DREAMMMER is an agentic AI image and video editor. It uses a conversation-driven approach: users describe the desired result in natural language, the agent understands the intent, searches the web for references when necessary, writes detailed English prompts, selects a suitable model, and assists the user in refining the result within the same conversation until they are satisfied. The left side of the interface is the chat area, and the right side is an infinite canvas, which visualizes the entire creative process as card nodes, showing each step's request, prompt, render, and modifications, while preserving a complete lineage that can branch from any step. Users can also hover over any rendered result to directly edit, remove the background, upscale, generate a video, or attach the image to the agent for further processing. The canvas supports dragging nodes, snapping to a grid, two-finger pan and zoom, and full-screen or custom split-screen ratios.
The platform requires no prompt engineering; users can describe their ideas in their native language, and the agent automatically constructs lighting, composition, and style. Image and video creation are completed in the same conversation, supporting uploading reference images or using a just-generated image as input for subsequent steps, enabling the blending of people, clothing, products, and backgrounds. The agent can also generate videos starting from text or images, describing motion and camera direction, and supporting reference images, start frames, and end frames.
DREAMMMER offers a variety of templates, including product images, posters, avatars, 3D avatars, stickers, anime, logos, interior design, and more, helping users get started quickly. The model library includes Nano Banana 2.1 (Google's latest image model, 1K–4K, up to 10 reference images, starting at 7 credits), GPT Image 2.5 Flare, GPT Image 2.5 Sunburst, Nano Banana Pro, Seedream 5.0 Pro, Qwen Image 3.0 Pro, and others, suitable for concept art, product photos, packaging design, Chinese text layout, and other scenarios respectively. Users can set the model to automatic or choose the model and settings such as resolution themselves.
Usage flow: describe the idea, let the agent create, then refine together. You can browse templates and public works (no login required), and when creating, log in with Google to save conversations, works, and credit balance. Image and video generation consume credits, and the number of credits depends on the model and settings. Generation time varies by model and task, and videos may take several minutes. If a task fails or gets stuck, you can view task details or contact support. Credits for failed tasks are automatically refunded. Conversations are private by default, and users can decide whether to enable a public share link.
This tool is suitable for creating product visuals, posters, portraits, social content, and more, helping users go from the initial idea tothe final image or video, all completed in a single conversation. Detailed pricing and specific credit rates are not yet provided in a table; please check the pricing page.
