Vertical clips
The 9:16 format for vertical feeds and shorts, where a horizontal frame gets cropped at the edges.
CAPABILITY
AI video generation from text and photos
Describe the scene and you get a finished MP4: horizontal, vertical or square, and on some models with an audio track already in it. You can also start from a picture: attach a frame and the model puts it in motion. Some models take a final frame as well, and the movement is built between the two.
- up to 10 s
- length of one clip
- 3
- aspect ratios
- MP4
- the file you get
How it works
-
Describe the scene or attach a frame
A text description is required in both modes: it sets what happens in the frame. A starting image is needed only when you want to put a specific picture in motion.
-
Set the parameters
Aspect ratio, resolution and duration — the available options come from the settings of the model you picked. The sound toggle and the negative prompt sit in the same place, on models that support them.
-
Start it and leave the page
Rendering runs on the server. Progress is shown as a percentage, and the queue panel stays visible over the interface when you move to another section, so the job finishes without you.
-
Collect the clip
The finished file appears in your personal gallery, usually with a cover taken from the first frame. From there you can watch it, download it, or, in Telegram, get it as a message from the bot.
What it does
- Input
- A text description of the clip; optionally a starting frame, and on some models a final frame as well
- Modes
- Text to video and frame to video — one clip per run
- Aspect ratios
- 16:9, 9:16, 1:1; the model decides which of them are available
- Duration
- 5, 8 or 10 seconds, depending on the model
- Resolution
- The model you pick decides which sizes are available
- Frame rate
- 24 frames per second by default
- Sound
- Models with an audio track have a toggle: the clip arrives with the sound already in it
- Negative prompt
- A description of what should stay out of the frame; supported by some models
- Where to find it
- The dedicated generation screen, the video section of the dashboard, and the Studio
What people use it for
Put a photo in motion
Attach a photo or an illustration and the model adds movement. Some models take a final frame as well, and then the clip is built as a transition from the first to the second.
Intros and previews
Horizontal 16:9 of 5 to 10 seconds — for the top of an article or a short cutaway.
A clip with sound
Models with an audio track return video with the sound already in it, with no separate voice-over step.
Models for this task
Full catalogVideo
ByteDance Seedance 2.0 Mini
Compact Seedance 2.0 variant for faster, lighter multimodal video generation — T2V, I2V, V2V, synced audio
$0.04 / sec
Video
Gemini Omni Flash
Google multimodal video model: 10-second 720p clips with native audio, photo references and multi-turn video editing
$0.12 / sec
Video
Grok Imagine Video
AI video generation model
$0.06 / sec
Video
HeyGen Avatar IV
AI video generation model
$0.12 / sec
Video
Kling VIDEO 3.0 Pro
AI video generation model
$0.13 / sec
Video
Kling VIDEO 3.0 Standard
AI video generation model
$0.1 / sec
Video
Kling VIDEO O3 Pro
AI video generation model
$0.13 / sec
Video
Kling VIDEO O3 Standard
AI video generation model
$0.1 / sec
Video
Seedance 2.0
ByteDance multimodal video model with synced audio: multi-shot clips up to 15s from text, images, video and audio
$0.08 / sec
Video
SkyReels V4
AI video generation model
$0.13 / sec
Video
Sora 2
AI video generation model
$0.12 / sec
Video
Sora 2 Pro
AI video generation model
$0.36 / sec
What else it does
The things people usually discover only once they start working.
- In frame-to-video mode the clip inherits the geometry of the attached frame; on some models the dimensions are set alongside it
- One run takes a single source as its basis — a starting frame or a reference clip
- Generation from a reference clip is available on the dedicated generation screen and in the By reference tab
- A starting frame is a file of up to 10 MB
Frequently asked questions
How long is the clip?
5, 8 or 10 seconds — the set of options depends on the model. The maximum for a single run is 10 seconds.
Can I animate a photo?
Yes. Attach the photo as a starting frame and describe what should happen in the scene: the text description is needed in this mode too. Some models take a final frame as well, and then the movement is built from the first to the second.
Which models return a clip with sound?
Models with an audio track: their settings include a sound toggle, and the finished clip arrives with the sound already in it, with no separate voice-over step.
When am I charged for a clip?
You are charged after a successful result. Before the run starts, the service checks the estimated cost against your balance and, if a top-up is needed, shows the exact amount right away.
Do I have to stay on the page until the render finishes?
No. The job runs on the server, so you can move to another section — the queue panel stays on screen and the finished clip appears in the gallery.
What do I get at the end?
An MP4 file, 24 frames per second by default. You can watch it in the browser, download it, or get it as a message from the bot in Telegram.
Related capabilities
Try it on your own task
A starting balance is already on your account, no card required. You pay only for what you generate.