# AI video generator

Make AI-generated videos from any photo, video or prompt with ready-made Spaces, then build your own workflow in Spacelab so every result stays consistent.

Last updated: 2026-11-06

**Key facts**

- Any photo, video or prompt goes in and an AI-generated video comes out.
- A Space is a ready-made workflow for one job: pick one, add your input and get the output.
- Build your own workflow once in Spacelab and run it again and again, so every output stays consistent.

Web page: https://sloosh.ai/docs/ai-video-generator

Got an idea? Start with a prompt, any photo or any video and Sloosh turns it into an AI-generated video. Anyone can do it. The quickest way is a Space: pick one for the job, add your input and get the output.

## Make a video with a Space

- [Shadow Clone](https://sloosh.ai/spaces/shadow-clone): upload a reference clip of up to 30 seconds and your images, and get a new video with the same moves.
- [Talking Clip](https://sloosh.ai/spaces/talking-clip): upload a clip of a person facing the camera, type the words, and get it back with a natural voice and synced lips.
- [Extend Video](https://sloosh.ai/spaces/extend-video): upload a video up to 10 seconds, optionally say what happens next, and the scene continues past its last frame.
- [Add Captions](https://sloosh.ai/spaces/add-captions): upload a talking video and get it back with word-accurate captions burned in.
- [Video Upscale](https://sloosh.ai/spaces/video-upscale): upload a clip and get it back at twice the resolution, up to 4K, with the same length and frame rate.

[Spaces](https://sloosh.ai/docs/spaces.md) explains how they work.

## Make a video from a prompt or photo

1. Open [Video](https://sloosh.ai/studio/video) in Studio.
2. Describe the shot as a sequence: subject, place, action and camera. [Video prompts](https://sloosh.ai/docs/video-prompts.md) has examples.
3. Optional: use the + menu to Upload frames (a start or end frame) or Upload references (a product, a person, a place).
4. Pick a model in the chip. Length, resolution and the other settings come from that model.
5. Press Generate. The button shows the cost first.

Some models take start and end frames or references, not both, and Studio closes the other option once you add one. [References](https://sloosh.ai/docs/references.md) has the detail.

## Build your own workflow

When you have a process that works, build it once in [Spacelab](https://sloosh.ai/docs/spacelab.md) and run it again and again, so every output has the same look. Anyone can build one.

## Check before you publish

AI-generated results can get label text and fine details wrong. Check both before you publish.

## Model details

Each model has its own page. This is the short version, by job.

