DAELAB · 旧分镜批量（已停用）

This is the rare node article where the most useful sentence I can write is: don't try to use it.

If you found this page because a workflow throws 旧 LibTV 分镜批量入口已停用；请将表格导入生成节点，使用 Comfy 积分运行 - that's not a broken install, a missing dependency, or a version mismatch. That string is hard-coded. The node still registers, still shows its widgets, and raises that ValueError the instant it executes.

What happened

The pack used to batch a storyboard through LibTV: one row per shot, one video generation each, billed in LibTV credits. As of the 2026-10-22 update, all new cloud generation in this library runs on ComfyUI account credits instead. LibTV's channel was retired for cloud tasks. So the batch entry point was left in place as a signpost rather than deleted.

Look at what the class actually says about itself. The registered display name is no longer DAELAB · LibTV 分镜批量生成 - it's 旧分镜批量（已停用）, and the category moved from DAELab/LibTV to DAELab/Legacy. Its docstring-level description reads, in translation: the old LibTV entry is disabled; import your table into the generation node and run on Comfy credits.

The execute body is one line. That's it.

Why it still exists at all

Backwards compatibility of the graph. Node IDs are a stable contract in this pack, and deleting an ID breaks every workflow that referenced it. So the inputs remain what they always were - storyboard_json, project_uuid, request_id, model (defaulting to Seedance 2.0), mode (mixed2video), duration, resolution, ratio, sound - and the outputs are still batch_report (STRING) and video_paths (STRING).

The practical consequence is important: old workflows load. They don't fail to resolve a node, they don't lose their wiring, they don't turn into red boxes. They just fail at run time, once, with a message that tells you where to go. That's a much better failure mode than a missing-node error at load, and it's worth recognising as a deliberate design choice rather than sloppiness.

What to use instead

Two paths, both inside the same pack:

Table generation columns. The pack's multidimensional table node is now the generation surface: you lay out shots as rows, mark the reference-image columns, parse and edit the final prompt per row, and run a generation column. That's the workflow the README leads with for both images and video.

The storyboard import path. Import a script or storyboard document, then drive generation from the table it produces. The pack's docs split this into delivery scope, table/reference-column rules, and a prompt-parse plan, and there's a storyboard video template in examples/libtv/Storyboard Studio.json if you want a worked graph.

If you specifically want a single video rather than a batch, the pack still ships DAELAB.LibTV.VideoGenerate for one-off LibTV CLI generation, and the native Comfy credit generation node for cloud models. Batch-through-LibTV, as a concept, is what got retired.

Install

You don't install a retired node, but for completeness the pack installs like this:

cd ComfyUI/custom_nodes
git clone https://github.com/TeutonicMepheles/ComfyUI-DAELab-Custom-Nodes-Library.git

Restart ComfyUI and refresh the browser. Note the newer generation path expects a reasonably current ComfyUI: several nodes are V3 (comfy_api.latest), and the pack's README also says the shared canvas and upload controls moved to a separate ComfyUI-DAELab-Creative-Canvas repository - if you want the canvas panels, you install both and restart.

There are essentially zero community threads about this pack, so there's no forum folklore to sift here. The source and the pack's own docs/comfy-credit-generation.md are the ground truth, and they agree: new cloud tasks are Comfy credit only, old LibTV tasks are not resurrected as paid Comfy tasks, and you must deliberately create a new task.

The one useful debugging note

If you migrated a workflow and some LibTV nodes still execute while this one doesn't, that's expected - the single-generation VideoGenerate node is a different code path and wasn't tombstoned the same way. Don't "fix" this node by looking at that one. Fix it by replacing this node with table-driven generation, which is what the maintainer's own notice tells you to do.