Canvas - Beeble Documentation

Canvas

Canvas is a powerful node-graph editor that brings every Beeble AI model into a single, unified workspace. Build custom processing pipelines by wiring together tools like SwitchX, Image Generator, and Video Matte. Simply import your raw footage, map out your node network, and generate production-ready results. With the flexibility to branch off creative variations at any step, Canvas gives you complete control from raw input to final render.


Node Catalog

Canvas ships with four node groups.

Input

Node Output Notes
Text text Plain text prompt.
Image image PNG, JPEG, or WebP.
Video video MP4 or MOV.

AI Tools

Node Inputs Output Notes
SwitchX source video
reference image
alpha video
prompt
video Same model as the standalone SwitchX product. Source and alpha handles are video-only. If you need Fill Mode, leave the alpha input disconnected. Set a random seed for reproducibility.
LLM prompt
media
text Generate text from a prompt using an LLM. Supports up to 4 media inputs (image or video). Use @ to reference connected media in your prompt.
Image Generator prompt
images
image Supports 1 to 4 reference images.
Video Generator prompt
start / end frame
reference media
video Generate a short clip via external models (e.g., Seedance 2.0). Pick a model in the footer:
- Text-to-Video — prompt only
- Image-to-Video — needs a Start Frame (End Frame optional)
- Reference-to-Video — up to 5 reference slots (max 3 images, max 2 videos)
Video Matte video alpha
RGBA
Two outputs: Alpha (grayscale matte) and RGBA (source merged with alpha). Two modes: Auto (one-click) or Select (interactive SAM editor). Choose between MatAnyone (general footage) and CorridorKey (green-screen keys). Export the RGBA composite as MOV or a PNG sequence, and the alpha as a PNG sequence.
VFX Pass Generator image / video image / video × 6 Extracts PBR material maps (BaseColor, Normal, Depth, Roughness, Specular, Metallic) from an image or video using SwitchLight 3.0. Export passes individually or all at once with Export All, including PNG sequences.
Depth Generator image / video image / video Extracts a depth map from an image or video using Video Depth Anything. Export as a 16-bit EXR for higher-fidelity depth workflows.
Video Upscaler video video Upscale a video 1×–4× (up to 4K) with Topaz Video AI. Pick a Mode in the footer (Proteus, Artemis MQ, Nyx, Gaia HQ, Starlight HQ); codec and frame interpolation live under Advanced.
Image Upscaler image image Upscale an image 1×–4× (up to 8K) with Topaz Gigapixel AI. Pick a Mode in the footer (Standard V2, Low Resolution V2, CGI, High Fidelity V2, Text Refine, generative Recovery / Redefine / Wonder, and more).
Image Describer prompt
images
text Generate a detailed text description of connected image(s) using an LLM. Connect up to 4 images. Comes with a built-in, editable prompt — hit Reset to restore the default.
Video Describer prompt
videos
text Generate a detailed text description of connected video(s) using an LLM. Connect up to 4 videos, each up to 5 minutes. Comes with a built-in, editable prompt — hit Reset to restore the default.
Camera Angle (Qwen) image image Re-render a still from a different camera angle.

