@BartWronski@mastodon.gamedev.place
@BartWronski@mastodon.gamedev.place avatar

BartWronski

@BartWronski@mastodon.gamedev.place

Principal Research Scientist at NVIDIA.

Technology and art.
Computer Graphics by day, Techno and Industrial music by night.

https://linktr.ee/bartwronski

Ex Google Research, Ex games (Sony, Ubisoft, CD Projekt).
Politically leftist. He/they.

This profile is from a federated server and may be incomplete. Browse more on the original instance.

BartWronski, to random
@BartWronski@mastodon.gamedev.place avatar

In case you didn't notice, that a16z manifest written by a NIMBY-Karen directly cites the author of "fascist manifesto" (the OG Italian one) https://en.m.wikipedia.org/wiki/Filippo_Tommaso_Marinetti
as well as neo-reactionary accelerationism father Nick Land https://en.m.wikipedia.org/wiki/Dark_Enlightenment

You cannot make this stuff up. All masks are off. This is not "alarmism" but ideologies SV billionaires (including Thiel, Musk, Altman, Graham etc) believe in and publicly brag about.

BartWronski, to random
@BartWronski@mastodon.gamedev.place avatar

"we have the right to defend ourselves 🤡"

BartWronski, to random
@BartWronski@mastodon.gamedev.place avatar

I am excited to finally share our recent paper "Filtering After Shading With Stochastic Texture Filtering" (with Matt Pharr, @marcosalvi, and Marcos Fajardo), published at ACM I3D'24 / PACM CGIT, where we won the best paper award! 1/N

BartWronski,
@BartWronski@mastodon.gamedev.place avatar

"Everyone" knows blending and filtering do not commute with non-linear functions.
However, this is how texture filtering is taught and applied - we filter textures, then "shade" (apply non-linear functions). This introduces bias and error and often destroys the appearance. 2/N

BartWronski,
@BartWronski@mastodon.gamedev.place avatar

We reviewed 40y of graphics literature and unify the theory to propose "filtering after shading".
To make it practical and fast, we realize it through stochastic filtering and propose unbiased Monte Carlo estimators, together with two families of low variance methods. 3/N

BartWronski,
@BartWronski@mastodon.gamedev.place avatar

@aras I had this observation in an older blog post of mine that artists manually sharpening mipmaps (which seemed like a gross hack) is actually an intuitive compensation for ugly bilinear filter and correct, best-fit optimization-based solution gives similar results :) https://bartwronski.com/2021/07/20/processing-aware-image-filtering-compensating-for-the-upsampling/

BartWronski,
@BartWronski@mastodon.gamedev.place avatar

Many practitioners have used stochastic filters, but we generalize them, expand to negative lobe filters and infinite kernels, and propose an efficient way of sampling B-spline kernels.
We discuss the limitations of those techniques and cases where we do not recommend FAS. 4/N

BartWronski,
@BartWronski@mastodon.gamedev.place avatar

I think it's time we change how we teach and approach filtering textures.
Curious?
Check our paper and presentation slides: https://research.nvidia.com/labs/rtr/publication/pharr2024stochtex/ .
We also made shadertoys demonstrating two families of stochastic techniques: https://www.shadertoy.com/view/clXXDs https://www.shadertoy.com/view/MfyXzV 6/6

BartWronski, to random
@BartWronski@mastodon.gamedev.place avatar

I'm super excited to finally share our work, accepted to Siggraph 2023: https://research.nvidia.com/labs/rtr/neural_texture_compression/ Neural Texture Compression, where we create a novel compression algorithm and format explicitly designed for rendering and graphics.

Compressing whole material texture stacks and their mipmaps together, we beat the BC formats by 10-16x - while keeping the random access requirement.
At the lowest bitrates, we surpass JPEG XL and AVIF, despite those formats being entropy coded.

