Text to Image Generation

DreamStudio turns a written prompt into a rendered image using Stable Diffusion models. A Beam agent picks up a content request marked ready for artwork, pulls the prompt and any style notes from that record, and submits the job to DreamStudio through the connected account. Once the render finishes, the agent attaches the image file back to the originating record and updates its status, closing the loop between a brief and a finished asset. Prompts that come back flagged, blocked by moderation settings, or missing key details are sent to a person for a rewrite instead of a silent retry.

Image to Image Variation

Beyond text prompts, DreamStudio can restyle an existing image, useful when a team wants another version of an asset rather than a fresh one. A Beam agent reads the source image and the requested change from an approved brief, submits both to DreamStudio, and saves the resulting variation alongside the original so both stay linked in the record. This keeps a visual history of edits without anyone downloading and re-uploading files by hand. Requests that ask for edits outside the account's configured style rules, or that fail to render cleanly, are handed to a designer to finish.

Render Status and History

Every job submitted to DreamStudio carries a status: queued, rendering, complete, or failed. A Beam agent checks that status on a schedule, matches finished or failed jobs back to the record that requested them, and updates the record so requesters see progress without opening DreamStudio themselves. Completed renders get filed into the asset library tied to the original brief; failed ones are noted with the reason DreamStudio returned. A person still looks at any batch of failures that repeats, since that usually points to a prompt or account setting the agent cannot fix on its own without guidance.