15:[[47828896,{"by":"shivampkumar","descendants":40,"id":47828896,"kids":[47831308,47831800,47829794,47829384,47829724,47832999,47833688,47831235,47830291,47832578,47834479,47831979,47851658,47830015,47831513,47831182,47832801,47843907,47868775,47868771,47831149,47829869],"score":202,"text":"I ported Microsoft's TRELLIS.2 (4B parameter image-to-3D model) to run on Apple Silicon via PyTorch MPS. The original requires CUDA with flash_attn, nvdiffrast, and custom sparse convolution kernels: none of which work on Mac.

I replaced the CUDA-specific ops with pure-PyTorch alternatives: a gather-scatter sparse 3D convolution, SDPA attention for sparse transformers, and a Python-based mesh extraction replacing CUDA hashmap operations. Total changes are a few hundred lines across 9 files.

Generates ~400K vertex meshes from single photos in about 3.5 minutes on M4 Pro (24GB). Not as fast as H100 (where it takes seconds), but it works offline with no cloud dependency.

https://github.com/shivampkumar/trellis-mac","time":1776643636,"title":"Show HN: Run TRELLIS.2 Image-to-3D generation natively on Apple Silicon","type":"story","url":"https://github.com/shivampkumar/trellis-mac"}],[47291574,{"by":"stagas","descendants":94,"id":47291574,"kids":[47335880,47334229,47338971,47333403,47336292,47334266,47334677,47333709,47334623,47338290,47333794,47333858,47337096,47336927,47333768,47335423,47334417,47335909,47342924,47336903,47337203,47344653,47334274,47333721,47336702,47346313,47334998,47333991],"score":228,"time":1772918291,"title":"Building a TB-303 from Scratch","type":"story","url":"https://loopmaster.xyz/tutorials/tb303-from-scratch"}],[47302637,{"by":"Lusrodri","descendants":2,"id":47302637,"kids":[47304174],"score":3,"text":"I got tired of watermark removers that either:\n - Ask you to create an account just to try\n - Show you the result and then lock it behind a paywall\n - Bombard you with ads

So I built Goodbye Watermark. Upload an image, watermark is gone, download it. That's it.

It's free. No signup. No tricks.

https://goodbyewatermark.com

Would love brutal feedback — especially on quality of results vs. other tools you've tried.","time":1773011477,"title":"Show HN: I built a free AI watermark remover – no signup, no ads, no bullshit","type":"story"}],[47183793,{"by":"woodybury","descendants":9,"id":47183793,"kids":[47183843,47223113,47183794,47214027,47214088,47183965],"score":35,"time":1772216977,"title":"Running Neural Amp Modeler on embedded hardware","type":"story","url":"https://www.tone3000.com/blog/running-nam-on-embedded-hardware"}],[47158975,{"by":"iamskeole","descendants":671,"id":47158975,"kids":[47161477,47162295,47162032,47165755,47161447,47164103,47161629,47161626,47161271,47162505,47162223,47163499,47161420,47170159,47164318,47161596,47169004,47164009,47161945,47161876,47161816,47163770,47165971,47162095,47162967,47180225,47161470,47180209,47163630,47162066,47161250,47162033,47163296,47162392,47181610,47160757,47162908,47177761,47161517,47164096,47160992,47161715,47163407,47164891,47172858,47166925,47161464,47161501,47163796,47166771,47164213,47161311,47164890,47164146,47161168,47162507,47162182,47160690,47162598,47165634,47163090,47163508,47177679,47161238,47161653,47164116,47167016,47168910,47161310,47163833],"score":482,"time":1772058565,"title":"How will OpenAI compete?","type":"story","url":"https://www.ben-evans.com/benedictevans/2026/2/19/how-will-openai-compete-nkg2x"}],[47157224,{"by":"tintinnabula","descendants":251,"id":47157224,"kids":[47158285,47160325,47157756,47161622,47160559,47168410,47158114,47159298,47161845,47157644,47159412,47161531,47158023,47157946,47159115,47159623,47158169,47177616,47157627,47162529,47184517,47171620,47168164,47160211,47162501,47170935,47163179,47170207,47162285,47157564,47165850,47158219,47166585,47163593,47158583,47158034,47160081,47158164,47158302,47170021,47157269,47157636],"score":678,"time":1772050607,"title":"Jimi Hendrix was a systems engineer","type":"story","url":"https://spectrum.ieee.org/jimi-hendrix-systems-engineer"}]]

user:	kennyloginz
created:	August 2, 2022
karma:	275