seek://map/dsh-image-gen
dsh-image-gen
#image-generation #studio #comfyui #byok
TL;DR
A native AI image studio for DSH: chat generation with iterative editing, a batch Studio workbench, multi-model comparison, a prompt-inspiration library and gallery management — BYOK across Gemini, OpenAI Images, Seedream, DashScope plus a local ComfyUI route.
Image generation on DSH has mostly been point tools; this plugin ships the whole studio loop — chat-based generation and iterative editing, a batch workbench, a prompt-inspiration library, a searchable gallery, and multi-model comparison — under a BYOK model that spans Google Gemini, OpenAI Images/Compatible, ByteDance Seedream, Aliyun DashScope and a local ComfyUI route for private workflows. The README's provider support table is refreshingly precise about what each backend can and cannot do (local ComfyUI, for instance, covers chat generation only), and results are stored isolated per workspace.
Facts
- install
- dsh plugin --profile web add dsh-image-gen@latest
- license
- Apache-2.0
- note
- 据其 README：Studio 与多模型对比目前仅支持云端 Provider，本地 ComfyUI 仅覆盖对话生图与单图编辑；环境要求为 DeepSeek Harness 稳定版本与 Node.js 22.19+/24+。
Key points
- Covers the whole image workflow in a single install: chat generation with iterative editing, a Studio batch workbench, a prompt-inspiration library and a workspace-isolated gallery
- BYOK across providers: Google Gemini, OpenAI Images/Compatible, ByteDance Seedream and Aliyun DashScope, plus a local ComfyUI route for private workflows
- Multi-model comparison runs one prompt across configured cloud providers; per the README's support table, local ComfyUI covers chat generation and single-image editing only
- The provider table ships default models and endpoints — gemini-3.1-flash-image and gpt-image-2 among them — overridable in settings
- Distributed on npm under the MIT license with CI badges; generated results are stored isolated per workspace
FAQ
Do I need to bring my own API key?
Yes — the plugin is BYOK: pick a provider in settings, enter your own API key, and adjust models and endpoints as needed; local ComfyUI additionally needs a service address reachable from the DSH host and an API-format workflow JSON import.
Do Studio and multi-model compare work with local ComfyUI?
Not yet. Per the README's support table, local ComfyUI currently covers chat generation and single-image editing only; Studio and multi-model comparison are cloud-provider features.
How do I install and configure it?
Run the README's plugin-install command in the DSH project root, restart, then complete provider setup under Settings → Plugins → Plugin configuration → Image generation.
Official references