| Job | Models to try | What to know |
| --- | --- | --- |
| A long clip | [Seedance 2.5](https://sloosh.ai/docs/seedance-2-5.md), [Kling 4.0](https://sloosh.ai/models/kling-4.0), [Wan 3](https://sloosh.ai/docs/wan-3.md), [Wan 3 Prime](https://sloosh.ai/models/wan-3.0-prime) | All four reach 30 seconds. Seedance 2.5 takes any length from 4 to 30 and follows timed beats. |
| Many references | [Seedance 2.5](https://sloosh.ai/docs/seedance-2-5.md), [Seedance 2.0](https://sloosh.ai/docs/seedance-2-0.md), [Grok Imagine 1.5](https://sloosh.ai/models/grok-imagine-1.5), [Kling 4.0](https://sloosh.ai/models/kling-4.0), [MiniMax H3 Reference](https://sloosh.ai/docs/minimax-h3-reference.md), [Kling V3 Omni](https://sloosh.ai/docs/kling-v3-omni.md), [Grok Imagine](https://sloosh.ai/docs/grok-imagine.md) | Seedance 2.5 takes the most: 30 images and 10 videos. Grok Imagine 1.5 takes 14 images and Kling 4.0 10. Each model's page has its limit. |
| Animate one photo | [Kling 3.0](https://sloosh.ai/docs/kling-3-0.md), [Kling 2.5](https://sloosh.ai/docs/kling-2-5.md) | Both need a start image. Kling 2.5 makes a silent 5 or 10 second clip. |
| Control the first and last frame | [Gemini Omni 1.1 Flash](https://sloosh.ai/docs/gemini-omni-1-1-flash.md), [Gemini Omni 1.2 Flash](https://sloosh.ai/docs/gemini-omni-1-2-flash.md), [Kling V3 Omni](https://sloosh.ai/docs/kling-v3-omni.md), [MiniMax H3](https://sloosh.ai/docs/minimax-h3.md), [Wan 3](https://sloosh.ai/docs/wan-3.md) | Set a start frame, or a start and end frame. |
| Up to 4K | [Seedance 2.0](https://sloosh.ai/docs/seedance-2-0.md), [MiniMax H3](https://sloosh.ai/docs/minimax-h3.md), [MiniMax H3 Reference](https://sloosh.ai/docs/minimax-h3-reference.md), [Gemini Omni 1.1 Flash](https://sloosh.ai/docs/gemini-omni-1-1-flash.md), [Gemini Omni 1.2 Flash](https://sloosh.ai/docs/gemini-omni-1-2-flash.md) | Every Sloosh workspace can make 4K on models that offer it. |
| Fast drafts | [Seedance 2.0 Fast](https://sloosh.ai/docs/seedance-2-0-fast.md), [Pruna P-Video 2 Pro](https://sloosh.ai/docs/p-video-2-pro.md) | Draft here, then make the final on a stronger model with the same prompt. |

| Model | Length | Resolution | References |
| --- | --- | --- | --- |
| [Seedance 2.5](https://sloosh.ai/docs/seedance-2-5.md) | 4 to 30 s | 480p to 1080p | 30 images, 10 videos |
| [Seedance 2.0](https://sloosh.ai/docs/seedance-2-0.md) | 4 to 15 s | 480p to 4K | 9 images, 3 videos |
| [Seedance 2.0 Fast](https://sloosh.ai/docs/seedance-2-0-fast.md) | 4 to 15 s | 480p, 720p | 9 images, 3 videos |
| [Kling 4.0](https://sloosh.ai/models/kling-4.0) | 5, 10, 15 or 30 s | 720p, 1080p | 10 images, 5 videos |
| [Kling V3 Omni](https://sloosh.ai/docs/kling-v3-omni.md) | 3 to 15 s | 720p, 1080p | 7 images |
| [Kling 3.0](https://sloosh.ai/docs/kling-3-0.md) | 3 to 15 s | No setting | None (start image required) |
| [Kling 2.5](https://sloosh.ai/docs/kling-2-5.md) | 5 or 10 s | No setting | None (start image required) |
| [Gemini Omni 1.1 Flash](https://sloosh.ai/docs/gemini-omni-1-1-flash.md) | 3 to 10 s | 360p to 4K | 6 images, 1 video |
| [Gemini Omni 1.2 Flash](https://sloosh.ai/docs/gemini-omni-1-2-flash.md) | 3, 5, 8, 10 or 15 s | 360p to 4K | 6 images, 1 video |
| [MiniMax H3](https://sloosh.ai/docs/minimax-h3.md) | 5 to 15 s | 480p to 4K | Start and end frames |
| [MiniMax H3 Reference](https://sloosh.ai/docs/minimax-h3-reference.md) | 5 to 15 s | 480p to 4K | 9 images, 3 videos |
| [Wan 3](https://sloosh.ai/docs/wan-3.md) | 5, 10, 15 or 30 s | 480p to 1080p | 10 images, 5 videos |
| [Wan 3 Prime](https://sloosh.ai/models/wan-3.0-prime) | 5, 10, 15 or 30 s | 480p to 1080p | 10 images, 5 videos |
| [Grok Imagine](https://sloosh.ai/docs/grok-imagine.md) | 4 to 10 s | 480p, 720p | 7 images |
| [Grok Imagine 1.5](https://sloosh.ai/models/grok-imagine-1.5) | 4 to 15 s | 480p to 1080p | 14 images |
| [Pruna P-Video 2 Pro](https://sloosh.ai/docs/p-video-2-pro.md) | 5 to 15 s | 480p, 768p | Start and end frames |

- [Studio](https://sloosh.ai/docs/studio.md): every control on the Image and Video pages.
- [Models](https://sloosh.ai/docs/models.md): the limits for each model.
- [AI image generator](https://sloosh.ai/docs/ai-image-generator.md): make a first frame, then animate it.

