TK Text Saver
TK Text Saver does nothing now — and that's fine
- texts
- filenames
Straight to it: TK Text Saver is a no-op. It takes a texts LIST and a filenames LIST, and it does absolutely nothing with them. If you open the source, the node's own comment says it plainly - "remains for workflow compatibility but performs no action." The real saving now happens inside the batch interrogator nodes, which write their .txt files as they go.
How it got this way
This pack originally had a two-step flow: batch loader → interrogator → Text Saver, with Text Saver writing the caption files. Then the author moved saving into the interrogators themselves, so every image you process gets its caption .txt written automatically - no extra node required. The old Text Saver stayed in the pack so existing workflows don't hard-fail when they load, and it's still in the README's worked example as step three. Don't follow that step; it's vestigial.
If your graph already has this node in it, leave it alone or delete it - either way the captions land on disk, because the interrogator wrote them before this node ever ran.
What it looks like
It's an output node (no outputs, and it marks itself is_output_node). Its inputs are:
- texts - LIST of caption strings.
- filenames - LIST of filenames, position-matched to
texts.
Both come straight from the texts and filenames outputs of TK QwenVL Interrogator, TK JoyCaption Interrogator, or TK WD14 Tagger. Wire them in and nothing happens - which is the entire point of this page.
Installation
If you somehow have a workflow that insists on the node existing, install the pack normally:
cd ComfyUI/custom_nodes
git clone https://github.com/tackcrypto1031/tk_comfyui_imageVL.git
cd tk_comfyui_imageVL
pip install -r requirements.txt
Restart ComfyUI, or use ComfyUI Manager and search tk_comfyui_imageVL.
The practical take
Think of this node as a compatibility stub. It isn't broken, it isn't going to save anything, and you shouldn't add it to a fresh graph. The thing that actually persists your captions is the interrogator you already used - it saves a {name}.txt next to each image in the batch's output folder. If you need the texts somewhere else, use a normal string/list save node from any of the big utility packs; TK Text Saver has nothing to offer that the interrogators don't already do.
Short version: if you're wondering why your Text Saver isn't writing files, it's because it was never going to. Check the output folder the interrogator was pointed at - your captions are already there.
Inputs (2)
| Name | Type | Default | Description |
|---|---|---|---|
| texts | LIST | — | |
| filenames | LIST | — |
Outputs (0)
No outputs