Hunyuan Batch Model Preview

What it is

Hunyuan Batch Model Preview is the browsing node for batch output - the counterpart to Hunyuan 3D Model Preview, but built for the case where Hunyuan Batch Image to 3D just handed you back ten GLBs at once instead of one. Rather than wiring up ten separate preview nodes, you point this one at the whole batch result and pick which model to look at by index.

It's a small, quality-of-life node in a pack that's otherwise about API calls and file paths - this is the one that makes a batch result actually usable inside the graph instead of just a list of paths you have to go find manually.

How it works

It takes the model_paths string a batch generation produced - the list of every GLB path from that run - and, using model_index, picks out one entry to load and display. Change the index, re-run, and you step through the batch one model at a time without touching the generation nodes again.

Inputs and outputs

• model_paths - the string of GLB paths from a Hunyuan Batch Image to 3D node's model_paths output. This is the whole batch, not a single file.
• model_index - which entry in that list to preview, 0 to 99 (default 0). Bump it up to page through the batch; if your batch produced fewer models than the index you set, that's the first thing to check if nothing loads.

Output is a single model_path - the one GLB path selected by the index, in the same string-passthrough style the single-image preview node uses, so you can chain it into anything else downstream that expects a path.

Install

Via ComfyUI Manager: search "Hunyuan 3D", install, restart. Manually:

cd ComfyUI/custom_nodes
git clone https://github.com/exedesign/Hunyuan-3D-v3.git
cd Hunyuan-3D-v3
pip install -r requirements.txt

Like the single-model preview node, this one doesn't talk to the network or need a Hunyuan Config - it's purely reading from the file paths a batch run already produced, so there's nothing to authenticate here.

Troubleshooting

If the preview comes up empty, the two things to check are whether model_index is actually within range for the batch you ran (an index of 7 does nothing useful against a batch of 5), and whether the upstream batch actually completed for that particular image - check the batch node's batch_summary output first, since a partial batch (say, one that ran out of Tencent Cloud credits partway through) will have gaps where some images simply never produced a model to preview. This node can only show you what the batch node actually finished; it has no way to distinguish "still processing" from "never going to finish."