Docs
Editor & Canvas Objects
The object types you can place on the canvas, how gestures work, and where each canvas tool is documented.
The canvas is an infinite board inside a workspace. Pan by dragging or scrolling, zoom with Ctrl/⌘ + scroll (or a trackpad pinch), and place objects anywhere — then connect them and generate renders from them.
Canvas objects
Every kind of object you can place and drag on the canvas. Add one with the
Add element button, or press W to open the quick-add ring at
your cursor — the ring places the object exactly where you pressed, the button
places it at the centre of the view.
| Object | What it is |
|---|---|
| Image / video layer | A generated or uploaded image (or video) — the primary render target. A Frame is a layer too: a blank one in a fixed aspect ratio, ready to generate into or sketch on. |
| Sticky note | A colored note for annotations and reminders. |
| Text box | Free-form rich text placed on the canvas. |
| Vector drawing / shape | Pen paths, shapes, arrows and freehand strokes. |
| Video compositor | Turns image frames into an AI-generated video clip. |
| 3D world | Generates an explorable 3D world from an image or prompt. |
| Prompt widget | A standalone prompt tile that renders connected images. |
| Router | Fans one input out to multiple downstream generations. |
| Empty Stack | A stack that exists before it holds anything — drop images in and it fills, empty it and it stays. The fillable slot a shareable pipeline template wires to. |
| Upscaler | Upscales one connected image and drops the result beside it, without touching the prompt bar. |
| 3D asset | Turns one to four reference images — or a whole stack — into a textured 3D mesh you can orbit on the canvas, remesh, and export as GLB, FBX, OBJ or USDZ. |
| Cutout | A person/object silhouette used as a placement reference. |
Frames appear in both add menus but not as a separate row above, because a frame is a layer — an empty one at a fixed aspect ratio. Everything on this page that applies to a layer applies to a frame. See Frames for the presets and how to render into one.
Gestures
Objects respond to four kinds of pointer gesture:
| Gesture | What it does |
|---|---|
move | Translate an element (drag its body). |
resize | Resize via corner/edge handles. |
transform | Rotate or scale (vector elements). |
endpoint | Redock a connection endpoint. |
Vector shapes
The shape and vector / drawing tools can create these shape types:
| Shape | What it draws |
|---|---|
| rectangle | A rectangle or square. |
| ellipse | An ellipse or circle. |
| line | A straight line. |
| arrow | An arrow (for callouts and direction). |
| polygon | A multi-point polygon. |
| path | A freeform Bézier path (pen tool). |
Every shape is stylable: stroke color and width, fill color (for closed shapes), opacity, line style (solid, dashed or dotted), corner radius (rectangles), arrowheads (arrows and lines), and a roughness control that ranges from clean, precise lines to a loose, hand-drawn / sketchy look.
In this section
Deeper guides for each canvas tool:
Images and layers:
- Images & layers — add, duplicate, reorder and delete layers.
- Crop an image — trim a layer with aspect-ratio presets.
- Frames — empty render targets at a fixed aspect ratio.
- Stacks — compare, merge and expand image variations, and place an Empty Stack before you have the images.
- Adjustment layers — non-destructive colour grading.
- Layer masks — hide or reveal parts of a layer.
- Annotate the canvas — sticky notes, text and People cutouts.
Widgets that turn images into something else:
- Video Compositor — turn a still, or a chain of frames, into short AI video clips.
- 3D assets — one to four images into a textured mesh you can orbit and export.
- 3D worlds — an explorable environment you can walk through.
- Upscale a render — enlarge and enhance a finished image.
Wiring objects together:
- Prompt widgets & routers — standalone prompt tiles, and fanning one input out to many.
- Connections — draw, select and restyle the lines between objects.
- Wire a pipeline — chain the above into a repeatable multi-step flow.
Arranging:
- Groups — combine objects into one movable unit, and nest them.
- Alignment & snapping — line objects up with guides.
On tablets & phones
INTER\STITIAL works on desktop and on tablets / iPads (a big screen helps). On phones, a workspace is capped at 15 layers and images render at a lower resolution to save memory.
Notes
Double-click a layer to zoom to it; double-click a group to enter it. Hold
Shift while dragging to lock movement to one axis. Drag a layer near the top-left
corner to snap it to the origin, and press Tab for smart alignment guides.