Gameplay & ideation
Visualize a mechanic before you build it: feel out pace, camera, and atmosphere in seconds, straight from a generated screenshot.
Turn a text prompt, an image or a set of reference images into a 5 to 15 second video with a soundtrack.
Ludo's Video Generator, shown making the videos of a witch game in each of its modes:
An AI video generator turns a text prompt, a still image or reference images into a short, animated MP4 clip. Ludo's Video Generator uses the Griffin HD model to make 768p clips from 5 to 15 seconds with a soundtrack. You can set a custom end frame, compose scenes from up to five reference images, edit a finished video with a prompt, and upscale it 2x.
Last updated November 2026
Visualize a mechanic before you build it: feel out pace, camera, and atmosphere in seconds, straight from a generated screenshot.
Cinematic intros, story cutscenes, and high-energy trailer beats, straight from a single image or prompt.
Bring still character art to life: subtle breathing, hair, glow, eye movement. Perfect for dialogue scenes.
A picture or a finished clip is a starting point.

Give it a picture you already have and describe what moves: the character, the camera, the weather. Refine the starting frame in the image editor first if you like.
Describe a change to a clip you made: the season, the time of day, an outfit. The motion stays the same, and up to 5 reference images can guide the edit.
Doubles the resolution of a square 768x768 clip, for store pages and trailers.
Cinematic video in any style at 768p: 768x768 square, 1344x768 landscape or 768x1344 portrait. Pick a length from 5 to 15 seconds, in whole seconds. Every clip comes with a soundtrack.
Clips download as standard MP4 files. A custom final frame makes the clip travel from your first picture to your last, and From References builds a scene from up to five images in 16:9, 9:16, 1:1, 4:3, 3:4 or 21:9.
Every generator in your terminal, your AI agent, your own code and the Unity editor. One account, one set of credits.
Explore developer toolsnpm install -g @ludoai/cliMCP serverAny MCP client, including assistants that can't run commands.https://mcp.ludo.ai/mcpREST APIYour own app, game server or pipeline.https://api.ludo.ai/api/Unity pluginGenerate and import assets inside the Unity editor.The short answers. For the full detail on every mode and setting, see the documentation below.
Four modes. Text to Video generates a first frame from your prompt and animates it, Image to Video animates a source image (uploaded or generated in Ludo) guided by a motion prompt, From References composes a scene from 1 to 5 reference images and a prompt, and Edit Video changes a clip you already generated with a text prompt. In Image to Video you can also set a custom final frame so the clip is interpolated between your start and end images.
Griffin HD makes clips from 5 to 15 seconds, in whole-second steps. You pick the exact duration before generating.
Griffin HD makes cinematic video in any style at 768p: 768x768 square, 1344x768 landscape or 768x1344 portrait. It runs every mode, from text and image to video to From References and video editing, and it generates a soundtrack with every clip.
Yes. Griffin HD generates a soundtrack together with every video, so you do not need a separate audio step. If you want dedicated game audio, Ludo's Audio Generator makes sound effects, ambiance loops and music.
Yes. Edit Video takes a clip you generated, a text prompt and up to 5 optional reference images, and returns a new clip that follows the source length, up to the Max Duration you set. You can also upscale a clip 2x when both of its sides are under 960 px, which means square 768x768 clips; landscape and portrait clips are already too large to upscale.
Videos are standard MP4 files you can download and drop into trailers, social posts or prototypes. From References also lets you choose the shape of the output: default, 16:9, 9:16, 1:1, 4:3, 3:4 or 21:9.
The generator is built for game creators: visualize gameplay concepts, animate characters and scenes, produce marketing content and trailers, or prototype motion ideas from a single image or prompt. You own the videos you generate with Ludo and can use them however you like, including in commercial games.
Yes. The Ludo public API, MCP server and CLI expose createVideo (from an image), createVideoFromReferences, editVideo and upscaleVideo. API and MCP results are kept for 7 days, and edit and upscale work on videos you generated in that window.
The Video Generator is a powerful AI tool within the Ludo.ai platform designed to create short, animated game videos based on images and text descriptions. This feature allows you to quickly visualize motion, animate characters and assets, or create dynamic gameplay mockups.
With the Video Generator, you can:
To get started:
The Video Generator offers four modes. Pick one in the Generation Mode menu at the top left of the interface; each entry shows a short example of what the mode does.
This mode is ideal when you are starting from scratch with just an idea. You provide a single, detailed text prompt that describes both the visual content of the video's first frame and the motion you want to see. The tool then generates an initial static image based on your description, which you can then approve to generate the final animated video.
Use this mode when you already have a specific image you want to animate. This could be an image you've previously generated on Ludo.ai, or one you upload. You provide the starting image, an optional final frame, and a separate text prompt to describe the desired motion. The tool will then generate a video that animates your provided image.
Use this mode when you have the cast but not the shot: up to five reference images of characters, objects or scenes, plus a prompt describing the video you want them in. Unlike Image to Video, the references are not the first frame — the model composes a new shot with those subjects in it. See From References.
Use this mode to change an existing video with a text instruction — swap an item, recolor, add an effect, restyle — while keeping its motion and timing. You can add reference images to show exactly what you mean. See Editing a Video.
