VELVET VICE LTX — v0.1.9 Verified Auto NVENC Output

If you open the node list for velvet-vice-ltx you'll see four nodes named VelvetViceLTXAutoVideoCombine plus a version number. That looks like clutter, but it's deliberate: the Velvet Vice LTX workflow is released in versioned drops, and each drop pins its own final-output node so that a behavior change in a future pack version can never silently alter what an older workflow produces. VelvetViceLTXAutoVideoCombineV019 is the one the v0.1.9 workflow uses.

What's different about v0.1.9

Mechanically, almost nothing. It's a subclass of the base Auto NVENC output node with an implementation version flag. The v0.1.9 description highlights the three guarantees that version locked in: it encodes the selected stream exactly once (no double-encode), it uses a Windows-safe output path (invalid filename characters scrubbed), and it probes NVENC with a real 256×256 test frame before choosing hardware or CPU encoding.

So everything that applies to the base node applies here: it delegates to VideoHelperSuite's VHS_VideoCombine, AUTO mode probes h264_nvenc and falls back to CPU H.264 (CRF), NVIDIA NVENC ONLY raises if the probe fails, and FP16 frames get the fast RGB8 conversion path.

Inputs are identical to the base node: images, frame_rate, loop_count, filename_prefix (video/LTX23-FINAL), encoder_mode, nvenc_bitrate_mbps_at_24fps (60), cpu_crf (15), pix_fmt (yuv420p), pingpong, save_metadata, trim_to_audio, save_output, and optional audio. One output, Filenames (VHS_FILENAMES).

Which one should you use?

If your workflow file came from the v0.1.9 release, use this node - that's the point of pinning. If you're building fresh, use the current default (v0.1.12) or the plain VelvetViceLTXAutoVideoCombine, since the version-bound variants exist to preserve old behavior, not to be the forward-looking choice. The base node and v0.1.12 behave the same today; the pinned ones only diverge if the pack changes behavior later.

Installing it

Part of VELVET VICE - LTX (velvet-vice-ltx), via ComfyUI Manager or:

cd ComfyUI/custom_nodes
git clone https://github.com/Velvet-Vice/velvet-vice-ltx

Restart, hard-refresh with Ctrl+F5. No pip deps in the pack, but VideoHelperSuite and FFmpeg are required at runtime - the node raises a clear "VideoHelperSuite is required" error if it's missing.

Troubleshooting

• NVENC ONLY errors - the probe is honest; your FFmpeg or driver lacks h264_nvenc. Use AUTO or CPU.
• Missing node on workflow load - you likely merged an old Velvet Vice folder into a new release. The README is blunt: never do that; reinstall clean.

One final note: the four versions differ only in which behavior the workflow author locked in. When someone on the internet complains "the video node changed," this is the mechanism - the pin, not the code, is what protects you.