dvrp

8 ms·

91.
by dvrp 10mo ago
gaussian splats alone are >500GBs lol (lot of bad quality ones though)
92.
by dvrp 10mo ago
Yes, also many were PPM images (or encoded as such) in PDFs and then I used (cheap/light) multimodal LLMs to classify documents from photos. It was surprisingly cheap: <$1 for a few thousand PDFs / Images.
93.
by dvrp 10mo ago
We're also working to have a RAG system on Jemini so you can talk to the files. many of the files have been embedded into vectors thru turbopuffer.
94.
by dvrp 10mo ago
big motivation for the whole project is to help structure the mess that was released
95.
by dvrp 10mo ago
A lot of the files are hosted on Cloudflare so their $0 egress cost helps a lot. Especially given that so many of the files can be considered immutable assets, so we can configure aggressive caching. I know that Luke was working on stuff s
96.
by dvrp 10mo ago
correct i can say that 90% of the work was in making sense of the data, indexing, processing, etc
97.
by dvrp 10mo ago
Also which platform? Mobile? I’ve noticed some burger menus failing on mobile.
98.
by dvrp 10mo ago
Would be nice to explain at some point how we did the structuring of the destructured data. For now we’re focusing on fixing the bugs because we’re already seeing an insane wave of traffic so most of us are focused on keeping the site alive
99.
by dvrp 10mo ago
just link dupe, but not content dupe read rest of the thread with more context on why
100.
by dvrp 10mo ago
I saw, but the project has changed since then. There's now Google Photos, improvements to Drive, Jemini (LLM that can assist when understanding the files), and JFlights for flight tracking. I sent an email to the hn email with more det
101.
by dvrp 10mo ago
I participated on this project, but I'm not the main contributor. In fact, I'm working on running an ML model to get some nice gaussian splat visualizations of some of the relevant images based on the recent release by Apple'
102.
by dvrp 10mo ago
See also: - JPhotos: https://www.jmail.world/photos - JDrive: https://www.jmail.world/drive - JAmazon: https://www.jmail.world/jamazon
103.
by dvrp 10mo ago
I saw some people at a company called Pruna AI got it down to 8 seconds with Cloudflare/Replicate, but I don't know if it was on consumer hardware or an A100/H100/H200, and I don't know if the inference optimization
104.
by dvrp 10mo ago
See also: - Paper page: https://huggingface.co/papers/2512.15603 - Model page: https://huggingface.co/Qwen/Qwen-Image-Layered - Quantized model page: https://huggingface.co/QuantSt
105.
by dvrp 10mo ago
Qwen-Image-Layered is a diffusion model that, unlike most SOTA-ish models out there (e.g. Flux, Krea 1, ChatGPT, Qwen-Image) it's (1) open-weight (unlike ChatGPT Image or Nano Banana) and Apache 2.0; and has 2 distinct inference-time f
106.
Qwen-Image-Layered: transparency and layer aware open diffusion model(huggingface.co)
130 points by dvrp 10mo ago|21 comments
107.
TC39: Signals(github.com)
3 points by dvrp 10mo ago|1 comments
108.
TRELLIS.2: state-of-the-art large 3D generative model (4B)(github.com)
86 points by dvrp 10mo ago|17 comments
109.
SHARP, an approach to photorealistic view synthesis from a single image(apple.github.io)
531 points by dvrp 10mo ago|108 comments
110.
CMYK Simulator(cmyk.danielpetho.com)
2 points by dvrp 10mo ago|0 comments
111.
by dvrp 10mo ago
“Our findings [show] that light […] provides a credible and independent source of information on early modern economic activity.” Wow!
112.
The Mundanity of Excellence(fermatslibrary.com)
2 points by dvrp 10mo ago|1 comments
113.
by dvrp 10mo ago
Conway’s Law yet again!
114.
Swift Implementation of DINOv3 with MLX Swift(github.com)
1 points by dvrp 10mo ago|0 comments
115.
by dvrp 10mo ago
Krea | Distributed Systems & Backend Engineers | San Francisco, CA (onsite+remote) I'm the cofounder & CTO of Krea ( https://www.krea.ai ). We are a startup in San Francisco building browser-based AI creative tools fo
116.
Ty – A fast Python type checker, written in Rust(docs.astral.sh)
77 points by dvrp 10mo ago|15 comments
117.
by dvrp 11mo ago
They messed up. We (Krea) were also surprised. They put our logo after we pointed it out. Nice eye!
118.
by dvrp 11mo ago
https://en.wikipedia.org/wiki/Productivity_paradox
119.
by dvrp 11mo ago
“ When Zork arrived, it didn’t just ask players to win; it asked them to imagine” Sigh… it’s all ChatGPT nowadays ain’t it.
120.
Back to Basics: Let Denoising Generative Models Denoise(arxiv.org)
4 points by dvrp 11mo ago|0 comments

More ›