The Image to Video, From References and Edit Video modes all run on Griffin. It renders at 768p (768×768 square, 1344×768 landscape, 768×1344 portrait), always with a synchronized audio track, and takes durations from 5 to 15 seconds. It follows multi-action prompts, camera directions and dialogue, and costs the same per second in every mode.
Griffin follows the aspect ratio of your first frame (Image to Video) or the Aspect Ratio you choose (From References).
Image to Video also offers Ignis, the budget model: 720p with audio, durations from 3 to 10 seconds, at 1 credit per second against Griffin's 2. It is best for simple scenes and quick tests.
Open the Model dropdown to compare the models side by side. Each card shows the model's name, the duration range it supports (for example, 5 to 15s), a short description of what it is good at, and the cost in credits per second, plus a minimum charge where applicable.
Older models — Blitz, Anvil, Eagle and Eagle with Audio — have been retired to a collapsed Legacy models group at the bottom of the dropdown. They exist only so that projects already built on them can keep generating matching videos, and should not be used for new work. New accounts don't see the legacy group at all — if the dropdown only lists Griffin and Ignis, nothing is missing.
After you pick a model, the Duration control shows only the durations that model supports. Griffin takes whole seconds from 5 to 15 and Ignis from 3 to 10, so the slider snaps to those.
The credit cost shown next to the duration updates live based on your model + duration choice, so you always know what a generation will cost before clicking Generate Video. If you switch models and your current duration is not supported by the new model, Ludo.ai automatically selects the closest valid duration.
The quality of your prompt is key to getting great results. What a good prompt looks like depends on the mode: in Text to Video the prompt paints the first frame, in the other modes it directs the clip.
Here the prompt is used first to generate a still image, so it needs the visuals as well as the action. A good structure includes:
a hero sprite).in a top-down pixel art perspective).executes a rapid dash attack through a cluster of blocky enemies, triggering bright particle effects).a health bar depletes).The image already fixes the look, so don't re-describe it. Write what happens, and use the whole duration. Ludo turns your instruction into a paced, shot-by-shot description for the video model, so a longer clip can carry a real sequence rather than one motion stretched thin:
the shopkeeper leans in and whispers: "Not for sale." Lines are reproduced verbatim and lip-synced; nothing is invented that you didn't write.What to avoid: mood words and adjectives ("epic", "beautiful", "cinematic") do nothing; concrete nouns and verbs do the work. Don't write the duration or timestamps yourself — the pacing is derived from the duration you selected.
Sprites and game assets. The asset's type (a sprite, a background, a VFX element, an icon) tells Ludo what a typical animation for it looks like — an idle loop for a background, a static camera for a sprite — and that is used as a default. Your instruction always wins: ask a background to explode and it explodes.
Simple prompts are excellent for brainstorming and discovering unexpected ideas. Don't be afraid to start with a broad concept and let the AI fill in the details.
Use detailed prompts when you have a specific vision in mind, and use simple prompts when you want the AI to be a creative partner in the ideation process.
The "Text to Video" mode uses a two-step workflow so you can review and refine the starting visuals before committing credits to a full video render.
Step 1: Generate the First Frame
Step 2: Turn the Frame into a Video
The "Image to Video" mode is a direct way to animate an existing visual. It is also where you land after clicking "Use as First Frame" on a frame generated in the Text to Video flow.
Before generating the video, you can precisely control the starting and ending visuals using the integrated Image Editor. After uploading or selecting an image for the First Frame or Final Frame, click the "Open In Editor" button on its card. This allows you to:
Editing your keyframes ensures the animation begins and ends with the exact visuals you want. Once you save your edits, you will return to the Video Generator to describe the motion and create the video.
To ensure the best results and adhere to platform policies, please follow these guidelines for your input images:
The From References mode generates a new video featuring the subjects in your reference images, rather than animating one image. Give it your character sheet, an enemy design and a background, describe the shot, and you get a clip with all of them in it.
When to use it
How to Use:
Tips
The Edit Video mode changes what's in an existing video while keeping its motion, timing and framing. Instead of re-generating and hoping for the same movement, you edit the clip you already have.
What you can do with a text instruction
Reference images (optional, up to 5): show the AI exactly what to bring in — your game's actual axe, a target art style, an outfit — instead of describing it.
How to Use:
Edits run on Griffin. Every edit re-renders the whole clip, so colours and detail drift a little with each pass — if an edit didn't work, go back to the original rather than stacking another edit on top.
Upscale 2x on a video card renders the same video at twice the resolution, keeping its frame rate, duration and soundtrack. It needs a video under 960 pixels on both sides: a square Griffin video (768×768) qualifies, and so do older 480p videos. Landscape and portrait Griffin videos are too large, so their cards don't offer it.
The Video Generator uses a focused set of options to steer the AI. The options available depend on which mode you are in.
These options shape the first frame generated from your prompt:
In Image to Video, the first frame is already chosen, so the options focus on the animation itself:
Once a video is generated, it appears in the gallery with several options:
.mp4) to your computer.Between Step 1 and Step 2 of the Text to Video flow, the gallery shows the generated first frames as image-only thumbnails. These have their own actions:
The Video Generator is a versatile tool that can accelerate various stages of the game creation process, from initial design to marketing.
If you encounter issues while using the Video Generator, consider these steps:
If problems persist, please contact Ludo.ai support or ask on our Discord server for assistance.