Show HN: Run TRELLIS.2 Image-to-3D generation natively on Apple Silicon

4/20/2026 at 1:59:48 PM

solid port. the sdpa swap for sparse attention — did you notice a meaningful quality difference, or is it basically equivalent to the cuda version? curious if the pure-pytorch path added any noticeable latency hit on the m3 max

by paolatauru

4/20/2026 at 7:26:06 AM

This is the kind of porting work I always hope for when I see a CUDA-only release. Have you thought about publishing the gather-scatter sparse 3D convolution and SDPA attention swaps as a standalone toolkit or writeup? A lot of folks running models locally on Apple Silicon hit the same wall with flash_attn, nvdiffrast, and custom sparse kernels and end up redoing the same work.

by sebakubisz

4/20/2026 at 8:33:02 AM

that makes so much sense...I am exploring if I can find someone who has done this well...If not I'll try to do it myself.

by shivampkumar

4/20/2026 at 8:58:48 AM

Most 'runs on Mac' ports are a wrapper around a cloud call or a quantized shell of the original model. Going after the CUDA-specific kernels with pure-PyTorch alternatives is the kind of work that ages well, because the next CUDA-locked research release is three weeks away. One question: how much of the gather-scatter sparse conv is reusable for other TRELLIS-like architectures, or is it bespoke to this one?

by sergiopreira

4/20/2026 at 11:54:52 AM

Cool project. I've been working on something similar in spirit — a personal video cloud (strimoza.com) — and the hardest part was also getting local playback to work reliably without internet. How are you handling memory pressure on the M-series chips with larger models?

by strimoza

4/20/2026 at 1:02:17 PM

I wonder how these outputs compare between Apple Sharp

https://github.com/apple/ml-sharp

No matter what it is cool seeing so much them work on different devices

by pcoyne

4/20/2026 at 2:32:49 AM

Nice work. Although this model is not very good, I tried a lot of different image-to-3d models, the one from meshy.ai is the best, trellis is in the useless tier, really hope there could be some good open source models in this domain.

by gondar

4/20/2026 at 3:03:56 AM

Hey, thanks for sharing this. I'm sure TRELLIS.2 definitely has room to improve, especially on texturing.

From what I've seen personally, and community benchmarks, it does fair on geometry and visual fidelity among open-source options, but I agree it's not perfect for every use case.

Meshy is solid, I used it to print my girlfriend a mini 3d model of her on her birthday last year!

Though worth noting it's a paid service, and free tier has usage limitations while TRELLIS.2 is MIT licensed with unlimited local generation. Different tradeoffs for different workflows. Hopefully the open-source side keeps improving.

by shivampkumar

4/20/2026 at 6:35:38 AM

Meshy is indeed great but I am terminally put off by their alltogether terrible, sleazy, gamified, opaque web UI. It's like aliexpress and a lootbox game had a baby that's into mesh generation. Ugh.

by isoprophlex

4/20/2026 at 10:41:39 AM

[dead]

by jseabra

4/20/2026 at 7:12:00 AM

This is fantastic, great work. I will attempt to run it on my 16GB M1 but I doubt it'll run.

Out of curiosity, how did you go about replacing the CUDA specific ops? Any resources you relied on or just experience? Would love to learn more.

by petargyurov

4/20/2026 at 2:40:50 AM

So much effort, but no examples in the landing page.

by kennyloginz

4/20/2026 at 3:05:47 AM

You're right, thanks for flagging this, let me run something and push images

by shivampkumar

4/20/2026 at 3:57:40 AM

added! will add more, maybe even a GIF

by shivampkumar

4/20/2026 at 9:24:06 AM

Great. Potentially can go much faster rewriting it in terms of Metal shaders.

by antirez

4/20/2026 at 4:04:31 AM

How much RAM does this use? Only sitting on 8 GB right now, I'm trying to figure out if I should buy 24 GB when it's time for a replacement or spring for 32.

