luma is the Luma service entry point for video creation, suitable for turning text concepts, still images, and existing generated clips into dynamic assets that can be further developed. Its practical features are first-and-last-frame control and video extension: you can start from a scene description, organize motion around specified images, and continue generating from existing clips. Results include video links, covers, and task information, making it easy to integrate with asset libraries and content production workflows.
Clarify capacity, inputs and outputs, and calling methods before selecting a model.
The above are the creation and calling specifications for the luma entry point and do not represent the full feature scope of all native Ray series versions.
Discover what luma can bring to your work.
In addition to text descriptions, luma can also accept first-frame and last-frame image links, allowing creation to start from specific visuals. The first frame establishes the initial composition, while the last frame conveys the end state; prompts supplement the subject's actions, environmental changes, and camera intent. Ideal for tasks with existing visual drafts that need further exploration of dynamic expression.
When a generated clip is worth keeping but its content has not yet fully developed, you can use extend to continue creating without starting over from text. Submit the clip's video ID or link, then describe the subsequent action; you can also add a last frame to express the ending image of the next segment. This approach makes it easy to explore different directions for shot development in stages.
luma provides controls for aspect ratio, looping, and clarity enhancement, and also supports asynchronous submission and task tracking. Applications can save the task ID first, then query results or receive callbacks, sending completed video links and thumbnails into an asset list. Especially useful for creative tools that need to continue handling other work while generation is in progress.
Start with specific tasks to find where the model can be most effective.
Use a product scene image as the first frame, combined with text descriptions of display actions, environmental atmosphere, and camera direction, to generate dynamic showcase assets; if an ending composition already exists, you can also provide a last frame. The delivered videos and covers can enter a marketing asset candidate library for team selection, then be completed with editing, subtitles, and brand packaging.
Use the starting and ending storyboard images as inputs, describe how characters or objects move and how the scene changes, and use the generated results to observe dynamic expression between the two visual states. When you need to extend an idea, continue creating from the generated clip. Suitable for concept previs and shot proposals, rather than directly replacing precise animation production.
For the same idea, choose landscape, portrait, or square aspect ratios, and enter prompts and reference images suited to each composition to generate assets for different placements. Background animations can use loop controls; the system saves task status, video links, and thumbnails, making it easy for editors to review candidate results and organize subsequent delivery.
Choose based on task complexity, input materials, and expected results.
If you already have static visual assets, or want to continue developing motion from a generated clip, luma's first-and-last-frame and extension workflows are better suited to this type of task. If you only have a text concept, you can create a new video first, then add reference images after clarifying the visual direction. When choosing, focus on visual constraints and segmented creation needs rather than only comparing model names.
luma is a service entry name and is not equivalent to a fixed version such as Ray2, Ray3, or Ray3.2. If a project explicitly requires multiple keyframes, HDR, or a specific video editing method, choose a product that clearly supports those features; if the core needs are text generation, first-and-last-frame guidance, and continuation of existing clips, you can organize creation around these luma workflows.
From a small-scale task to formal integration.
Define the objective, required inputs, and output requirements, using real business examples as a starting point.
Open the trial page, confirm the parameters supported by this entry, then submit a small-scale task to review the results.
Keep the complete model ID, use the request format specified in the documentation, and confirm billing rules on the Pricing page.
Before formal use, understand the output quality and capability scope.
Answers to common questions about using luma.
That is not the naming relationship. luma is the name of a video service entry point and cannot be directly equated with a specific Ray version. When using it, you can plan tasks around text generation, first and last frame control, and video extension; if you need exclusive features of a particular Ray version, choose a product that explicitly supports those features.
You can provide image links through start_image_url and end_image_url, then use prompt to describe the actions and changes you want to happen in between. You can also start creating with only a first frame. Two images are suitable for expressing the starting and ending states of a shot, but the intermediate process is still generated rather than specified frame by frame.
Set action to extend, provide the video_id or video_url of an already generated video, and enter a prompt for the subsequent content. If you want the continuation to end on a specified image, you can also add end_image_url. Save the new video information returned by each generation to facilitate continued creation and asset management.
loop is used to request a looping video, while enhancement is used to control clarity enhancement; they serve playback mode and visual processing respectively and should not be confused as a single feature. When creating dynamic backgrounds, you can try looping and check the transition effect; enabling enhancement also does not guarantee a specific fixed output resolution.
No. You can set async=true to first obtain a task_id, then retrieve status and results through task queries; you can also use callback_url to receive completion notifications. The task ID is used to associate this creation, while the video ID identifies the generated asset. They should be saved separately to avoid mixing them up in subsequent extensions.
Model information · Updated: 2026-10-01. For call parameters and billing rules, see the API and pricing sections.
Start with a clear goal and assess from real results whether it suits your work.
Pricing
$0.11· per generation
You pay$13,584for 142,800 credits · $0.0951 / credits46% OFF
You're billed in credits. Pick a top-up tier to see its per-credit rate — the larger the top-up, the cheaper each credit.
This is only a basic call example. See the full docs for more parameters and advanced usage.
View Docs