Introduction
Video to Video allows you to change the style of your videos by using a text prompt or input image as the first frame.
This article outlines how to use Video to Video on the Gen-3 Alpha and Turbo models, the available settings, and more.
Spec Information
| Spec | Gen-3 Alpha | Gen-3 Alpha Turbo |
|---|---|---|
| Cost | 10 credits per second, rounded up to 5 second increment | 5 credits per second, rounded up to 5 second increment |
| Supported durations | Up to 20 seconds | |
| Explore Mode on Unlimited Plans | Yes | |
| Platform availability | Web | |
| Supported video inputs |
See here |
|
| Output resolutions | 1280x768 | 1280x768 768x1280 |
| Maximum input size | 64mb | |
Step 1 – Selecting the Input Video
Begin by navigating to Generative Session in your Dashboard.
From here, make sure the Gen-3 Alpha or Gen-3 Alpha Turbo model is selected in the bottom left corner dropdown.
Select a supported video file from your assets or drag and drop a video to upload it.
You'll be prompted to crop your input video if it is not in a supported resolution.
Credit notes
When choosing a video, please note that credits are charged per second, but rounded up to the nearest increment of 5. Below are some examples of credit costs based on the length of the video and selected model:
- 4 seconds of Gen-3 Alpha Turbo: 25 credits
- 5 seconds of Gen-3 Alpha Turbo: 25 credits
- 12.6 seconds of Gen-3 Alpha: 150 credits
- 15 seconds of Gen-3 Alpha: 150 credits
We recommend choosing videos in 5 second increments for optimized credit usage. Videos longer than 20 seconds will be automatically trimmed to the first 20 seconds of the input.
Step 2 – Drafting the Prompt
Video to Video can dramatically change the style of your input with a strong text prompt or stylized input image.
Text Prompts
A descriptive prompt that focuses on the desired style will provide the best results in most cases.
This tutorial uses an example with a green screen mask, but you can use any type of video input.
| Input | Prompt | Output |
| 3D halftone CMYK style. halftone print dot. comic book. vibrant colors in layers of cyan blue, yellow, magenta purple, and black circular dots. |
Tip: Try out a Custom Preset to load a curated example prompt. Be sure to replace the purple placeholder text with any customizations you wish to include.
Simple prompts tend to work well, but detailed prompts are more effective and will yield results more closely aligned with the concept you’re envisioning.
Image prompt
Alternatively, you can upload an image to apply a style. A restyled first frame is your video's opening frame reimagined in your desired artistic style. For best results:
- Maintain the same composition as your original first frame
- Clearly showcase your desired style (cartoon, painting, sketch, etc.)
- Use your video's actual first frame as a reference
The input image will act as the first frame of the video, so using a stylized image that matches the first frame of the input video will yield the highest level of consistency and control:
| Input | Input video first frame | Restyled first frame | Output |
Using an input image that doesn’t match the first frame will yield more creative results, and may result in a brief skip effect after the first frame. This skip can later be trimmed out by using Actions > Edit Video.
If using an input image that doesn’t match the first frame, try increasing the Structure transformation value for more flexibility in how motion from the video is applied with the image.
Step 3 – Configuring the Settings
Video to Video has additional Settings that you should review before starting your generation.
Structure transformation
This setting controls the structural consistency between the input and output. Lower values will result in an output that maintains the structure of the original video. Higher values will result in more abstract outputs.
| Structure transformation value | Output |
| 0 | |
| 5 | |
| 10 |
Fixed seed
Using a fixed seed will allow you to create similar generations. This is unchecked by default to give you a wide variety of results.
Copy and paste the seed of a previous output if you'd like to receive generations with similar style and movement. Pasting the seed will automatically check the box.
Aspect ratio
On the Turbo model, you'll have the option to choose between a landscape and vertical aspect ratio.
Step 4 – Generating the Video
After drafting your text prompt and configuring your settings, you're now ready to generate your video.
The final result will match the length of the input video if it's 20 seconds or less. For videos longer than 20 seconds, only the first 20 seconds will be used.
Your generations will be scrollable through your session as you continue to generate. You can also access completed videos in your Assets, where they will save to the Generative Video folder by default.