## Frequently asked questions

### How do I make an AI video from a photo in Sloosh?

Open [Video](https://sloosh.ai/studio/video) in Studio, use the + menu to Upload frames, and add your photo as the start frame. Describe the motion in the prompt, pick a model that takes a start frame, and press Generate. [Kling 3.0](https://sloosh.ai/docs/kling-3-0.md) and [Kling 2.5](https://sloosh.ai/docs/kling-2-5.md) need a start image.

### Do I need editing skills to make AI videos in Sloosh?

No. Write a prompt, add any photo or video, or pick a Space and add your input.

### Which AI video model takes reference images in Sloosh?

[Seedance 2.5](https://sloosh.ai/docs/seedance-2-5.md) takes up to 30 reference images and 10 reference videos. [Seedance 2.0](https://sloosh.ai/docs/seedance-2-0.md) takes 9 and 3. [Grok Imagine 1.5](https://sloosh.ai/models/grok-imagine-1.5) takes 14 images, [Kling 4.0](https://sloosh.ai/models/kling-4.0) 10, [MiniMax H3 Reference](https://sloosh.ai/docs/minimax-h3-reference.md) 9, and [Kling V3 Omni](https://sloosh.ai/docs/kling-v3-omni.md) and [Grok Imagine](https://sloosh.ai/docs/grok-imagine.md) 7.

### How long can an AI video be in Sloosh?

It depends on the model. [Seedance 2.5](https://sloosh.ai/docs/seedance-2-5.md), [Kling 4.0](https://sloosh.ai/models/kling-4.0), [Wan 3](https://sloosh.ai/docs/wan-3.md) and [Wan 3 Prime](https://sloosh.ai/models/wan-3.0-prime) go up to 30 seconds. Seedance 2.0, MiniMax H3, Grok Imagine 1.5 and Gemini Omni 1.2 Flash go up to 15, Gemini Omni 1.1 Flash up to 10, and Kling 2.5 makes 5 or 10 seconds. Every plan can use a model's full range.

### Can I set the first and last frame of an AI video?

Yes, on models that take frames, such as [Gemini Omni 1.1 Flash](https://sloosh.ai/docs/gemini-omni-1-1-flash.md), [Gemini Omni 1.2 Flash](https://sloosh.ai/docs/gemini-omni-1-2-flash.md), [Kling V3 Omni](https://sloosh.ai/docs/kling-v3-omni.md), [MiniMax H3](https://sloosh.ai/docs/minimax-h3.md) and [Wan 3](https://sloosh.ai/docs/wan-3.md). Studio shows what each model accepts. Some models take frames or references, not both.

### Can I make a 4K AI video in Sloosh?

Yes, on the models that offer it: [Seedance 2.0](https://sloosh.ai/docs/seedance-2-0.md), [MiniMax H3](https://sloosh.ai/docs/minimax-h3.md), [MiniMax H3 Reference](https://sloosh.ai/docs/minimax-h3-reference.md), [Gemini Omni 1.1 Flash](https://sloosh.ai/docs/gemini-omni-1-1-flash.md) and [Gemini Omni 1.2 Flash](https://sloosh.ai/docs/gemini-omni-1-2-flash.md). Every Sloosh workspace can make 4K, whatever the plan.

### Does Sloosh make video with sound?

Several do. Seedance 2.5 makes sound by default, and you can turn it off. Kling 2.5 makes silent clips. Each model's page says what it does.