dsh-shell-canvas: A Local Text-to-Video Canvas for DeepSeek Harness
feely0208/dsh-shell-canvas
DeepWhale Canvas is a DSH plugin that turns a script plus your own assets into ready-to-post videos in both landscape and portrait, with everything staying local and offline.
dsh-shell-canvas is a local text-to-video plugin for DeepSeek Harness, built for the situation where you have a script but no footage. Paste some text, pick a few assets, and it delivers both landscape and portrait cuts in one pass, with narration, quote card, and step list compositions to choose from. Scripts, assets, and finished videos never leave your machine, no network is required, and image and voice generation run through your own endpoints with keys kept locally. The free tier renders unlimited videos with a watermark and needs no registration.
How to Install
dsh plugin --profile web add @deepwhale-cn/dsh-shell-canvas- Category
- Tools & Capabilities
- Platform
- DSH Plugin
- Author
- feely0208
- Distribution
- Plugin
dsh-shell-canvas Key Features
dsh-shell-canvas Repo Summary
What Does It Do?
DeepWhale Canvas is a DSH plugin for DeepSeek Harness that turns a piece of text into a publish-ready video. It solves the "I have a script but no footage" problem: paste a script, pick a few assets, and generate both landscape and portrait cuts in one click. Scripts, assets, and finished videos all stay on your own computer — nothing is uploaded and no network is required, so it still renders with the cable unplugged. The free tier renders unlimited videos with a watermark and needs no registration.
Core Features
- Three layouts in one run: narration, quote card, and step list, each with its own composition.
- Per-sentence visuals: assign assets and durations line by line in the storyboard so media matches subtitles one to one.
- Multiple images or a video clip in a single slot: images rotate over time, while video assets make the frame actually move.
- Per-shot layout changes: size, position, and entrance style rotate shot by shot.
- Landscape and portrait together: the same script yields both versions at once.
- Model-agnostic: connect your own image and voice endpoints with keys kept on your machine; batch rendering keeps one failure from sinking the whole run, and macOS / Windows / Linux are all supported.
How to Use This Plugin?
After enabling it in any DSH-capable host, paste a script, choose a few assets, and click once to get both landscape and portrait videos. For finer control, arrange assets and durations sentence by sentence in the storyboard and adjust size, position, and entrance style per shot; for bulk work, drop in a whole batch and a single failed item will not stop the rest. The free tier renders unlimited videos with a watermark, while the licensed tier removes the watermark and unlocks batch rendering plus per-item lookup, with a time-based license key that is not tied to a machine.
How to Troubleshoot This Plugin?
If one item fails during batch rendering, the remaining tasks still finish, so you can re-run just the failed one. If image or voice generation returns nothing, check that your local endpoint configuration and keys are valid — keys stay on your machine and are never uploaded with the finished video.
This page is an independent rewrite of the plugin's official README — for authoritative documentation and the latest changes, refer to the source: feely0208/dsh-shell-canvas. The plugin is third-party code that runs on your machine once installed; inclusion does not imply endorsement — please review the source before installing.