Dreamina brings canvas planning, targeted edits and preview renders to AI filmmaking

01One workspace for your scene references

Dreamina’s Octo workspace brings images, video, scripts, mood boards and production notes onto a single canvas. Uploaded material provides context for generation, while grouping tools keep related assets together.

For an episodic short, a recurring presenter or a brand campaign, that means keeping character references, locations and shot instructions alongside the work they describe. A team can explore several versions of a scene without scattering its assets across unrelated chats and folders.

02An assistant inside the production board

Octo’s creative chat helps develop ideas, generate assets and refine concepts through conversation. Its built-in web search can bring research and references into the project, and canvas notes hold feedback and production requirements.

A director could collect visual references, describe the intended scene and develop a shot plan in the same workspace. The benefit is having the planning context beside the generated material when it is time to revise.

03Targeted edits inside a generated clip

Dreamina’s official Seedance 2.5 announcement describes edits to specific timestamps, characters and other elements inside a video. It also lists clips up to 30 seconds and support for up to 50 multimodal references, including images, video, audio and text.

Targeted revision matters when most of a shot already works. If the framing and performance are right but one visual detail is wrong, a local edit gives the creator a narrower request than starting again. Think of a distracting object, an unwanted character detail or a short section that needs another pass.

04Generate storyboard images on the same canvas

Octo also supports image generation and editing. Dreamina’s guide names Seedream 5.0 and GPT Image 2 among the tools available for developing visuals.

That is useful for a sequence that starts as a rough idea: create concept images, explore a location or build storyboard frames before moving into video. Keeping those images beside the script and references makes it easier to see whether the visual direction still fits the scene.

05480p previews before a full-quality render

Dreamina’s own guide describes Seedance 2.5 Preview as a lower-cost 480p generation that can become a native 1080p result using retained model information. This differs from upscaling an already completed video: the preview belongs to the generation process itself.

The attraction is deciding whether an idea works before paying for its final quality. Composition, pacing and the broad performance are decisions that can often be made from a rough version. Fine texture, small facial changes and edge detail need a closer look at the final output.

The guide does not provide a specific saving or a general rollout date for this mode. Check the available controls and credit quote in your account before relying on it for a project.

06Where to try it

To start a video in Dreamina, open AI Video, add your references, describe the scene and select Seedance 2.5. Set the duration and aspect ratio before generating. For a project with several scenes, use Octo to organize the references and planning material.

The preview-render option is separate from the standard Seedance 2.5 model selector. Look for a Preview mode in the generation screen; its presence and quoted credit cost determine whether you can use the new two-stage process.

The strongest use case is a creator making several related scenes: shared references reduce repeated setup, the canvas keeps the pieces organized, and targeted edits offer a way to refine a promising result.