Prompting Guide
Prompting Guide for Charm AI Beta 0.1
To get the best results from Charm AI Beta 0.1, crafting effective prompts is essential. Here’s a guide to help you maximise the potential of the platform and ensure your video generations align closely with your vision:
Start Your Prompt with “Who” Begin your prompt by describing the character in the uploaded image. This helps the model anchor the video generation to the correct visual style and character. For example:
If your image is an NFT of a penguin, start with “2D Penguin character.”
For a robot, use “2D Robot character.”
For characters that are more humanlike, use "2D anime girl/boy"
Keep Prompts Short and Clear Avoid overly elaborate or descriptive prompts. Instead, focus on a few key details to guide the generation. For example:
Good Prompt: “2D Penguin character typing on a keyboard.”
Avoid: “A cute 2D penguin sitting in a modern office typing rapidly on a mechanical keyboard while looking very busy and glancing around.”
Stay Close to the Initial Image The best results come from prompts that align closely with the uploaded image. If you want the character to perform a specific action, make sure the initial image supports that context. For instance:
If you want a character typing on a keyboard, ensure the uploaded image shows the character sitting near one.
If the image is of a standing character, stick to actions that match that position (e.g., walking, gesturing).
Experiment and Adjust
When using the Turbo model, you may need to adjust and refine your prompts to find the best combination for your desired outcome. Experiment with different phrasing to achieve consistency.
The Performance model is more forgiving, but aligning your prompt with the image is still crucial for optimal results.
By following these guidelines, you’ll improve the quality and consistency of your generated videos while making the most of Charm AI Beta 0.1’s capabilities.
Last updated