VC-Filename Generator I
Name the two clips you're about to stitch
- pipe
When you render video with a model like Wan, you usually save short clips - the community standard is 81 frames a pop, a couple of seconds - and then stitch pairs of them into longer takes. That stitching needs to know exactly which files to combine. This node generates that filename list, and the source comment says it plainly: "v1 is used for combine only, it automatically sets the 2nd filename (filename_suffix_start + 1), it also provides the mask settings."
So VC-Filename Generator I is purpose-built for a combine operation: it names exactly two clips - filename_prefix_suffix_start.mp4 and the next number up - and carries the frame-masking settings used to blend the seam where the two clips meet. It does not touch video. It computes paths and builds a pipe.
The settings you actually set:
- path - the folder your rendered clips live in. Raw text, no picker. Required; empty path raises an error.
- filename_prefix (default
vc) - the stem your clips were saved under. - filename_suffix_start (default 1) - the first clip's number. The second clip is automatically
+1. That's the whole range; there is no end number, by design. - file_extension (
.mp4), frame_load_cap (81) - how many frames per clip to load. 81 is the Wan default and the sensible value here. - mask_first_frames (10) / mask_last_frames (0) - mask out frames at the start and end of the clip so the combine node can blend the cut instead of hard-cutting it.
Output is a pipe that you unpack with the matching Pipe Out VC-Filename Generator I [RvTools] node - it splits into path, rel_path, frame_load_cap, mask_first_frames, mask_last_frames, and a files string that feeds straight into a VHS_VideoCombine-style node.
One deliberate design choice: the generator overrides IS_CHANGED to return NaN, which forces ComfyUI to re-execute it every single run. Your filename list refreshes each queue instead of caching stale names - important when clips get re-rendered between runs.
Installing it
Part of ComfyUI-RvTools_v2. Install once: ComfyUI Manager → search ComfyUI-RvTools_v2, or clone https://github.com/r-vage/ComfyUI-RvTools_v2 into ComfyUI/custom_nodes, restart. No models, no keys; the pack's deps (torch, numpy, Pillow, opencv-python, pilgram) are already in stock ComfyUI.
The pack is unmaintained (README → ComfyUI_Eclipse). The [RvTools] suffix is from v2's rename pass after v1's repo disappeared.
Common issues
- "Path is missing" - path is a required raw string. There's no folder browser; type it.
- Windows-flavored paths. Names are built with hardcoded
\separators and the shipped workflow usesD:\AI\output\.... It works on Windows out of the box; elsewhere, expect separator fights. - Only two names, ever. If you expected a range, that's the II version's job. I is deliberately minimal: combine needs exactly two clips.
Inputs (7)
| Name | Type | Default | Description |
|---|---|---|---|
| path | STRING | — | |
| filename_prefix | STRING | vc | — |
| filename_suffix_start | INT | 11–18446744073709550000 | — |
| file_extension | STRING | .mp4 | — |
| frame_load_cap | INT | 81 | — |
| mask_first_frames | INT | 10 | — |
| mask_last_frames | INT | 0 | — |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| pipe | pipe | — |