seconds
- Resolution
- 1080p
- Aspect ratios
- 9:16, 3:4, 1:1, 4:3
- Weights
- Not released
- API
- Yes
- Commercial use
- Not established
Per-signal conditioning is available for pose, depth, normals, trajectory and face.
No licence or terms page was retrieved for this model during research, so the commercial-use position is not established here. Check the vendor's own terms.
Specification
- What it takes in
- video, text, image
- What it makes
- video
- Maximum duration
- 18 seconds
- Maximum resolution
- 1080p
- Aspect ratios
- 9:16, 3:4, 1:1, 4:3, 16:9, 21:9
- Controls
- keyframes, reference-character-consistency, inpainting
- Weights
- Not released
- API
- Yes
- Pricing
- API pay-as-you-go: video editing $0.54-$2.16 per 5 seconds depending on resolution; also included in web app credit plans
- Released or updated
- not established
Strengths
- Nine explicit strength presets rather than a single opaque slider
- Per-signal conditioning on pose, depth, normals, trajectory and face
- auto_controls derives conditioning from the source automatically
- Up to 64 keyframes to pin guide images at exact frames
- Handles 18-second inputs, longer than most generation models produce
Weaknesses
- Input capped at 18 seconds
- No audio handling documented
- $2.16 per 5 seconds at the top resolution is costly for long footage
- Luma's marketing page claims 20s where the API docs say 18s
Best for
- restyling live-action footage
- motion transfer onto new looks
- previz to final look conversion
Where the sources disagree
Recorded rather than resolved. Picking the tidier number would hide the fact that the vendor's own pages do not agree.
- Luma's llm-info page states "Modify Video V2 for transforming existing footage, up to 20 seconds at 1080p" while the Luma API editing guide states the input must be "18 seconds or shorter" - https://lumalabs.ai/llm-info vs https://docs.agents.lumalabs.ai/guides/videos/editing/index.md
- Aspect ratios listed here are inherited from the Ray3.2 generation parameter set; the editing guide does not restate a ratio list and in practice output follows the source - https://docs.agents.lumalabs.ai/guides/videos/editing/index.md
What to understand before using it
Explainers naming the mechanisms this model's own documentation mentions.
MCP servers that front Luma AI
Servers in the media MCP directory whose own description or homepage names this vendor. Absence of a server here is not evidence that none exists: most publish no tool list at all.
-
Luma
MCP server for Luma Dream Machine AI video generation.
-
Picsart GenAI
Generate and edit images, videos and audio with 150+ AI models.
-
Shortgenius
Generate AI short-form videos and auto-publish them to TikTok, YouTube, Instagram, Facebook and X.
-
Varosity
One key for AI video, voice, music, and image across 40+ frontier models. BYOK zero markup.
-
Vivideo
AI video generation via the Vivideo API: auto or manual mode, avatars, voices, brand kits.
See it from the vendor
Shotcraft hosts no sample output. Judging a model means watching its own showreel with your own eyes, and a clip re-hosted here would be a clip we chose.
Nearby in the directory
Adobe Premiere Pro AI tools (Generative Extend)
Premiere Pro carries a set of Firefly-backed AI features, the best known being Generative Extend, which lengthens a clip's head or tail by generating
Editing and post
DaVinci Resolve 21 AI tools (Neural Engine)
DaVinci Resolve 21 bundles a large set of Neural Engine AI tools inside a full edit, grade, VFX and audio suite. Version 21 adds AI IntelliSearch for
Editing and post
Descript Underlord
AI co-editor inside Descript's transcript-based editor that performs editing actions on command: removing filler words, adding captions, generating sh
Editing and post