Synthetic Talent enables you to create a fully virtual model using two flexible approaches:
Image-based generation: Upload a reference image to serve as a visual starting point, with optional instructions to guide the final result
Prompt-based generation: Describe the talent you want using text prompts, defining attributes such as appearance, style, and mood
This flexibility allows you to either build from an existing visual direction or create entirely new talent from scratch.
By Upload
Upload a clear face shot of a single person. For best results, use an image with good lighting and minimal distractions, as this will improve generation quality.
You can optionally add instructions to guide the result (e.g., “make the beard longer”). Prompts can be short and direct or more detailed, depending on your needs.
Disclaimer:
By uploading an image, you confirm that you have the right to use this content. Dreem will analyze the image and any additional comments to generate synthetic talent options based on the provided references.
Once your photo has been analyzed, Dreem will generate four talent options. If you’re happy with one of the results, you can move forward by selecting Use.
If the generated options aren’t what you’re looking for, you have several ways to refine or save your results:
👎 Feedback (Thumbs Down): If the generated options aren’t what you’re looking for, you can provide feedback by selecting the thumbs down icon. This explicit feedback helps improve model performance over time.
❤️ Favorite (Heart Icon): Save a talent option to revisit or use later.
➕ Generate More (Plus Icon): Create additional variations similar to a selected result.
You can also use the prompt box to add optional instructions for the next generation, helping guide the output more precisely.
When you choose to Use a generated talent option, you’ll have the ability to further customize the profile before saving:
Update the talent name
Edit the profile description
Adjust selected attributes, which can be used as filters within the Talent Library
You can also explore the Lookbook for inspiration. The Lookbook is designed to showcase a range of variations using the same talent—similar to real-world model agency lookbooks—helping you understand how the talent can be styled and applied across different use cases.
By Description
You can create synthetic talent by using either a brief description or a more detailed prompt—both approaches work effectively and can be used depending on your needs.
Similar to the image-based method, you have full flexibility in how you guide the output, whether you prefer simple inputs or more specific direction.
Disclaimer:
Generated talents are AI-created and are not intended to represent or replicate real individuals.
From here, the same options and steps apply as with Synthetic Talent (image-based), including reviewing generated results, providing feedback, generating additional variations, and finalizing your selected talent.
From there, you can save the talent, and it will live in the Talent Library for use when creating on-model content.