Edit

Node Inputs Output Notes
Extract Frame video image Extract a single frame from a video
Trim video video Trim a video to a specific frame range
Crop image / video image / video Crop an image or video. Interactive overlay or numeric inputs.
Resize image / video image / video Resize an image or video to target dimensions
Curves image / video image / video Adjust the tonal range with interactive RGB and per-channel (R/G/B) curves. Drag points to shape the curve; click empty space to add a point; double-click or drag a point off the graph to remove it.
Invert image / video image / video Invert RGB colors of an image or video.
Rotate / Flip image / video image / video Rotate an image or video in 90° increments, or flip it horizontally or vertically.
Erode / Dilate image / video
mask
image / video Dilate (grow) or erode (shrink) bright regions. Typically chained after Video Matte for matte cleanup. Strength ranges from −200 to +200. Optional mask input.
Channels image / video image / video Isolate a single RGBA channel (R, G, B, or Alpha) and view it as grayscale.
Blur image / video
mask
image / video Apply a Box or Gaussian blur (Size 1–100). Optional mask input to confine the blur to specific regions.
Merge background
foreground
mask
image / video Composite foreground over background with 16 blend modes (Normal, Multiply, Screen, etc.). Inputs accept image or video independently. Optional grayscale mask.
Merge Alpha RGB
alpha matte
image / video Combine an RGB source and a grayscale alpha matte into a single RGBA stream. Both inputs accept image or video.
Compositor layers video Composite multiple layers into a single video with per-layer blend modes, opacity, and timing.
Paint image result
mask
Freehand drawing with pen and eraser. Connect an image to paint over it, or set dimensions for a blank canvas. Outputs the painted result (RGBA) and a grayscale mask.
Text Iterator text text Split text into multiple items by delimiter. Drives multi-prompt batches.
Image Iterator images image Upload multiple images at once and batch-process the full set through connected nodes — no need to upload one-by-one.
Video Iterator videos video Upload multiple videos at once and batch-process the full set through connected nodes — no need to upload one-by-one.
Text Concatenator text × 2 text Concatenate two prompts — use one as a system prompt and the other as a user prompt.

Utility

Node Inputs Output Notes
Sticky Note Sticky note for annotations
Router any any Reroute connections for cleaner graphs (. shortcut)
Viewer image / video Display connected media at larger size with playback controls
Compare image / video Side-by-side A/B comparison
Export image / video Download connected media directly. Use the Save As dialog to choose where files are saved.

Handles & Data Types

Each node’s handle color represents its data type. Required inputs are marked with an asterisk on the label (e.g., Source *).

Type colors

Type Color Examples
Image ● Green SwitchX reference, Extract Frame output
Video ● Orange SwitchX source / alpha / output, Video Matte output
Text ● Blue LLM output, prompt inputs

Keyboard Shortcuts

Canvas

Shortcut Action
V Switch to Pointer tool
H Switch to Hand (pan) tool
/ or Tab Open the Add Node menu at the cursor
. Add a Router node at the cursor
T Add a Sticky Note at the cursor
Q Toggle the Job Queue panel
\ Toggle the Viewport panel
1 Connect selected node to Viewer and open Viewport
Space Hold to pan; on selected video nodes, toggles playback (multi-select plays in sync)

View

Shortcut Action
F Fit view to all nodes
Cmd/Ctrl + 0 Zoom to 100%
Cmd/Ctrl + 1 Fit all nodes
Cmd/Ctrl + 2 Fit selected nodes
Cmd/Ctrl + = / Cmd/Ctrl + - Zoom in / out

Selection

Shortcut Action
Esc Deselect all nodes and edges
Cmd/Ctrl + A Select all nodes
Cmd/Ctrl + C / Cmd/Ctrl + V Copy / Paste (preserves output history)
Cmd/Ctrl + Shift + V Paste without output history
Cmd/Ctrl + D Duplicate selected nodes with their output content
Cmd/Ctrl + G Group / Ungroup selected nodes
Delete / Backspace Delete selected nodes and edges

Alignment

Shortcut Action
Alt/Opt + A Align selected nodes to the left
Alt/Opt + D Align selected nodes to the right
Alt/Opt + W Align selected nodes to the top
Alt/Opt + S Align selected nodes to the bottom
Alt/Opt + H Align selected nodes to horizontal center
Alt/Opt + V Align selected nodes to vertical center
Cmd/Ctrl + Alt/Opt + H Distribute selected nodes with equal horizontal spacing
Cmd/Ctrl + Alt/Opt + V Distribute selected nodes with equal vertical spacing

General

Shortcut Action
Cmd/Ctrl + S Save
Cmd/Ctrl + Z / Cmd/Ctrl + Shift + Z Undo / Redo
Cmd/Ctrl + , Open Settings dialog
Shift + ? Open the Keyboard Shortcuts dialog