Justin Ling

let's go beyond the screen

[Konrad Reczko @reczko_konrad]

Depth-aware light injection in TypeGPU

I got a 448x448 monocular depth model down to ~8 ms on my M4 Pro across ~250 dispatches, which is fast enough to use in realtime :D
Since the inference is written directly in TypeGPU, I can just feed the depth buffer straight into the lighting pass. It never has to leave the GPU or go through any extra synchronization/interop step

Inference, lighting and draw all go through the same command encoder.

[Justin Ling @upload_j]

@helloharuko i did the same (i even had to delete youtube, because naturally yt shorts filled the gap). Now pinterest is like my nicotine patch.

[Justin Ling @upload_j]

i use the word "optimize" way too often.

[Brad Lynch @SadlyItsBradley]

I think it’s really coming along quite nicely. Three modes of play now!

I am really trying to make Trainer/Wild Pokemon encounters take full advantage of XR. While also trying to keep the original style intact..

So please be patient with me as I continue cooking that 🥹

[Justin Ling @upload_j]

trying to figure out if its me or the model that is currently underperforming.

[Ben Lau @benlaudesign]

Today on building whatever I want, a Chinese calligraphy art and learning tool

[Fred Moon 🚀 @fw3d]

i've never bookmarked a site so quick

[Sergey U @ustnvs]

iOS & Android CLI for your coding agent

noqa action tap -d "Blue login button at the bottom"

Any frameworks and platforms, real devices, no locators or coordinates headache.

[Alex Kotliarskyi 🇺🇦 @alex_frantic]

How to graph-max with Codex and 5.6 Sol:

1. Draw a graph (literally in any tool, even on paper)
2. Send it to Codex and say "write a code mode script that implements this workflow, run it with <your inputs>"

There's no step 3, it just works.

Seriously it's that simple.

[Justin Ling @upload_j]

@joshuavoydik a significant amount of time that has been converted into quality.

[MΞTΛCO @MetacoGen]

Vertical hustle

[Ben Springwater @benspringwater]

And on the seventh day, man rested from the timeline.

Introducing Sabbath: Block X (and other addicting apps) one day a week.

 Now on the App Store. Link below.

[Andrej Karpathy @karpathy]

One pattern I find useful for working with LLMs is a nice long ramble session. Sometimes the LLM needs more bits to understand what you're trying to achieve, but you're too lazy to type them. In these cases I like to lean back, switch to /voice and just ramble for like 10 minutes, total mess, anything goes, full stream of consciousness. Sometimes I declare it up top, something like "switching to speech recognition sorry for any typos...". Sometimes I turn it into a small interview of a few turns. But I find that the LLMs are somehow very good at reconstructing long incoherent rambles and often their echo of your own tangle of thoughts comes out quite a bit cleaner than what you started with. The result is that you improve the mind meld and have to correct things less from that point on.

[Justin Ling @upload_j]

is codex your sidechick or main b

[Justin Ling @upload_j]

humans are great synthesizers

[Josephine Ong @josephineonggo]

Merry go round and round and round and round
here's an ASCII generator i'm gonna will into existence :)

[Justin Ling @upload_j]

Are better writers also better thinkers?

[Justin Ling @upload_j]

if you were a super-intelligence that could take the physical form of any human. Why would you ever choose a man?

[Justin Ling @upload_j]

Are OpenAI researchers allowed to use claude code?