Reference Image AI Generator – DreamMMer

Reference Image AI Generator, DreamMMer lets you start from an existing photo and modify only the elements you specify. You upload a picture of a person, product, or style, tell the system what should stay the same, and describe what you want to change. The tool then creates new images or short clips that keep the original subject intact while applying the requested edits.

The service is built around an “agent” that translates your plain‑language instructions into a detailed prompt and selects an appropriate model. You can work with a single reference or add several images in one chat message. Models such as Nano Banana 2.1, GPT Image 2.5 Sunburst, Nano Banana Pro, and Seedream 5.0 Pro are available; Nano Banana 2.1 can handle up to ten references at once. When you leave the model setting on Auto, the agent picks the model that best fits your input.

Typical use cases include headshots where you keep the face, hairstyle, and expression but change clothing, background, or lighting; product photography that preserves the exact silhouette, colors, logo, and label while moving the item into new studio or lifestyle scenes; and style matching, where the color and texture of one image are applied to a different subject. The platform also supports turning a reference portrait into a short video by describing movement and camera changes; the original frame remains the starting point for the animation.

The workflow consists of three steps. First, upload your reference image(s) through the chat box. Second, write a brief description of what to keep and what to alter, for example “keep the same bottle and label, place it on wet stone at sunset.” The agent then generates a prompt, selects a model, and produces an output. Third, you can refine the result within the same conversation, asking for adjustments until the image meets your expectations. Earlier results stay in the thread, allowing you to branch or revert to previous versions.

Key features

• Upload one or multiple reference photos and define keep/change instructions in plain text
• Agent writes detailed prompts and automatically selects a suitable AI model
• Supports image generation, photo edits, and short video clips from a reference