Anders Langlands

@maxliani I’ve only ever done single wavelength at dispersion events.

I was wondering, how do you sample dispersion with hero wavelength path sampling? Do you scatter on the hero wavelength alone, or do you resort to path splitting?

@self_shadow Congrats! Well deserved

@maxliani @jakub_vondra @Jiayin_Cao That’s what I did in alSurface back in the day

Yes, it can be considered a form of resampling. One draw the 2d random number to sample BXDF lobes, samples all of them (spercular, diffuse, etc…) with the same random number, build a CDF with those candidate throughputs and use a separate random number to sample which one to trace the scattered path with. The technique has pros and cons. Pros: it’s simpler to implement as you don’t need extra code to approximate the weighting between lobes. It may be faster because (at least on the GPU) all of your lobes sampling may be running divergent, plus the approximate picking logic. With this execution is coherent. It tends to produce lower variance. Cons: you need 3 random number dimensions instead of 2. If you happen to have 3d stratified sampling, cool; otherwise you may have more noise due to the lack of stratification between 2d sampling direction and the 1d resampling.

@maxliani @kenpex @yiningkarlli Yeah we’re probably not going to start adding transmission to meta shaders any time soon :)

@anderslanglands @kenpex @yiningkarlli Years ago I was convinced that a sheet of metal thin enough would transmit, then somebody convinced me otherwise. It seems I have to change my mind again🤭. Not that it changes anything in practice tho.

Fun fact: even most rocks have at least a small amount of subsurface (and yes, we've been putting at least a small amount of subsurface in rocks at WDAS since the original Moana).

@maxliani @kenpex @yiningkarlli AIUI the chance of a photon not interacting with free electrons is incredibly low so mean free path is effectively (but not quite) zero.

@maxliani @kenpex @yiningkarlli physics.stackexchange.com/questions/1614….

@maxliani @kenpex @yiningkarlli Metals also transmit a teeny amount of light. You can see light through gold leaf, for example.

@kenpex @yiningkarlli Except conductors (metals). Most materials are dielectrics

@j3ffdr Heh looks like our greenscreen shots on The Martian

Great Perseverance rover shots. Though one might ask why they look like CG? A few reasons why lighting on Mars would give this impression: smaller sun = less contrast, sharper shadows (very CG-looking). And sky doesn't scatter blue light like ours, so it doesn't look "outdoors".

@danielelliott3d @DMeville This is the right answer. Doesn’t matter which you do as long as the parameter is named correctly.

@DMeville It really becomes a matter of semantics. ie Is the parameter called scaling or tiling? if its scaling, then I want it to scale. if its tiling then the number is how many times you want it to tile.

When you scale texture coordinates for tiling a texture in a material do you do you:

@slider1978 What are you using for the wire format between host and subscribers?

and you might don't want to stop to one "subscriber", you could have as many as you need for different purposes (live rendering, disk rendering, whatever else) (9 and last, sorry 😅)

I've been asked about this "gameengine" in #hydra and what that means. Not easy via a twitter/x post obviously, but I've prepared a few images to explain it as best as I can here 😅 It's all about #openusd and #hydra being the common languages across hosts/dccs/renders (1)

Virtual Town Halls kick off tomorrow! Join #ASWF projects as they share highlights from the past year, provide a look ahead at what's to come and answer questions from the community. View the schedule and register at hubs.la/Q02HxL910. #opensource #vfx #animation

Myself, @SheriffStone, Carol Payne, Nick Porcino and @AdrienHerubel will present on the latest developments in standardization for lights, cameras, colour, materials and more in USD. We expect and encourage lively discussion of these essential topics for all your rendering needs!

Going to #SIGGRAPH2024? Interested in rendering and #OpenUSD? Come to our BOF on Seamless Rendering Interchange in OpenUSD, 1pm Mon 29th to hear about and discuss the latest news and future plans around effectively sharing scenes for rendering in USD!

@kenpex @BrianKaris @KostasAAA I like the idea of a book club for graphics papers

@BrianKaris @KostasAAA I've thought a few times to create a "club" to make proper summaries of papers, collaboratively - with eventually even the hope of creating a standard people use in their project pages. Abstracts are not designed to give a proper summary

How to keep up with the state of the art in rendering is a common theme in the DMs I receive. Conferences, books, blog posts all available at a click of a link, we live in a golden age of (graphics) knowledge sharing. This also makes it overwhelming and hard to keep up. 1/4

@mmalex my brain just experienced an IRL smash cut to being 5 years old watching Bertha

i just woke up and my brain informed me that the 'bertha' and ' neighbours' theme songs are the same song. you're welcome. youtu.be/AIQoljauPyM?si…

OpenPBR in @nvidia Omniverse. A single Uber-Material for use everywhere.

@maxliani @rgba8 @lauriewired Megakernels considered harmful considered harmful?

@rgba8 @lauriewired Hybrid in this context = wavefront for first N bounces, then monolithic for the long tail.

Pixar doesn't use GPUs (much). Their render farm compute is mostly CPUs with a ton of cores + memory, using AVX-512 and SSE 4.2 optimizations when appropriate to optimize render time. Why? GPU render compute doesn't speed things up as much as you might expect. Yes, in certain instances, you can get a 3-4x benefit, however you have to fit all the polys in GPU memory. Paging back out to system RAM brings you back down to CPU speeds. Scenes can often contain > 150,000,000 polys. Only very recently have gpu memory sizes gotten big enough to be "worth it" over CPU rendering, and even then, it's often less flexible/tolerant of mixing older architectures in the cluster.

@robpieke 8 minutes of amazing looking features I will never have the time to try 🥲

@TylerGlaiel youtu.be/QT9p0tCiIPc?si…

something I've always found curious about the way we do graphics on computers is that despite how photorealistic we've been able to make stuff look, I don't think any rendering systems out there can do something like monochromatic sodium light without doing it as a weird hack

OpenPBR 1.0 is now available! This update introduces a reference implementation for OpenPBR that's integrated within @MaterialXcg – meaning that tools in the MaterialX and OpenUSD ecosystem will automatically support OpenPBR. Read more on our blog: aswf.io/blog/academy-s…