Sebastian Sylvan's Avatar

Sebastian Sylvan

@ssylvan.bsky.social

Graphics (VR/AR) at Apple.

201 Followers  |  204 Following  |  29 Posts  |  Joined: 10.10.2023  |  2.556

Latest posts by ssylvan.bsky.social on Bluesky

BILL OF RIGHTS STATUS

❌ First Amendment
✅ Second Amentment
❔ Third Amendment
❌ Fourth Amendment
❌ Fifth Amendment
❌ Sixth Amendment
❔ Seventh Amendment
❌ Eighth Amendment
❌ Ninth Amendment
❌ Tenth Amendment

Things are about to get very, very dark in the US.

18.09.2025 00:52 — 👍 82    🔁 25    💬 3    📌 0

It’s a free speech issue when it’s done so blatantly at the behest of the government.

18.09.2025 17:49 — 👍 6    🔁 0    💬 0    📌 0

Big release! Congrats!

21.08.2025 03:30 — 👍 1    🔁 0    💬 0    📌 0

Did someone say Journey?

10.08.2025 01:41 — 👍 2    🔁 0    💬 0    📌 0

I hope you’re exited about AI!

23.07.2025 22:00 — 👍 1    🔁 0    💬 1    📌 0

I’m pretty sure that suppressors are mandatory in like Norway or something. And I agree, all firearms should be NFA items. It’s not particularly hard to do, but hard enough to limit things to people who actually use it for hunting or sport or whatever.

03.07.2025 04:29 — 👍 0    🔁 0    💬 1    📌 0

Wait what? What’s vulkan doing?

30.06.2025 02:21 — 👍 0    🔁 0    💬 1    📌 0

That’s some good looking grass though.

01.06.2025 02:54 — 👍 0    🔁 0    💬 0    📌 0

TLDR rendering isn’t what’s cool about splats IMO. It’s the optimization part where they shine. We “just” have to find a way to convert to a more easily rendered rep.

30.05.2025 21:18 — 👍 0    🔁 0    💬 1    📌 0

Some kind of learned spherical function per texel combined with a triangle mesh extracted from splats perhaps?

30.05.2025 21:17 — 👍 0    🔁 0    💬 0    📌 0

Ie use the super flexible but inefficient representation to find a high quality scene. But then find a way to convert to an efficient representation (possibly with a final optimization pass on material parameters once structure is frozen). No idea what that would look like in practice tbh!

30.05.2025 21:16 — 👍 1    🔁 0    💬 1    📌 0

I think these kinds of splatting approaches are amazing for the optimization part. And of course you need a way to render to optimize the representation. But IMO it’s a mistake to assume that the final render needs the same representation.

30.05.2025 21:14 — 👍 0    🔁 0    💬 2    📌 0

It’s because Russia doesn’t want the war to end. The only way this ends is if Russia is punished enough that they want peace. Which means arming Ukraine to the teeth.

11.05.2025 03:57 — 👍 1    🔁 0    💬 0    📌 0

But. Circles have both a circumference and an area. How do they get around pi?

02.05.2025 02:52 — 👍 0    🔁 0    💬 2    📌 0

Mous is pretty solid. It s backpack, lots of pockets and two slots for laptops etc. Not huge but enough for a few days of clothes and tech etc.

02.05.2025 02:37 — 👍 1    🔁 0    💬 1    📌 0
Post image

PhD Timeline xkcd.com/3081

25.04.2025 15:32 — 👍 60404    🔁 20710    💬 602    📌 832

Oh wait. True Lies is even better.

23.02.2025 06:23 — 👍 1    🔁 0    💬 0    📌 0

Not super early but The Sixth Day is peak Arnold.

23.02.2025 06:14 — 👍 0    🔁 0    💬 1    📌 0
Preview
Nullable reference types - C# reference Learn about C# nullable reference types and how to use them

You can turn that off. learn.microsoft.com/en-us/dotnet...

23.01.2025 23:52 — 👍 1    🔁 0    💬 0    📌 0

You could use frame N and frame N-1 to compute optical flow, and use that to forward predict to frame N+1. Definitely not as straightforward (eg disocclusion artifacts) but 4 frames of interpolation seems a bit silly

07.01.2025 22:48 — 👍 1    🔁 0    💬 0    📌 0

Mrs Doubtfire

25.12.2024 04:06 — 👍 0    🔁 0    💬 0    📌 0

Ooooh. Weird!

04.12.2024 02:33 — 👍 1    🔁 0    💬 0    📌 0

I don't think that's what they're doing? The left half is without ray tracing, and the right side is with ray tracing.

04.12.2024 02:25 — 👍 0    🔁 0    💬 1    📌 0

Similar with vehicles. In theory quite variable in size but for something 500m away you don’t really care that much as long as it’s in the ballpark.

25.11.2024 22:08 — 👍 0    🔁 0    💬 0    📌 0

The distance between the left side of a lane and the right side is approximately constant over distances you care about. And by the time you care about more accuracy you’ll be closer and can do easier things.

25.11.2024 22:07 — 👍 0    🔁 0    💬 1    📌 0

Roads kinda do this implicitly too. You know how wide the road is nearby so you can use to infer distant depth (taking lane changes etc into account)

25.11.2024 20:06 — 👍 1    🔁 0    💬 1    📌 0

(0.5+0.1)*15= 7.5+1.5=9

25.11.2024 16:38 — 👍 4    🔁 0    💬 0    📌 0

Reposting the SLAM Handbook again for the robotics people arriving here :)

24.11.2024 09:19 — 👍 33    🔁 8    💬 2    📌 0

He has a ton of fun lectures, all of which with the same great explanations and examples. The sparsity stuff for example. Worth browsing if you like it.

21.11.2024 13:38 — 👍 1    🔁 0    💬 0    📌 0
Preview
Control Bootcamp - YouTube This course provides a rapid overview of optimal control (controllability, observability, LQR, Kalman filter, etc.). It is not meant to be an exhaustive trea...

Forgot link and can’t edit here apparently. youtube.com/playlist?lis...

21.11.2024 13:23 — 👍 1    🔁 0    💬 1    📌 0

@ssylvan is following 20 prominent accounts