Start with
words.
No reference image? Describe the subject, action and camera movement. AI Text to Video is a useful starting point for a scene you want to create from scratch.
AI Video Generator
Create short videos from words or reference images.
Choose your
starting point.
Start with a written scene, animate one image, or add several visual references. This AI video maker supports 3–12 second clips; choose the input that fits what you already have, then describe what should happen.
Start with a promptStart with
Start with
one image.
Use a photo or illustration and explain how it should move. With one reference, the image determines the video’s aspect ratio. Explore AI Image to Video for this approach.
Start with
references.
At 480p or 720p, add two to seven images and explain which appearance, setting or style each should guide. At 1080p, use at most one reference image. With multiple references, you can choose a supported aspect ratio, but references are not separate first-frame and last-frame controls.
Small subjects.
Clear actions.
For a miniature scene like the chefs shown here, describe one main action before adding background activity. Say who moves, what they interact with and how quickly. After generation, check hands, utensils and contact with nearby objects—not just the overall look.
A visual style.
A sense of motion.
A painterly scene needs more than a style label. Describe what changes over the shot, whether the camera stays still and which textures should remain visible. Review the clip for flicker, drifting details and abrupt changes; visual references guide a result, rather than guarantee an exact match.
Moving subjects.
A following camera.
For a chase-style idea, separate the subject’s movement from the camera’s movement: what travels through the scene, in which direction, and whether the camera follows or stays fixed. Inspect the subject’s shape and position throughout the clip. Change one prompt detail at a time when refining a result.
Describe the scene.
Direct the motion.
A useful video prompt explains what changes over time, not only what a still frame looks like. Start with one main action and one camera movement, then add the details that support them. The example below teaches a prompt structure; it is not the original generation record for a displayed clip.
Start with a promptDefine the subject.
Name the main subject, its appearance and its position. Keep the important details close together: “An orange drink can stands upright on a wet stone surface.” If using references, explain which image should guide the subject and which should guide the setting.
Describe the action.
Say what moves, in which direction and at what speed. Distinguish the main action from background movement: “A small splash rises behind the can and falls naturally.” For a short clip, one readable action is easier to review than a sequence of unrelated events.
Direct the camera.
Specify the viewpoint and whether the camera is fixed, pushing in, tracking or orbiting. For example: “The camera slowly pushes in while the can stays centered.” These are written directions, not separate camera controls, and the result may need another iteration.
Set the surroundings.
Add the environment, light and secondary motion without competing with the main action: “Cool side lighting catches the droplets, with a simple dark background.” Mention moving steam, water or wind only when it serves the scene; avoid a list of contradictory visual styles.
Keep the shot focused.
Keep the timing manageable and review continuity. Combined teaching prompt: “An orange drink can stands upright on a wet stone surface. A small splash rises behind it and falls naturally. The camera slowly pushes in while the can stays centered. Cool side lighting catches the droplets, with a simple dark background. Keep the shot continuous, without scene changes.”
Plan a short clip.
Give it a purpose.
These still-image concepts suggest directions to explore, not finished video results. Decide what the viewer should notice, then choose a frame shape and a single action that suit the intended use.
Product concepts.
Focus on one detail.
For a drink concept, try a brief splash or a slow camera push rather than several competing effects. Leave space for text you plan to add later, and check packaging, lettering and product shape before using the clip in a campaign.
Social stories.
Follow the action.
For a skateboard scene, make the direction of travel and camera position explicit. A 9:16 frame can suit a vertical post, while 16:9 gives a wider view. With a single reference image, prepare its composition first because it determines the ratio.
Imagined worlds.
Make scale clear.
For a tiny astronaut beside a coffee cup, establish the size relationship and choose a small, readable movement. A gentle step or drifting steam can carry a short scene. Check that the scale and object positions remain coherent from beginning to end.
How to create an AI video.
Use this page to start your idea, then continue in the workspace to sign in, confirm your inputs and settings, and submit the generation. On mobile, add references and set your options in the workspace.
Choose your starting point
Write a scene from scratch or add up to seven reference images at 480p or 720p, or one at 1080p. Explain what each image should guide. If you need to create a starting visual first, use the AI Image Generator, then bring it into your video workflow.
Describe and set your clip
Describe the subject, action and camera. In the workspace, select 3–12 seconds and a 480p, 720p or 1080p output setting, then review the displayed credit cost. Grok Imagine 1.5 generates at the selected setting without a separate upscaling step; higher resolution does not guarantee corrected motion or details.
Review and download
Generate in the workspace, then review the full clip for movement, consistency and unwanted changes. Download the completed MP4 within 72 hours of completion. Save it locally before that window ends; platform storage is not permanent.
Frequently
asked
questions
Can I create videos from text or images?
Yes. Start with a written prompt alone, one image, or two to seven reference images at 480p or 720p. At 1080p, use at most one reference image. A written prompt is required when you submit in the workspace. One image determines the aspect ratio; with text alone or multiple references, you can choose a supported ratio. The current workflow accepts images, not video uploads, and does not offer dedicated first-frame, last-frame, clip-extension or video-editing controls.
Which reference images can I upload?
Upload JPEG, PNG or WebP files, up to 20 MiB per image: at most seven images at 480p or 720p, or one at 1080p. Use material you have permission to upload. Choose clear references and describe what each should guide; multiple images do not guarantee exact subject identity or visual consistency. Your video prompt can contain up to 4,096 characters.
What duration, aspect ratios and resolutions can I choose?
Select a duration from 3 to 12 seconds in whole-second increments. Available ratios are 1:1, 16:9, 9:16, 3:2 and 2:3, with no automatic option; single-image mode uses the image’s ratio. Output settings are 480p, 720p and 1080p. Grok Imagine 1.5 generates directly at the selected setting without a separate upscaling step. The default is 5 seconds at 480p, with a 16:9 ratio when no single image determines it. Actual pixel dimensions depend on the generated output.
Is there a free allowance, and how many credits does a video cost?
New accounts receive 50 credits valid for 30 days. If unused, that allowance covers one default 5-second video at 480p for 40 credits, with 10 credits remaining; it is not a daily or unlimited offer. At 480p, 3, 5 and 12 seconds cost 20, 40 and 90 credits. At 720p, they cost 40, 70 and 170; at 1080p, 80, 130 and 300. Cost depends on duration and output setting, not the number of references. Check the workspace’s displayed cost before generating.
How do I download a video, and how long is it saved?
Completed videos can be viewed and downloaded as MP4 files in the workspace. They are stored for 72 hours after completion, not after you first submit the task. Download and keep a local copy before expiry. Generation time varies; there is no guaranteed completion time. Review the entire video before sharing it or using it in a project.
Can I use my videos commercially, and are they private?
Free and paid generated outputs may be used commercially under our Terms of Service. You remain responsible for permissions covering uploaded material and the rights involved in your intended use; commercial permission is not a guarantee of exclusive copyright. Uploads and outputs are private by default, and prompts, uploads and outputs are not used to train AI models under our platform policy. See the Privacy Policy for details.