Skip to content
Rook Rook Rook

RookVision

RookVision gives Rook a second sight. The viewport becomes a canvas and the canvas a viewport: you can generate imagery from inside Rhino — capture a view, describe the look you want, and bring the result back into the model.

“Render this elevation as a watercolour, dawn light, faint mist.”

The loop

Section titled “The loop”

RookVision works as a round-trip:

  1. Capture — an in-memory grab of a viewport: a named view, a display mode, a raytraced render.
  2. Enhance — a prompt enhancer composes the scene context, view direction, units, and your intent into a strong prompt.
  3. Generate — provider-agnostic image generation: multiple models, multiple keys, bring-your-own-key.
  4. Return — approved results come back into the document as a PictureFrame, a material, or a reference.

What it’s for

Section titled “What it’s for”
  • Presentation — turn a working view into a rendered image without leaving Rhino
  • Exploration — try several looks for the same elevation, fast
  • Communication — produce something to share or react to

Video

Section titled “Video”

RookVision also generates video with Veo 3.1 (Standard, Fast, and Lite): text-to-video, image-to-video, and first/last-frame video.

  • Source framing — a source image is prepared automatically to fit the output shape. Fit keeps the whole image and pads it; Fill crops to the shape. You see the prepared image before you confirm.
  • Quotes and cost caps — every job is quoted before it runs, and your cost cap applies.

Providers and fallback

Section titled “Providers and fallback”

Google is the primary provider for Nano Banana images and Veo video. You can use it with your own Google AI Studio key, or through Google Enterprise: your organization’s own Google Cloud project (Vertex AI), billed to that project. Sign in with a Google account or a Microsoft 365 / Entra identity in RookChat Settings → Enterprise Google, then choose a “Google Enterprise” model.

If you store a fal.ai key, RookVision falls back to fal.ai for Nano Banana images, prompt enhancement, and Veo 3.1 video in two cases:

  • no Google key is stored, or Google rejects the stored key;
  • for video, Google could not be reached before it accepted the job.

A job Google has already accepted is never sent a second time. Video from reference images and Veo 2.0 / 3.0 have no fal.ai fallback.

Status

Section titled “Status”
  • Image — live
  • Video — live
  • 3D — roadmap (see Hunyuan 3D)
Section titled “Related”