ml-sharp-web
AI
image to 3d gaussian splat
SHOW HN2026
How HN took it
3 positive, 7 neutral or asking, 1 mixed, 4 critical, out of 15 comments about the project. Thread score -0.01.
Idea: 2 for, 1 againstPerformance: 0 for, 3 againstExecution: 1 for, 0 against
The thread, comment by comment
Each row is a top-level comment and its replies, in HN's order. Select a cell to open that comment on HN.
positive neutral or mixed critical about another commenter or something else the maker
What HN said
What people liked
Impressive browser tech showcase and fast inference
From 2 comments
> inference itself is a few seconds on a recent Mac This is impressive as hell Very cool demo. It works in about ~9 seconds on my machine. A few asks if you're going to devote…
echelon on HNDid not work in Firefox on Linux, but it runs on Chrome. Have to admit, I dont get it. I tried it with 3 landscape photos I have and the results were nowhere close to the results…
javier2 on HN
Mesmerising spatial viewing of photos in VR
From 1 comment
I vibecoded a simple web app using Sharp that allowed be to quickly browse any local image folder and view them as "almost" volumetric 3d scenes in a VR headset. I precomputed…
andybak on HN
Objections
Model is about 2.4 to 2.5GB and crashes browsers or runs out of memory
From 4 comments
Loading the model crashes my browser tab from memory usage :/
herpdyderp on HNWhat are the requirements for running this? Chrome throws a whole bunch of "out of memory" errors into the console when I try to execute these. I'm guessing 4GiB of VRAM is not…
jeroenhd on HNMy laptop's GPU has 4GiB of VRAM but it still failed to allocate enough memory. Seems like I'll have to pass on demos like these until someone figures out a way to use less…
jeroenhd on HNIt froze up my computer, had to hard-boot lol (16MB M1 Macbook, Chrome)
anentropic on HN
Single photo makes up depth that is not real
From 2 comments
I don't like that it uses only a single photo. This means it is going to make up a lot of stuff. E.g. if I show it a photo of a poster, then it will make that poster 3D. With…
amelius on HNMaybe, but what is wrong with wanting real depth instead of "made up depth"? One extra photo mostly solves that.
amelius on HN
Results differ from demo; poor on some photos
From 1 comment
Did not work in Firefox on Linux, but it runs on Chrome. Have to admit, I dont get it. I tried it with 3 landscape photos I have and the results were nowhere close to the results…
javier2 on HN
Did not work in Firefox on Linux
From 1 comment
Did not work in Firefox on Linux, but it runs on Chrome. Have to admit, I dont get it. I tried it with 3 landscape photos I have and the results were nowhere close to the results…
javier2 on HN
Requests
Lower memory or compressed model for low-RAM machines
From 2 comments
Yeah, I think you need at least 8gb ram unfortunately, but I tested it only on a 32gb M2, so 8gb might also not be enough. I might create a compressed version of the model, that…
bring-shrubbery on HN2.4GB in browser is crazy. I’m curious whether quantization works here or if the model quality falls apart too much.
mattbaconz on HN
Full orbital camera and double-click drag to move camera
From 1 comment
> inference itself is a few seconds on a recent Mac This is impressive as hell Very cool demo. It works in about ~9 seconds on my machine. A few asks if you're going to devote…
echelon on HN
Questions
What are the memory requirements?
From 1 comment
What are the requirements for running this? Chrome throws a whole bunch of "out of memory" errors into the console when I try to execute these. I'm guessing 4GiB of VRAM is not…
jeroenhd on HN
Does it use WASM and WebGPU?
From 1 comment
Does it use both WASM and WebGPU?
sroussey on HN
Does quantization work without losing quality?
From 1 comment
2.4GB in browser is crazy. I’m curious whether quantization works here or if the model quality falls apart too much.
mattbaconz on HN
Alternatives named
Original 3D Gaussian Splatting, Polycam, Luma, Postshot for multi-photo
From 1 comment
For multi-photo, the go-to is still the original 3D Gaussian Splatting (Kerbl et al., 2023) — most consumer tools like Polycam, Luma, and Postshot wrap that under the hood.
bring-shrubbery on HN
HY-World 2.0 and WorldLabs Marble
From 1 comment
> inference itself is a few seconds on a recent Mac This is impressive as hell Very cool demo. It works in about ~9 seconds on my machine. A few asks if you're going to devote…
echelon on HN