# Luma Video Canonical page: https://fuser.studio/models/luma-video Luma Video in Fuser generates video from text or keyframe images with Luma's Dream Machine model. Interpolate between stills and chain edits on one canvas. ## Specs | Spec | Value | | --- | --- | | Inputs | Prompt (text), Image Prompt (image), End Image Prompt (image) | | Output | video | | Aspect ratios | 16:9, 9:16, 4:3, 3:4, 21:9, 9:21 | | Cost | 689 credits per run | Transform prompts and still keyframes into cinema-grade motion shots with physical camera control Creator: Luma AI Also known as: Dream Machine, Ray 2, Ray 3 [Direct video shots](https://app.fuser.studio) ## How Luma Video works Go from a single concept or still image to a fully realized cinematic clip in three directed moves. ### Anchor your visual frame Input a descriptive text prompt or upload a reference keyframe to preserve character styling and lighting. ### Direct camera and kinetics Specify camera techniques like dolly zooms or tracking shots alongside temporal light and subject movement. ### Format and render Select your aspect ratio across widescreen, vertical, or standard crops, and toggle seamless looping. ## What Luma Video is good at From physics-based subject kinetics to flexible aspect ratios, control every dimension of cinematic video generation. ### Keyframe-Anchored Animation Bring high-fidelity still images to life using image-to-video mode, preserving original styling, lighting, and textures while directing natural temporal motion. ### Dynamic Camera Tracking Generate entirely from scratch with text-to-video to hallucinate sweeping FPV drone sweeps, steady tracking shots, and cinematic crane maneuvers. ### Atmospheric Looping Engine Create continuous 5-second video loops that smoothly blend start and end frames for ambient visual installations, flowing fluids, and atmospheric backgrounds. ### Multi-Aspect Framing Compose scenes natively across widescreen 21:9, standard 16:9, or vertical 9:16 mobile viewports without post-crop distortion or loss of framing intent. ## Made with Luma Video Explore how physics-aware motion, authentic film grain, and dynamic camera paths come together across styles. ### High-speed automotive motion captured in wet atmospheric light ### Surreal musical instrument animation with dynamic physics ### Slow-motion fluid dynamics and packaging commercial frame ### Acrobatic kinetics and camera rotation for vertical short-form ### Editorial fashion motion with organic material textures ## What people build with Luma Video See how directors, designers, and visual artists turn static concepts into dynamic film sequences. ### Fashion Editorial & Lookbooks Convert static concept photography and lookbook stills into sweeping runway and editorial video sequences with subtle garment motion. ### Cinematic Worldbuilding & Previs Generate establishing shots, atmospheric landscapes, and cinematic transitions directly from script descriptions for pre-visualization. ### Product Design & Automotive Spots Showcase physical objects in motion with realistic surface reflections, mechanical articulation, and dynamic camera angles. ### Architectural Walkthroughs & Ambiance Bring interior visualizations and architectural models to life with realistic natural light progression and seamless environmental loops. ### Brand Commercials & Music Videos Produce surreal visual metaphors and high-impact hero shots with striking color tension and physical motion dynamics. ## Frequently Asked Questions ### Is Luma Video the same as Dream Machine? Yes, Luma Video is the video generation model family developed by Luma AI, commonly known as Dream Machine or its underlying Ray architectures. The system translates text descriptions and static images into photorealistic video sequences with native understanding of real-world physics and camera movements. ### What is the difference between text-to-video and image-to-video workflows? Text-to-video generates entirely new visual scenes and camera paths from a written description, offering creative freedom at the expense of strict compositional consistency. Image-to-video takes an existing still image as an anchor, animating motion and camera shifts while strictly preserving the subject's styling, textures, and lighting. ### Which workflow should I use for character and composition consistency? You should use the image-to-video workflow whenever character consistency, brand fidelity, or specific scene composition is essential. Starting from a high-fidelity keyframe allows the model to focus purely on temporal motion, avoiding the visual drift that can occur when generating characters from scratch. ### What visual styles and camera techniques work best with the model? The model excels at cinematic cinematography, including tracking shots, dolly zooms, crane sweeps, and FPV drone paths. It naturally produces authentic film grain, soft exposure gradients, and realistic depth of field, working exceptionally well on vehicles, landscapes, environmental fluids, and subtle character portraits. ### What are the limitations of Luma Video and what should I avoid? Luma Video is not designed for rendering legible typography, brand logos, or high-speed chaotic crowd scenes. In addition, generations are strictly silent without native audio, and complex anatomical movements during rapid action sequences can produce morphing artifacts. ### Who developed Luma Video and what versions exist? Luma Video was created by Luma AI, first launching in June 2024 as Dream Machine and evolving through subsequent iterations including the Ray 2 and Ray 3 model generations. It is built on a scalable multimodal transformer architecture trained directly on extensive video sequences. ## Try Luma Video on Fuser Transform prompts and still keyframes into cinema-grade motion shots with physical camera control [Direct video shots](https://app.fuser.studio) ## Related models - [Luma Ray](https://fuser.studio/models/luma-ray.md) - [MiniMax H3 Recast](https://fuser.studio/models/minimax-h3-recast.md) - [Kling O3 Edit](https://fuser.studio/models/kling-o3-edit.md) - [Kling O3](https://fuser.studio/models/kling-o3.md) - [Kling 3.0 Video](https://fuser.studio/models/kling-3-0-video.md) - [Seedance 2](https://fuser.studio/models/seedance-2.md) Node reference: [Luma Video docs](https://docs.fuser.studio/docs/nodes/video/luma-video.md)