
AI reference to video generator
Show the model your character, your product or your place and get new shots built around them. Choose from 13 models, all on one account.
You bring
Reference images
Who or what the clip should show. Most models also take reference clips, and some take audio.
A prompt
The scene they appear in.

Pick a model
13 models run reference to video here. The table below compares their limits and prices.

Add your references
Upload pictures of who or what should appear and write the scene. Most models take reference clips too.

Get your clip
You see the price before you run it, and a generation that fails costs nothing.
AI models for reference to video
Ardscale runs reference to video on 13 models, made by Alibaba, Kuaishou, MiniMax, ByteDance, Google, and ShengShu.

HappyHorse 1.1· Video
Videos up to 15 seconds, up to 1080p

Kling 3.0· Video
Videos up to 15 seconds, up to 4K

Kling 4.0 Flash· Video
Videos up to 20 seconds, up to 720p

Kling 4.0· Video
Videos up to 30 seconds, up to 1080p
| Model | Maker | Length | Top resolution | Aspect ratios | Price |
|---|---|---|---|---|---|
| HappyHorse 1.1 | Alibaba | 3–15 sec | 1080p | 16:9, 9:16, 1:1, 4:3, 3:4, 4:5, 5:4, 9:21, 21:9 | From 19 credits per second |
| Kling 3.0 | Kuaishou | 3–15 sec | 4K | 16:9, 9:16, 1:1 | From 8 credits per second |
| Kling 4.0 Flash | Kuaishou | 3–20 sec | 720p | 16:9, 9:16, 1:1, 21:9 | From 12 credits per second |
| Kling 4.0 | Kuaishou | 3–30 sec | 1080p | 16:9, 9:16, 1:1, 21:9 | From 11 credits per second |
| MiniMax H3 Max | MiniMax | 5–15 sec | 768p | 21:9, 16:9, 4:3, 1:1, 3:4, 9:16 | From 8 credits per second |
| MiniMax H3 | MiniMax | 4–15 sec | 2K | 21:9, 16:9, 4:3, 1:1, 3:4, 9:16 | From 11 credits per second |
| Seedance 2.0 | ByteDance | 4–15 sec | 4K | 16:9, 4:3, 1:1, 3:4, 9:16, 21:9 | From 10 credits per second |
| Seedance 2.5 | ByteDance | 4–30 sec | 1080p | 16:9, 4:3, 1:1, 3:4, 9:16, 21:9 | From 12 credits per second |
| Veo 3.1 Fast | 4–8 sec | 4K | 16:9, 9:16 | From 10 credits per second | |
| Vidu Q4 Preview | ShengShu | 3–16 sec | 4K | 16:9, 9:16, 4:3, 3:4, 1:1 | From 7 credits per second |
| Wan 2.6 Flash | Alibaba | 2–15 sec | 1080p | 16:9, 9:16, 1:1, 4:3, 3:4 | From 5 credits per second |
| Wan 3.0 Prime | Alibaba | 2–30 sec | 1080p | 21:9, 16:9, 4:3, 1:1, 3:4, 9:16 | From 11 credits per second |
| Wan 3.0 | Alibaba | 2–30 sec | 1080p | 21:9, 16:9, 4:3, 1:1, 3:4, 9:16 | From 7 credits per second |
Length, resolution and aspect ratios show each model’s full range. One mode can narrow it, and a clip that starts from your image usually takes the image’s shape.
More video tools

Text to video
Write a scene and get the clip.

Image to video
Turn a photo into a moving shot.

First and last frame to video
Set both ends of a shot and get the motion between.

Video to video
Change a clip you already have.

Motion control
Make your character copy the moves in a video.
Which AI models run reference to video on Ardscale?
Ardscale runs reference to video on 13 models: HappyHorse 1.1, Kling 3.0, Kling 4.0 Flash, Kling 4.0, MiniMax H3 Max, MiniMax H3, Seedance 2.0, Seedance 2.5, Veo 3.1 Fast, Vidu Q4 Preview, Wan 2.6 Flash, Wan 3.0 Prime, and Wan 3.0.
How long can the clips be, and at what resolution?
Clips run up to 30 seconds, on Kling 4.0, Seedance 2.5, Wan 3.0 Prime, and Wan 3.0. The sharpest output is 4K, from Kling 3.0, Seedance 2.0, Veo 3.1 Fast, and Vidu Q4 Preview.
How much does reference to video cost?
On Ardscale, reference to video starts at 5 credits per second, on Wan 2.6 Flash. Every model has its own price and you see it before you generate. A generation that fails costs nothing.
What is the difference between reference to video and image to video?
In image to video, your picture is the first frame of the clip. In reference to video, your pictures show the model who or what to include, and it composes new shots around them.