bring-shrubbery

Thanks for feedback! I'll fix most of these issues.

The mac version is available btw, it's actually better than iOS one :D I'll add a link to it on the website, but you can actually just find it on AppStore.

Regarding sign in - the LLM that composes widgets is running on the server side, so we need accounts for attribution and rate-limiting. The widget layout is stored on the server side on your account too. No personal data is stored btw, they are just bindings, so if you say "my location" it just stores a binding to "currentLocation".

We do have ability to add custom API btw, I'll write up a guide for how to do that. That's kinda how the Claude uptime widget template works.

EDIT: Also if there are any public APIs that you'd have in mind, let me know and I'll add them to work out of the box!

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.

Thanks for trying it out! How much ram do you have? Pretty sure it's the only issue that can occur. The quality varies depending on the image too, so it might have been unlucky photos :(

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 would work on low-ram machines.

There's nothing to download - you can run everything from your browser, and the photo you upload is not uploaded anywhere, it stays in your browser.

Ahh, yeah I forgot to mention it. The model is 2.5gb so I assume you'd need at least free 3gb with all the surroudning stuff, with the rest of your system using more ram I'd guess 4gb is way too low - maybe even 8gb would be in some occasions.

I personally tested it on 32gb Apple M2, and it's able to run much heavier stuff.

Interesting. Yeah in-browser is not the best, but getting much easier over time!

Yeah I completely agree, but I think this model solves a different problem. AFAIK it's specifically there for the case where you only have one photo, but still need a 3D gaussian splat scene.

Nice, would love to see it, feel free to link it here <3

Interesting, what use cases are you using onnx for btw?

Yeah it's pretty cool what a 2gb NN can do from a single image

I think it's essentially a transformer, so it just stores a bunch of weights. As a model that's supposed to be able to convert any image to 3d scene, it's pretty nice size actually.

Regarding the ios lockscreen - I believe they are different models. I think Apple use this one to generate those Vision Pro 3d photos though, but I'm not too sure.

Hi HN, author here. SHARP is Apple's recent single-image 3D Gaussian splatting model (https://arxiv.org/abs/2512.10685). Their reference code is PyTorch + a pretty heavy pipeline; I wanted to see if it could run in a browser with no server hop, so I exported the predictor to ONNX and ran it via onnxruntime-web with the WebGPU EP.

What works: drop in an image, get a .ply you can download or preview live, all on your machine — your image never leaves the tab. The model is large (~2.4 GB sidecar) so first load is slow on a cold cache, but inference itself is a few seconds on a recent Mac.

Caveats: SHARP's released weights are research-use only (Apple's model license, not the code's). I host the exported ONNX on R2 so thedemo "just works", but you can also export your own from the upstream Apple repo and upload locally.

Happy to talk about it in the comments :)

Thank you <3 Yeah, the use case is really not for everyone though, but for Linear-heads it's really useful :D

yeah, me next xDDDD

Yeah for sure, I don't mind open sourcing it one day, but I think you'd understand why I chose current set up, since we're both on HN :D. I do think the free tier is more than enough for most people.

When you just copy the html, you get a ton of useless repeating data. This tool removed a lot of duplicated things, structures the styling in a better way where we don't spam the styling rules. The goal is to provide just the useful info, without the noise.

Having said that, it does do essentially what you described, but when prototyping fast, and not designing from scratch it's extremely useful.

If you love plagiarising, I will not be able to stop you :(

It's not, it actually currently doesn't even take screenshots. It actually looks at the html and includes actual structure and values from the webpage.

Thanks for asking, maybe I should make it clearer on the website I would also add a screenshot functionality later, if people feel like current approach is not enough.

Yeah, it should perform well with complex tailwind classes, and if not, there's a "this did not work" feedback button. I'll work on fixing anything that get's lost in translation :)

Valid concern, but I guess screenshot feature on every computer could also be considered as copyright violation.

You're right though - don't use this tool to violate copyright please :)

Hey HN! I built Pluck — a free Chrome extension that lets you click any component on any website and copy it as a structured prompt for AI coding tools (Claude, Cursor, v0, Bolt, etc.). You can also export directly into Figma!

It (should) work on any page, including ones behind auth, since it runs entirely in your browser.

Tech stack for the curious: Plasmo for the extension, Next.js + Hono + tRPC for the web/API, Drizzle + Postgres for the DB, all in a Turborepo monorepo.

Free tier gives you 50 captures/month (should be plenty). Unlimited is $10/mo.

Would love feedback — especially on the capture quality and whether the prompts it generates actually produce good results in your tool of choice.

A while ago I needed to get very custom SVG into my SwiftUI app, I also needed to programmatically control it, so making it into an SF Symbol was not an option. So as every self-respecting developer, I've created an automated tool to convert SVG code into SwiftUI Shape code.

It's been several years since the first release, but I think the tool still does its' job well enough. Let me know what you think!