BartWronski, to random
@BartWronski@mastodon.gamedev.place avatar

Some really cool work of my teammates on differentiable shading language, compatible with HLSL, and interoperable with CUDA/Python/PyTorch, and even C++.
https://developer.nvidia.com/blog/differentiable-slang-a-shading-language-for-renderers-that-learn/

This makes ML + graphics significantly easier. I'm excited to see what researchers and engineers do with it :)

BartWronski, to random
@BartWronski@mastodon.gamedev.place avatar

New blog post: https://bartwronski.com/2024/01/22/how-i-use-chatgpt-daily-scientist-coder-perspective/
"How I use ChatGPT daily (scientist/coder perspective)."

I recommend it to anyone working with technology, but especially if you think that LLMs are "useless" and are open-minded to see how they can be helpful, delightful, and playful.

BartWronski, to random
@BartWronski@mastodon.gamedev.place avatar

We have recorded a 48min version of the presentation for our paper "Filtering After Shading with Stochastic Texture Filtering":
https://www.youtube.com/watch?v=e3391_C6aXE

Properly paced, going deeper into history, context, gamedev practices, and explaining recommendations - check it out! :)
Project page:
https://research.nvidia.com/labs/rtr/publication/pharr2024stochtex/

BartWronski,
@BartWronski@mastodon.gamedev.place avatar

In our original 20min slides, we didn't have enough time to explain aliasing from non-linearities properly and why it's more apparent if done before magnification. This was not clear and confused some folks, so we fixed it. Here is a relevant timestamp: https://youtu.be/e3391_C6aXE?t=1812

BartWronski, to random
@BartWronski@mastodon.gamedev.place avatar

Not many papers are so visually satisfying and beautiful as this one, using Diffusion generative model to create smoothly interpolated, spatially varying, correlated noise patterns of different classes: https://armanmaesumi.github.io/onenoise/

image/png
image/png
image/png

BartWronski, to random
@BartWronski@mastodon.gamedev.place avatar

Are there any shader toy experts here?

Two of my recent shader toys display just black on iOS Safari (no compilation errors), while a separate iOS app shows them "okay" (some bugs, but at least not a black screen).
https://www.shadertoy.com/view/clXXDs
https://www.shadertoy.com/view/MfyXzV
Nothing crazy, pretty simple, but using feedback for TAA accumulation.

BartWronski, to random
@BartWronski@mastodon.gamedev.place avatar

Also, this is why every industry needs strong unions: https://www.cnn.com/2023/11/28/business/tesla-sweden-postal-workers-ruling/index.html

We need unions, even if your job is cozy and safe (now), so that unions can cooperate and make abusive businesses and union-breakers difficult to operate.

BartWronski, to random
@BartWronski@mastodon.gamedev.place avatar

https://mastodon.lawprofs.org/@AnnemarieBridy/110890637706842148

Copyright and "authors' rights" were never about protecting authors.
They are monopolistic, anti-consumer extortion coming from huge corporations and making the world a worse place.

BartWronski, to random
@BartWronski@mastodon.gamedev.place avatar

Such a fantastic use of diffusion models and GenAI and an example of how much understanding about the real world, scenes, and light transport they have randomly encoded somewhere: https://diffusionlight.github.io/
Estimating scene light environment map by asking a generative model to in-paint a chrome ball inside a scene. 🤯

Hallucination? Obviously! This ill-posed inverse problem has no non-hallucinatory solutions.
But a plausible and super useful one, as they show with object relighting.

BartWronski, to random
@BartWronski@mastodon.gamedev.place avatar

The biggest existential threat to humanity?

People taking seriously the ramblings of tech bros.

BartWronski, to random
@BartWronski@mastodon.gamedev.place avatar

Israel Defense Minister has ordered a complete blockade on the Gaza Strip: "There will be no electricity, no food, no fuel, everything is closed. We are fighting human animals and we act accordingly."
This language and treatment of innocent civilians feels somehow historically familiar...

