Use PixVerse Creative Studio to apply custom styles to your effect. Start by entering prompt words, then personalize it with additional settings for your Agent.
In this guide, we will show you how to customize the Riding Dinosaurs theme Agent and create your own version!
Creating a Agent
- Open PixVerse Creative Studio and navigate to Agent.
- Click Create Agent to open the Agent Editor window.

-
Now the Agent Editor window opens. Enter a prompt in the <Template Prompt> Box to test your idea.
-
For example, for the Riding Dinosaurs theme, you can try and input the following prompt:
Character in the {input_image} are riding different kinds of dinosaurs.
-
Adjust other parameters to better align the template with your expectations.
Template Prompt |
Required, input your prompt here. |
Model |
Currently only PixVerse V5 model is supported, stay tuned for updates |
Script Type |
You have two options: Basic Mode and Long Shot Mode. |
- Basic Mode splits the scenes based on your shot count.
- Long Shot Mode uses the Transition feature to connect different shots. |
| Shot Model | You can choose from 4 image generation models for this process.
Why I need to choose image generation models?
— The Agent's creation process generates images based on the prompt, then converts them into video. |
| Shot Count | This parameter determines the number of shots in your video. Up to 4 shots are supported. |
| Duration | Minimum 5 seconds, maximum 30 seconds |
| Resolution | For testing purposes.
Some prompts may yield different results with different resolution options. Please consider this when testing. |
| Ratio | This parameter determines the aspect ratio of your video. |
| User Input Type | Specify how users will interact with the Agent. We generally recommend selecting Image.
For example, with Riding Dinosaurs:
- If Text is selected, the output will be: "{input_text} are riding different kinds of dinosaurs."
- If Image is selected, the output will be: "Character in the {input_image} are riding different kinds of dinosaurs." |
| User Image Count | Here, you can specify how many images users can upload. |
-
Upload one of your photos or select a photo from Official Assets to simulate the user’s input image and test if the prompt can generate the preset effect in most cases.
<aside>
📌
Tips:
- When testing, try selecting several images with different genders, skin tones, species, styles, and aspect ratios.
- You can start with simple prompts. The goal of the Agent is to create a video with one click, so the simpler the prompt, the more likely it is to produce surprising results.
</aside>

This is Template Editor window.

Enter a prompt, select a photo, then click Preview Template to test the effect.

Check your generated results in the history section.
-
If you need to take a break, click Save. When you're done testing and everything is ready, click Submit. Both options will save your creation as a Project Draft.
<aside>
📌
Tip:You can click the duplicate button in the top-right corner of the Draft to create a new draft and continue working from the previous one.
</aside>


Submitting Your Agent
- Once you’ve tested your Agent idea and confirmed that the prompt works to transform the input image into the preset effect in most cases, you can download 1-2 videos for future submission as cover samples (choose watermark-free when downloading as cover samples).

<aside>
📌
Tips:
- The cover display video for the Agent does not need to be fixed at 3:4 ratio; it should simply match the aspect ratio of your Agent creation.
- Ensure the uploaded videos fully preview the template effect.
</aside>
- Select background music for your template and submit an MP3 file with a duration no less than the template's length.

<aside>
📌
Tip:
- Choose music that complements your template theme.
- You can also browse TikTok’s popular music charts to find potential viral content: TikTok Popular Music.
</aside>
- Give your template a name and description, using the language you are most comfortable with. Once the template is approved and launched, it will be translated into multiple languages by AI.
