Create Videos
Make AI videos from a text prompt, an image (image to video) or reference media. Animate a photo or product shot, or make clips for TikTok, Reels and YouTube Shorts.
The Video app makes short clips from a text prompt, from still frames, or from images, video and audio you upload. To turn an image into a video, use Frames.
Who can use it
Video is a premium app. You can use it if you have a paid plan or bought credits. On a free plan with no bought credits, you see Upgrade to unlock premium apps & models. See Available Plans and Buying Extra Credits.
Open the Video app
Select Video in the sidebar, or pick Video in the type picker on Home. Both open Generate Video. From Home, Ideogram copies your prompt in and starts the run; if you attached images, it uses References mode.
On Home, the quick action Generate video opens the app in Text mode. Edit video opens Seedance 2.5 in References mode.
Make a video
Select Video in the sidebar. Generate Video opens.
Optional: open the Model picker and choose a model. Auto picks one for you. See Models.
Choose an input mode: Text, Frames or References. Every model starts in Text mode. See Input modes.
In Frames mode, add a Start frame. In References mode, add media under Add media. See Frames and References.
Describe the video in the Prompt box, up to 2,000 characters. In References mode, describe what you want to change. See Prompt.
Optional: change Duration, Aspect ratio, Resolution or Generate audio. See Settings.
Click Generate.
Find the finished video in the My assets tab on the right. See Results and download.
Models
Open the Model picker. Auto sits at the top, then Recent, then All models. You can search with "Search all models..." or filter by provider.
The Video app offers:
Auto: picks a model for you
Seedance 2.5
Seedance 2.0
Kling v3 Standard
MiniMax H3
The model list can change over time.
Input modes
Choose how you start with the Text, Frames and References control. Every model starts in Text mode.
Text
Makes a video from your prompt alone.
All
Frames
Animates from a Start frame, with an optional End frame.
All
References
Uses images, video clips and audio you add as the basis for the video, then applies the changes you describe.
Generate Video, Seedance 2.5, Seedance 2.0, MiniMax H3
Frames
Frames is image to video: your image becomes the first frame, and the model animates from there.
Turn an image into a video
Select Frames.
Click Start frame and choose an image.
Optional: add an End frame if you want the clip to finish on a set image.
You can also paste an image in. One button swaps the two frames. The frame sets the shape of the video, so Aspect ratio locks to "Auto".
Animate a product photo
To bring a product shot to life, use Frames:
Select Frames.
Add your product photo as the Start frame. The video opens on that exact image.
Write how it should move, for example
the sneaker slowly rotates 360°, camera locked, white background.Turn off Generate audio if you want a silent clip.
Click Generate.
Use References instead when you want the product placed in a new scene rather than animated as it is.
References
Add reference media
Select References.
Select the plus under Add media.
Select References, then the plus under Add media. Upload files, choose from your library, or select Record audio to record with your microphone.
If the browser blocks the microphone, allow access in your browser settings.
Seedance 2.0 and 2.5 need at least one image or video. MiniMax H3 needs one only when you add audio.
You can refer to your references in the prompt. The app rewrites those mentions into the notation each model expects.
Images
9
9
9
Video clips
3
10
3
Audio clips
3
10
3
Total references
12
50
12
Total video length
15 s
30 s
15 s
Total audio length
15 s
30 s
15 s
Audio and video together
No limit
No limit
15 s
File limits
Frames and reference images
JPEG, PNG, WebP
50 MB
Reference video
MP4
50 MB, at least 2 s long
Reference audio
MP3, WAV
15 MB, at least 2 s long
Prompt
Describe the video in the Prompt box. Prompts can be up to 2,000 characters. In References mode, describe what you want to change.
Video prompts work best when they describe motion and the camera, not just the scene:
The bottle slowly rotates 360°, camera locked, soft studio lightSlow push-in on the coffee cup as steam rises, shallow depth of field
Say what stays still as well as what moves. "Camera locked" or "static shot" stops the camera drifting.
Settings
Settings differ by model. Each model defaults to 10 seconds and 16:9. With Generate Video, the panel shows the settings that apply to your run.
Duration
Auto or 4–15 s
Auto or 4–30 s
3–15 s
5–15 s
Aspect ratio
Auto, 21:9, 16:9, 9:16, 1:1, 4:3, 3:4
Auto, 21:9, 16:9, 9:16, 1:1, 4:3, 3:4
16:9, 9:16, 1:1
21:9, 16:9, 9:16, 1:1, 4:3, 3:4
Resolution
480p, 720p (default 720p)
480p, 720p, 1080p (default 720p)
Not shown
480p, 768p, 2K, 4K (default 2K)
Generate audio
On by default
On by default
On by default
Not shown
Advanced settings
None
None
Shot planning (Custom or Intelligent), Prompt adherence (0–1, default 0.5), Negative prompt (what the video should avoid, up to 2,000 characters)
Prompt expansion (Off, Fast, Balanced or Quality; default Balanced), References mode only
Cost and generating
Cost depends on the model, duration and resolution. To see video prices, go to Settings → Usage costs. See How credits work.
To save the result straight into a collection, click Add to Collection next to Generate.
Results and download
Finished videos appear in the My assets tab on the right of the app. They are private by default. Open a video to see its length and resolution, like it, add it to a collection or delete it.
Download a video as one MP4 file:
Open the video.
Click Download.
If you see "Video is not available for download yet", wait for the video to finish.
To load a generated video into MiniMax H3 as a reference:
Open the video.
Select Edit.
You can't upscale a video, remove its background or use it as a reference image.
Last updated