Hacker News: alexkranias

New comment by alexkranias in "Show HN: Diffuji – a diffusion-powered instant camera"

alexkranias — Mon, 16 Feb 2026 18:53:05 +0000

We were looking into OnnxStream (https://github.com/vitoplantamura/OnnxStream) and modifying it to support img2img. We got pretty close but yeah capability of running diffusion models on a Raspi are quite limited lol.

Alternatively we could use compute from your iPhone, but it adds additional dependencies to external hardware that I don't quite like. We could use a Jetson, but then power draw is quite high. I agree with you that on-device inference is the holy grail, but figuring out the best approach is something we are still trying to figure out.

New comment by alexkranias in "Show HN: Diffuji – a diffusion-powered instant camera"

alexkranias — Mon, 16 Feb 2026 18:50:51 +0000

Hey it's Alex, I was one of the creators of this project. We thought it would be fun to experiment with this odd mix of real, physical media and altering reality with diffusion models. I'm very happy with the results lol!

The entire thing can be built for around ~$35-40.

Here's our parts list: - Raspberry pi 2w for its low power draw, wifi capabilities, decent memory (512mb), and low cost ($15). Also with a arducam - Cheap TTL Aliexpress thermal printer along with sticker thermal paper - Rotary encoder for switching modes, push button for our shutter, and tactile switch for our power switch - I2C OLED display for display our modes and settings - 2 18650 batteries and a 3amp 5v UPS supply, enabling good battery life and regulation. We also add a 1000uF capacity to guard against current spikes from the printer.

We have a longer write up on our devpost: https://devpost.com/software/diffuji?ref_content=user-portfo...

We submitted this for Treehacks 2026, and ended up winning the Most Creative grand prize :)

Recreating Hi8

alexkranias — Wed, 11 Feb 2026 15:47:53 +0000

Article URL: https://alexkranias.com/essays/hi8.html

Comments URL: https://news.ycombinator.com/item?id=46976414

Points: 1

# Comments: 0

Attention Sinks and Compression Valleys in LLMs

alexkranias — Wed, 11 Feb 2026 15:45:50 +0000

Article URL: https://arxiv.org/abs/2510.06477

Comments URL: https://news.ycombinator.com/item?id=46976396

Points: 1

# Comments: 0

Show HN: Compressing a Histogram from 16GB to 2KB

alexkranias — Thu, 15 Jan 2026 15:54:46 +0000

Thought I'd dive into this cool streaming/approximation algorithms problem I encountered a few months ago.

TLDR: we can create our own custom floating point representation for encoding and decoding integers and use this to index into a tiny 2D histogram, upper-bounding approximation error based on the number of bits per integer we specify to keep.

---

Broadly useful for aggregating statistics from massive data streams of user data. Also turns out this is incredibly similar to data structures used in production, like HDRHistogram.

Comments URL: https://news.ycombinator.com/item?id=46634366

Points: 5

# Comments: 0

"Numbers Everyone Should Know" from Jeff Dean

alexkranias — Sat, 15 Nov 2025 20:03:16 +0000

Article URL: https://brenocon.com/dean_perf.html

Comments URL: https://news.ycombinator.com/item?id=45940169

Points: 3

# Comments: 0

New comment by alexkranias in "The FFT Strikes Back: An Efficient Alternative to Self-Attention"

alexkranias — Mon, 03 Mar 2025 13:40:29 +0000

Is this not just a global convolution?