They also just gave out arms to the illegal West Bank settlers who committed government-approved ethnic cleansing genocides in the last months.

(Terrorists and murderers of innocents on both sides are evil.)

BartWronski, to random
@BartWronski@mastodon.gamedev.place avatar

“I tend to think that most fears about A.I. are best understood as fears about capitalism. And I think that this is actually true of most fears of technology, too. Most of our fears or anxieties about technology are best understood as fears or anxiety about how capitalism will use technology against us. And technology and capitalism have been so closely intertwined that it’s hard to distinguish the two.”
Ted Chiang

bitinn, to random
@bitinn@mastodon.gamedev.place avatar

“Tone Mapping Considerations for Physically-Based Rendering”

Cool interactive comparison.

https://modelviewer.dev/examples/tone-mapping

BartWronski,
@BartWronski@mastodon.gamedev.place avatar

@bitinn This screenshotted excerpt is terrifying. "Some folks don't understand PBR and creating correct assets where diffuse is more saturated to compensate for specular, so we bias a supposedly natural tonemapper so that incorrectly authored assets display better, while hurting any asset that is scanned or authored correctly". :(

BartWronski, to random
@BartWronski@mastodon.gamedev.place avatar

ACM I3D'24 committee ready for the conference! Hope to see some of you tomorrow! :)

BartWronski, to random
@BartWronski@mastodon.gamedev.place avatar

Ugh, I really liked Kind Bars, but they are canceled. :(

https://en.wikipedia.org/wiki/Daniel_Lubetzky

"In May 2024, Lubetzky was one of a group of billionaires and business interests who privately pressured New York City Mayor Eric Adams to deploy police on pro-Palestinian protests at Columbia University"

BartWronski,
@BartWronski@mastodon.gamedev.place avatar

It is so fucking infuriating that the same politicians accuse student protestors of being under the influence of some "agents" while they are under clear and documented influences of interest groups (and literal agents like AIPAC).
AIPAC should be delegalized as foreign agents under the Espionage Act, not TikTok.

pervognsen, (edited ) to random
@pervognsen@mastodon.social avatar

Fun visualization of low-rank matrices from the lecture I boosted earlier today.

Incidentally, I remember there were some papers on separable/low-rank BRDF approximations: BRDFs are 4-dimensional and you can approximate them as a sum of outer products of 2-dimensional texture maps. I remember one paper by Michael McCool but I don't think it got picked up or went anywhere in practice. Nowadays I assume people would just use NN-based approximations.

BartWronski,
@BartWronski@mastodon.gamedev.place avatar

@pervognsen I think low-rank approximations are still important, even in the ML landscape.
I'll start with a shameless plug of 3 blog posts:
https://bartwronski.com/2020/02/03/separate-your-filters-svd-and-low-rank-approximation-of-image-filters/ low-rank approximations of arbitrary image filters.

https://bartwronski.com/2020/03/15/using-jax-numpy-and-optimization-techniques-to-improve-separable-image-filters/ improving low-rank approximations with SGD.

https://bartwronski.com/2022/02/15/light-transport-matrices-svd-spectral-analysis-and-matrix-completion/ low-rank approximations of light transport matrices.

Finally, for use in contemporary ML, LoRA: https://arxiv.org/abs/2106.09685 Memory-efficient low-rank fine-tuning of huge ML models.

  • All
  • Subscribed
  • Moderated
  • Favorites
  • •
  • anitta
  • thenastyranch
  • rosin
  • GTA5RPClips
  • osvaldo12
  • love
  • Youngstown
  • slotface
  • khanakhh
  • everett
  • kavyap
  • mdbf
  • DreamBathrooms
  • ngwrru68w68
  • megavids
  • magazineikmin
  • InstantRegret
  • normalnudes
  • tacticalgear
  • cubers
  • ethstaker
  • modclub
  • cisconetworking
  • Durango
  • provamag3
  • tester
  • Leos
  • JUstTest
  • All magazines
    •