VAE Name

The VAE is the quiet part of every workflow: you set it once, forget it, and it's the difference between clean colors and washed-out artifacts. When you want to record which VAE you used - in a filename, a metadata line, a comparison log - the VAE loader hands you a VAE object, not a name. VAE Name (class VAEName) is the lhyNodes utility for that gap: a dropdown of every VAE in your VAE list, with the selected name as its only output.

It's the same family pattern as the pack's LoraName, UNETName, and the rest: dropdown from the model folder, string out, typed * so it plugs into any text socket. Feed it into a String Format node and your save prefix can read {model}_vae_{vae_name} - so every output file tells you what decoded it. Swap VAEs mid-comparison and the filenames keep the story straight for you.

The inputs that matter

• vae_name - the dropdown of available VAEs. One quirk worth knowing: it's built from ComfyUI's VAELoader.vae_list(), which includes the built-in pixel_space entry as well as your actual VAE files. pixel_space is ComfyUI's placeholder "no VAE" option (for models that decode in pixel space), so if you see it in the list, that's normal - it just means "no VAE selected" and isn't a file you need to hunt for.

Output: vae_name, the selected name as a string.

Where it fits

The VAE matters more than most people give it credit for - the wrong or missing VAE is a classic source of washed-out or oversaturated output, and community wisdom is to pick a good VAE and keep it consistent across a project. This node turns that consistency into record-keeping: wire it into your filename template and your outputs carry the VAE identity automatically. It's also handy in multi-VAE A/B tests, where you want the chosen VAE reflected in the save name without retyping it each run.

Like its siblings, it's a deliberately boring node - dropdown in, string out. If you never auto-name outputs, you don't need it. If you do, it's one more place where the graph knows what it's doing instead of you remembering it.

Installing it

Part of lhaoyun6/ComfyUI-lhyNodes. Install via ComfyUI Manager by searching for lhyNodes, or:

cd ComfyUI/custom_nodes
git clone https://github.com/lihaoyun6/ComfyUI-lhyNodes.git
python -m pip install -r ComfyUI-lhyNodes/requirements.txt

Restart ComfyUI. The pack's shared requirements are ultralytics, opencv-python, numpy, and yarl; no models or downloads involved.

It's the kind of node that feels pointless until the day you're staring at fifty outputs trying to remember which VAE made the good ones. Add it once, wire it into your naming, and that day never comes.