Skip to main content

Working with the Dreem Chat Agent V3

The Dreem Chat Agent lets you create product content by describing what you want in plain language. It detects your product angles, works with several images in one request, suggests talent and motion, and shows you each step before any credits are used.

Written by Phuong Anh (Sofia)

Watch the tutorial

Prefer a quick walkthrough? Watch this short tutorial to see how Content Template works in Dreem:

You can also follow the guide below.

What the Chat Agent can do

Detect product angles automatically

When you upload a garment, the Chat Agent identifies whether the image shows the front, back, or side view. You don’t need to label your images.

Work with multiple images in one request

You can refer to several uploaded images in a single message, for example:

“Put the dress from the 2nd image on the model in the 1st image.”

Generate Content Kits step by step

Content Kits run in three stages: Product Shots from your flat lays first, then model shots, then video. You can review and adjust the results at each stage before moving on.

Suggest talent, poses, and motion

Based on the style of your product, the Chat Agent suggests suitable models, poses, and video motions, such as a 360° Turn. You can use a suggestion or choose your own.

Preview your prompt before generating

For general editing requests, the Chat Agent refines your prompt and shows how it will interpret your request. Nothing is generated and no credits are used until you confirm.

Show what a request will cost

If you don’t have enough credits for a request, the Chat Agent tells you how many credits it needs and how many you have available.

How to get the best results

  1. Upload clear product images. Angle detection works best with sharp, well-lit images that show the full product.

  2. Check the detected views. Confirm each image was recognized as the right angle before generating, especially for garments that look similar from the front and back.

  3. Refer to images by order. In multi-image requests, say which image is which, such as “the 1st image” or “the 2nd image”. Keep each instruction to one clear action.

  4. Review each Content Kit stage. Approve the Product Shots before moving to model shots, and the model shots before moving to video. Fixing something early saves credits later.

  5. Read the prompt preview. If it matches what you want, click Generate. If not, edit the prompt first.

Summary

The Chat Agent is built around how ecommerce content gets made: upload your products, describe what you need, and review each step as you go. It handles angle detection, suggests talent and motion, and previews your request before generating, so you can focus on approving the content you want. If something doesn’t look right, adjust the result at the stage where it happens, or contact us through the chat with your images, prompt, and the result you expected.

Did this answer your question?