插件 - Blender - v0.10.0 - 在 Blender 内实现 AI 图像、视频、3D 和音频

Scenario for Blender 0.10.0 is the first supported release of our Blender extension. One add-on for Blender 5.0, 5.1 and 5.2 brings Scenario's image, video, 3D, material and audio models into the viewport. It takes its inputs from your scene and puts results back where you work. It installs and updates natively from Blender, with a full handbook at blender.scenario.com.

Install in Blender

Native install and updates: in Edit › Preferences › Get Extensions › Repositories, choose Add Remote Repository and enter https://blender.scenario.com/repo/index.json. Then install Scenario and paste your API key and secret. Later releases arrive through Blender's own extension updates.
Verified downloads: each release also ships as a ZIP with SHA-256 checksums and a build provenance attestation. It is validated on Blender 5.0, 5.1 and 5.2 before it is published.

Generate from your scene

Every creation lane in one tab: Image, Video, 3D, Materials (PBR texture sets), Audio (music, speech and sound effects) and Environment (360° panoramas).
Render Image and Render Video: turn a viewport or camera capture into a finished still or clip. A saved image can become the first frame of a video.
Scene-aware inputs: use a viewport capture, the scene camera, a playblast of your animation, the selected mesh, a local file, or an asset from your Scenario Library as a reference.
Your trained models: pick your LoRAs and compositions in the model picker, set their strengths, and save project defaults.
Blockout: describe a scene in words and get a reviewed greybox layout you can rebuild, refine, or clear.

Results land where you work

Images: come back as textures, image planes, or references for the next generation.
3D models: import from GLB, OBJ, FBX, glTF, and PLY, keeping rigs and animation. Gaussian splats arrive as point clouds.
Edit the meshes you already have: remesh, retexture, UV unwrap, rig, animate, or split into parts, then bring the result in next to the original or onto it, with native Undo.
Materials: apply saved PBR texture sets to the material slot you choose.
Panoramas: set a JPEG, PNG, or EXR panorama as your World. Your original World stays untouched.
Video and audio: drop results onto the Video Sequencer, with Play, waveform, and preview controls.

Know the cost, keep every job

Exact quotes: see the server's Creative Unit estimate before every paid run, from the sidebar, the composer, or an agent.
Durable jobs: generations survive closing the panel or restarting Blender, and finished results download and verify before they touch your scene.
Recovery and cloud history: Project history lists generations made in the web app, by agents, or on another machine, and lets you bring them back into Blender.

Studio, Library and workflows

Compact composer: a floating prompt pill in every viewport, with a job strip that follows the current generation.
Studio view: an expanded workspace for detailed settings, results and jobs.
Library: browse your Scenario assets, reuse them as references, and organize them into collections and tags.
Workflows: run your Scenario workflows from Blender. Upload local files and captures into their inputs, and approve the steps that wait for review.

Film (Experimental)

From shots to a finished cut: plan shots, capture them, generate takes, review them, and assemble an editable timeline.
Export and share: export an MP4 to the folder you approve, or a portable case-study bundle of the production.

Built for AI agents

Local MCP server: Claude Code, Codex, Cursor, Claude Desktop, or any MCP client can read your scene, plan camera moves, run Blender tools, and generate with Scenario.
Same safety as the UI: agents get the same quotes and approvals as you do. The server is authenticated and local, and Python execution stays off unless you enable it.

Requirements: Blender 5.0 or later, a Scenario account with an API key and secret, and Allow Online Access turned on in Blender. Some lanes are marked Experimental; the handbook lists the known limitations.