连环画HTML浏览导出

Once you've generated a comic with 豆包连环画创作, this is how you turn it into something you can actually read - a single interactive HTML file with flip-through pages, narration, and animations. 连环画HTML浏览导出 takes your IMAGE batch and the story text and bakes them into one .html document. No server, no dependencies, no web skills required: the file it produces is completely self-contained and shareable as-is.

How it works

The node reads your image tensor batch, encodes each page as a base64 data URI, and merges that with the story text into one HTML document. Because the images are embedded (not linked to files on disk), the result is a single file that works from any browser, on any machine, forever - this is the "send it to a client / post it somewhere" move. It writes out to your ComfyUI output directory (falling back to the pack's web/exports folder), with a timestamp prefix so re-runs don't clobber each other, and returns the full path.

The pages get the classic comic layout: image on the left, narration text on the right, with prev/next buttons, a page counter, and keyboard arrow navigation. Everything's responsive, and the default page-turn transition gives it a book feel.

Inputs that matter

• images (required) - the comic panel batch, straight from DoubaoComicBookNode.comic_images.
• story_structure (required) - the text. Wire in DoubaoComicBookNode.story_text or story_structure; both are accepted, and the node parses either the JSON structure or plain formatted text. The README notes that story_text is the recommended feed.
• title - shows in the page header. Default "我的连环画".
• output_filename - the HTML file name; gets a timestamp prefix in the actual path.
• transition_effect - page-turn, fade, slide, flip, or none.
• transition_duration - 0.1–3.0s, default 1.5.

Outputs are viewer_path (the absolute path to the HTML file - feed this into the pack's browser preview or in-canvas preview node) and summary (a status string with page count).

Install & setup

This is part of ComfyUI_doubao_seed, so install the pack once - ComfyUI Manager (search "Doubao") or:

cd ComfyUI/custom_nodes
git clone https://github.com/xuchenxu168/ComfyUI_doubao_seed
cd ComfyUI_doubao_seed
pip install -r requirements.txt

Restart ComfyUI and it appears under Ken-Chen/Doubao. No API key needed for this node - it's pure local HTML assembly, so it works even if the generation nodes are failing. Worth knowing if you're debugging the API side: you can still preview images you already have.

Gotchas

If you point the in-canvas preview (ComicHTMLPreviewNode) at this and see a blank window, the README's #1 troubleshooting entry is about that exact failure: the front-end WEB_DIRECTORY didn't load or the browser cached an old version. Hard-refresh (Ctrl+F5) and confirm web/comic_html_preview.js exists in the pack. One more: a 4K, 12-scene comic embeds a lot of base64, so the HTML file gets heavy - fine for sharing, less fine for loading in a low-end phone browser. If that's your audience, export the pages as images instead with the batch exporter.