Dreem is designed to generate high-quality fashion ecommerce content without requiring detailed prompts from you.
For most standard workflows, you do not need to write a full creative prompt. Dreem already includes structured generation flows, predefined output formats, pose options, talent options, product views, and prompt libraries that are designed around what good fashion ecommerce content should look like.
Prompts are most useful when you want to add extra creative direction, clarify a specific styling requirement, or refine a generated result.
When you do not need a detailed prompt
You usually do not need a detailed prompt when you are using one of Dreem’s main generation flows: Virtual Model, Content Kit, Image-to-Video, or Product Shot.
These workflows already include built-in guidance for fashion ecommerce content. For example, Virtual Model includes model and pose libraries, Image-to-Video includes a prompt library with expected output behavior, Product Shot includes predefined fashion product views such as flat lay and ghost mannequin, and Content Kit combines Dreem’s generation options into a faster content creation workflow.
In most cases, start by uploading clear product images and selecting the available options that best match your goal. Add a prompt only when you need to clarify something that is not already covered by the selected workflow, model, pose, view, or prompt option.
When prompting is helpful
Prompts are most helpful when your request does not fit one of Dreem's main generation flows (Virtual Model, Content Kit, Image-to-Video, or Product Shots). In those cases, you can describe what you want in chat, and Dreem will help turn the request into a useful next step. If more information is needed, Dreem may ask a clarifying question.
If you are using one of Dreem's main generation flows, start by following the workflow and selecting the available options instead of writing a long prompt. After the result is generated, use Reuse to create a new version with a short instruction, or use Image Editing to make a targeted change to part of the image.
Keep prompts short and direct. You do not need to write a complex brief. One clear instruction is usually better than a long prompt with many details.
What you want to do | Example prompt |
Change product color | “Change the vest to black, keep everything else the same.” or “change the skirt color to #ffe5e7 (HEX), keep everything the same” |
Create multiple color versions from one image | “Create five flat lay front images of this item in black, white, grey, blue, and green.” |
Change the background | “Change the background to subtle vertical or diagonal panels behind the model, with shadow” |
Add makeup to a model’s face (Learn more) | “Add natural makeup with soft blush and subtle lip color. Keep the outfit, pose, background, and lighting the same.” |
Refine the model’s facial expression | “Make the model’s expression softer and more relaxed. Keep everything else the same.” |
Extract an outfit from a model | “Extract all items from this outfit, create separated product shots. Do not combine all items in 1 image” |
Button up or unbutton a shirt, jacket, or blazer | “Unbutton only the top two buttons of the shirt.” / “Button all buttons except the bottom one.” |
Adjust the length of a fashion item | “Make the skirt slightly above the knees.” |
For model makeup, facial polish, or small styling changes to the face, use Image Editing instead of regenerating the full image when the rest of the output already looks good. Select the face area with the brush tool, then describe the change you want, such as natural makeup, bolder makeup, defined brows, blush, lip color, or a more polished campaign-ready look. Learn more
For fashion length adjustments, use visual language instead of exact measurements. Dreem may not perfectly understand instructions like “make the skirt 90 cm long.” It is usually better to describe the visible result you want, such as:
“Make the skirt slightly above the knees.”
“Make the dress ankle length.”
“Make the sleeves shorter and ending above the wrist.”
“Make the jacket cropped at the waist.”
In general, prompt for what you want to see in the final image, not technical measurements.
Review the prompt preview before generating
When you enter a prompt, Dreem may refine your instruction and show a preview of how it will interpret the request before generating.
Review the preview to make sure it matches what you want. If it looks right, click Generate. If something needs adjusting, edit the prompt before generating.
This is useful for prompts such as color changes, background changes, buttoning or unbuttoning, garment length adjustments, or creating multiple color variations from one image.
Use clear fashion product inputs
The quality of your uploaded product images matters more than writing a long prompt.
For best results, upload product images that are clear, high resolution, well lit, not heavily shadowed, not cropped in important areas, and show the full product shape. Make sure important details such as logos, prints, straps, closures, texture, or stitching are visible.
For fashion products, it is helpful to provide multiple angles when available, such as front, back, side, detail, or flat lay.
Use mannequin or worn references for complex garments
Some fashion items are harder to understand from a flat lay image alone. This is especially true when the product has a special structure or can be worn in multiple ways.
Examples include corsets, ponchos, shawls, wraps, multi-way dresses or tops, fake sweater scarves, layered garments, asymmetrical pieces, and products with ties, straps, belts, or unusual closures.
For these items, use a mannequin or worn reference when possible. This helps Dreem understand how the garment should sit, wrap, drape, fasten, or layer on the body.
💡 A prompt can help clarify the intended styling, but the visual reference is often more important.
Keep edit prompts focused
When refining an output, do not rewrite the full creative brief unless you want a completely new result.
For Image Editing or small corrections, focus on the exact change you want.
Instead of:
“Fix the bag.”
Try:
“Remove the extra strap on the left side of the bag. Keep the bag shape, logo, color, and hardware unchanged.”
Instead of:
“Make the outfit more premium.”
Try:
“Change the shoes to black loafers. Keep the model, pose, product, background, and lighting the same.”
Say what should stay the same
If most of the result is already correct, include what should not change. This helps keep the edit focused.
Examples:
“Keep the product shape, color, material, logo, and stitching the same.”
“Keep the model, pose, outfit, background, and lighting unchanged.”
“Keep the front view and product proportions the same. Only adjust the background.”
This is especially useful for color changes, product-detail corrections, unwanted extra elements, or small styling edits.
Summary
You do not need to be an expert prompt writer to use Dreem.
Dreem is built around fashion ecommerce workflows, with predefined generation types, model and pose libraries, product shot views, and video prompt options that already guide the output toward strong ecommerce content.
Use prompts as an extra layer of direction, not as the main setup method. The best prompts are short, specific, and focused on what Dreem cannot already infer from your uploaded product images, selected workflow, and chosen settings.
Data use note: Prompts, uploaded images, generated outputs, and other platform data are treated as customer data. For Pro customers, this data is not used to train AI models unless the customer opts in. For Free, Starter, and Team plans, Dreem may use aggregated and de-identified platform data to improve AI feature quality, with safeguards described in our Privacy Policy.