Canvas & Nodes
Deep dive into infinite canvas interaction: node creation and layout, multimodal model matrices (text, image, video, audio, 3D World), and multi-way transmission.
AZ8.Studio centers on the Infinite Canvas, shattering the constraints of single-column chat windows. Like a director laying out storyboards across a production whiteboard, you can freely place and connect text, image, video, audio, and 3D World nodes to assemble a visual, reproducible cinematic pipeline.
1. Canvas Workspace & Creating Nodes
In the infinite canvas, pan by dragging with the middle or right mouse button, and zoom smoothly with the scroll wheel. Create new nodes via:
- Sidebar Tools: Click modal icons (text, image, video, audio, 3D) on the floating left toolbar.
- Right-Click Canvas Menu: Right-click anywhere on the canvas to open the quick-create radial menu.
Infinite Canvas Workspace
Right-Click Node Creation Menu
2. Modality Nodes Overview
| Node Type | Key Operations | Typical Creative Use |
|---|---|---|
| Text Node | Duplicate, delete, multi-turn context piping | Script outlines, shot breakdown, prompt inversion, character lore |
| Image Node | Duplicate, delete, download, inpainting, save to asset library | Concept frames, character design, props, start/end frames |
| Video Node | Duplicate, delete, download, upscaling (1080p/2K), keyframe extraction | Dynamic shots, camera moves, panoramic video generation |
| Audio Node | Duplicate, delete, waveform audition, multi-lingual voiceover | Mood music, sound effects, character dialogue tracks |
| 3D World Node | Duplicate, delete, 3D exploration, multi-angle perspective capture | Camera reference baseline, spatial consistency anchor |
3. Multimodal Model Matrix
In AZ8, all node input fields support in-place model switching, allowing you to rapidly test different model aesthetics against the same prompt:
Text Node
- Leading Models: Gemini 3 Pro, GPT-5
- Key Features: Direct multimodal input (text, images, video, PDF); script structure breakdowns, shot dialog writing, and multilingual adaptation.
Initialized Text Node
Multi-turn Script Progression Output
Image Node
- Leading Models: Nano Banana 2, Seedream 5.0 Lite, GPT-image-2
- Key Features: Text-to-image and image-to-image; precision inpainting, multi-angle rotation, and 25-frame storyboard grid expansions.
Initialized Image Node
Floating Action Menu & Model Tuning
Video Node
- Leading Models: Sora 2, Seedance 2.0 Fast / 2.0, Kling 3.0 Pro / O3 Pro, Wan 2.6
- Key Features: Text-to-video, first/last frame constraints, multimodal driving; built-in upscaling to 1080p and 2K, with one-click keyframe extraction back to canvas.
Initialized Video Node
Motion Controls & Parameters
Audio Node
- Leading Models: Suno V5, MiniMax Audio, MiniMax Music
- Key Features: Instrumental background tracks, cinematic scores, multilingual voiceovers, and fine-grained duration controls.
Initialized Audio Node
Waveform Audition & Track Management
3D World Node
- Core Engine: MarbleWorld
- Key Features: Build 3D environments from text prompts or concept images. Guarantees physical and spatial consistency across large scenes and multi-character shots, serving as a virtual camera perspective source for downstream workflows.
1. Empty 3D World Node
2. Natural Language 3D Generation
3. Node Parameters & Viewport
Result: Text-Prompted 3D Space
Result: Image-Reconstructed 3D Space
