Maciej Halber

YouTube video by Game Industry Conference Physically Based Rendering: Putting Pieces Together - Paweł Stolecki || CD Projekt Red ||

Nice introduction (or refresher) to PBR: "Physically Based Rendering: Putting Pieces Together", Game Industry Conference 2024 presentation recording www.youtube.com/watch?v=OwyM...

GitHub - memononen/Skribidi: Nimble bidirectional text stack for UIs Nimble bidirectional text stack for UIs. Contribute to memononen/Skribidi development by creating an account on GitHub.

Also, Skribidi is a work of art.
github.com/memononen/Sk...

Falling down the rabbit hole of text rendering. Please send help.

YouTube video by I3D Symposium I3D 2025 Chris Wyman's Keynote "ReSTIR: Traveling the Path of Reuse"

Worth watching presentation, also provides some background info on how ReSTIR came to be and where it is heading: "ReSTIR: Traveling the Path of Reuse" www.youtube.com/watch?v=MRg9...

IMHO it is really hard to have a clear api for all the cases, so it is more up to the programmer to know what sort of transformations they have and what data is necessary for efficient inverses

That is up for the library author to decide, really. For a library that is meant for graphics, the main thing for inversion is a normal matrix, and that is only necessary if you have non uniform scale. If you were doing something like camera pose estimation then rotation + translations happen a lot

Technically for simple transformation that involve just rotation and translation you can get inverse via R.T, -R.T*t, but I’m sure you know that ;)

Yes! Here are all the SGP grad school recordings for every year: school.geometryprocessing.org

I'm quite impressed how good it is at guessing the scale. It's a bit off, but way closer than I'd expect.

GitHub - microsoft/MoGe: [CVPR'25 Oral] MoGe: Unlocking Accurate Monocular Geometry Estimation for Open-Domain Images with Optimal Training Supervision [CVPR'25 Oral] MoGe: Unlocking Accurate Monocular Geometry Estimation for Open-Domain Images with Optimal Training Supervision - microsoft/MoGe

MoGe-v2 is bananas
github.com/microsoft/MoGe

I def. am having identity crysis hearing you and Omer saying you're not engineers.

Caved in and tried zed with all the llm bells and whistles. Sadly, what gets me the most is that llm based edit predictions are faster than lsp suggestions :S

pyright is SO ... DANG ... SLOW

Where did we go wrong?

Let's call them challenging indoor handheld captures :)

They are crazy sequences with tons of motion blur. For these VGGT struggled, while MUSt3R produced pretty good results.

That being said, this is not a quantitative statement, mostly just vibes based on spot checking.

I sadly can't say I share the experience :/
MUSt3R is really good tho

Clearly must be the phase of the moon.

I wonder if a side-effect of LLMs will be lesser adoptions of new programming languages, esp. languages developed after LLM-boom. While the outputs of an LLM on C / JS are impressive, trying it on something like Odin is a bit of a dumpster fire.

Possible, but still impressive. In this case LLMs are truly magical. I remember just starting and being very intimidated by such formulas. Now you can plop it into an LLM and get solution + possible guidance. Might not be perfect, but nicer then just not knowing where to start.

Are you using free type to raster or the Skribidi rasterzier?

was looking for some slides, then stumbled upon this interesting repo/file:
GPU Optimization for GameDev (md format)
gist.github.com/silvesthu/50...

Another treasure trove is this list: tfpsly.free.fr/bookmarks.html although by now maybe a 4th of these links lead to defunct sites.

I have some gripes with unpaid work, can't pay for food with exposure.

Re.first point - is the idea that people will be able to skill up more easily thanks to LLMs? I hope so, but in my experience the LLMs are particularly useful when you know what question to ask.

While we are spinning up the coding agent's doing "intern's" work, what is the game plan to train people up if there is no entry level position?

Is the long term game to improve the models to completely remove necessity of coding to build software product?

A three panel image. Top left - classical orchestra with caption "COLMAP", top right - metal band with caption "Mono Depth", bottom row - symphonic metal band with caption "MP-SfM"

Really enjoyed MP-SfM! (arxiv.org/pdf/2504.20040)
Very cool problem formulation, awesome results!

Also, a great excuse for a joke for like 3 people.

Load-store conflicts meshoptimizer implements several geometry compression algorithms that are designed to take advantage of redundancies common in mesh data and decompress quickly - targeting many gigabytes per second in...

New blog post! "Load store conflicts", in which we look at some performance sensitive code that has surprisingly dramatic performance swings based on the compiler and the microarchitecture used. Reposts appreciated!

zeux.io/2025/05/03/l...

A closeup picture of a wall with nothing on it

No, not like that!

Mirrors were a mistake.
Lambertian everywhere, please.

Anytime someone mentions high speed rail I get reasonably upset at the state of railroads in the US.

Theydo crazythingswith tokenizersnowadays.

All my life I've struggled with math, but this seems like more than 3 words.

So excited to finally get some AI employees to summarize all my emails after long day of work, while I'm getting home in a fully self driving taxi, with my humanoid robot preparing my dinner at home. All surely coming next year, frl frl.