ComfyUI Trellis: One-Click 3D Workflow

I’ve been messing around with ComfyUI Trellis lately, and honestly, it’s one of those tools that just works. No endless tweaking, no convoluted steps—just drag, drop, and get a 3D model. Here’s how it went for me.

Trellis takes text or images and spits out 3D assets—meshes, radiance fields, the whole deal. It’s not magic, but it’s close. I threw a few prompts at it, and the results were surprisingly solid. Not perfect, but way better than I expected for something this straightforward.

First, you’ll need a decent GPU. I used an NVIDIA A100, but anything with 16GB VRAM should handle it. The docs say Windows works best, but Linux might too if you’re feeling adventurous.

I cloned the repo into my ComfyUI custom_nodes folder—standard stuff. The tricky part was the extra dependencies. Things like diffoctreerast and nvdiffrast took a minute to install, but the GitHub README walks you through it. If you’ve set up ComfyUI before, this’ll feel familiar.

After installation, I dragged the Trellis node into my canvas. No fancy configuration—just plugged in a text prompt and hit run. The first try was… slow. Like, “make coffee” slow. But the output? A fully textured 3D model with zero manual tweaking.

I tried a few more prompts, and the consistency was hit-or-miss. Simple shapes worked great; complex scenes needed more refinement. Still, for a one-click tool, it’s impressive. If you’re curious, the Trellis GitHub has the full setup details.

cd ComfyUI/custom_nodes

Most 3D tools feel like solving a puzzle blindfolded. Trellis removes the guesswork. It’s not replacing Blender, but for quick prototypes or brainstorming, it’s stupidly convenient. The outputs aren’t studio-quality, but they’re good enough to build on.

git clone --recurse-submodules https://github.com/if-ai/ComfyUI-IF_Trellis.git

pip install git+https://github.com/EasternJournalist/utils3d.git@9a4eb15e4021b67b12c460c7057d642626897ec8
New-Item -ItemType Directory -Force -Path C:\tmp\extensions
git clone --recurse-submodules https://github.com/JeffreyXiang/diffoctreerast.git
C:\tmp\extensions\diffoctreerast
pip install C:\tmp\extensions\diffoctreerast
git clone https://github.com/autonomousvision/mip-splatting.git C:\tmp\extensions\mip-splatting
pip install C:\tmp\extensions\mip-splatting\submodules\diff-gaussian-rasterization\
pip install kaolin -f https://nvidia-kaolin.s3.us-east-2.amazonaws.com/torch-2.4.0_cu121.html
git clone https://github.com/NVlabs/nvdiffrast.git C:\tmp\extensions\nvdiffrast
pip install C:\tmp\extensions\nvdiffrast

Biggest downside? The VRAM hunger. My A100 choked on high-res outputs, so temper your expectations if you’re on consumer hardware.