by post-it

4/20/2026 at 4:45:46 AM

The model needed about 15GB at peak during generation - the 4B model loads multiple sub-models (1.3B each for shape and texture flow). 8GB won't be enough, but both 24GB and 32GB both should be fine.

by shivampkumar

4/20/2026 at 5:07:54 AM

Thanks! Could it conceivably load the sub-models in series rather than parallel? 8 still won't be enough but I wonder if those with 16 could eke something out.

by post-it

4/20/2026 at 4:30:14 AM

[dead]

by Serhii-Set

4/20/2026 at 10:57:59 AM

Does it support multi-view input?

by drbscl

4/20/2026 at 1:34:25 AM

That’s always been possible with MPS backend, the reason people choose to omit it in HF spaces/demos is that HF doesn’t offer an MPS backend. People would rather have the thing work at best speeds than 10x worse speeds just for compatibility.

by villgax

4/20/2026 at 3:22:33 AM

IMO TRELLIS.2 is slightly different case from the HF models scenario. It depends on five compiled CUDA-only extensions -- flex_gemm for sparse convolution, flash_attn, o_voxel for CUDA hashmap ops, cumesh for mesh processing, and nvdiffrast for differentiable rasterization. These aren't PyTorch ops that fall back to MPS -- they're custom C++/CUDA kernels. The upstream setup.sh literally exits with "No supported GPU found" if nvidia-smi isn't present. The only reason I picked this up because I thought it was cool and no one was working on this open issue for Silicon back then (github.com/microsoft/TRELLIS.2/issues/74) requesting non-CUDA support.

by shivampkumar

4/20/2026 at 2:09:53 AM

Are you saying the original one worked with MPS? Or are you just saying it was always theoretically possible to build what OP posted?

by Reubend

4/20/2026 at 12:09:41 PM

Latter

by villgax

4/20/2026 at 1:54:06 AM

It’s always been possible, but it’s not possible because there’s no backend, and no one wants to it to be possible because everyone needs it 10x the speed of running on a Mac? I’m missing something, I think.

by refulgentis

4/20/2026 at 3:26:17 AM

I thought it was cool and then I found the open issue mentioned above, that convinced me its def something more people want.

It IS significantly slower, about 3.5 minutes on my MacBook vs seconds on an H100. That's partly the pure-PyTorch backend overhead and partly just the hardware difference.

For my use case the tradeoff works -- iterate locally without paying for cloud GPUs or waiting in queues.

by shivampkumar

4/20/2026 at 3:15:11 AM

Well done

by jmatthews

4/20/2026 at 3:15:56 AM

rad. how long does output take? trellis is a fun model.

by serf

4/20/2026 at 3:37:06 AM

i was able to get it in 3.5 mins from a single image on my 24gb m4 pro macbook

I'm still working on this to try to replicate nvdiffrast better. Found an open source port, might look it tonight

by shivampkumar

4/20/2026 at 3:32:15 AM

thanks!

by shivampkumar

4/20/2026 at 11:30:34 AM

[dead]

by takahitoyoneda

4/20/2026 at 8:04:20 AM

[dead]

by techpulselab

4/20/2026 at 7:03:24 AM

[dead]

by jiexiang

4/20/2026 at 6:58:13 AM

[dead]

by vrr044

4/20/2026 at 2:52:10 AM

[flagged]

by hank808

4/20/2026 at 4:22:45 AM

Sunday night, and its kinda cool idk man

by refulgentis

4/20/2026 at 3:27:53 AM

I mean I can see that it's niche. Did not expect so many upvotes, but ig it's less niche than I tought

If you're not working with 3D on Apple Silicon this isn't relevant to you. For the subset of people who are, running this 4B parameter 3D generation model locally on a Mac was previously blocked by hard CUDA dependencies with no workaround.

by shivampkumar

4/20/2026 at 4:27:03 AM

Right but it is at most a couple of hours with claude code and posted on Sunday night.

by svnt

4/20/2026 at 11:03:29 AM

Exactly, I know because I did the same thing!

by atultw

4/20/2026 at 2:57:26 AM

Good question.

by kennyloginz