Imagine if all of OOT was like that, in first person, like they originally planned! It’d prob end up being like metroid prime lmao

Here’s a fun litmus test for you: does your sim produce the rainshadow effect & horse latitudes?

I wonder how well this system would work with a little bit of z axis flow too? You’ve shown that it produces trade winds, so a two-layer sim may produce full atmospheric cells. It’d make rendering easy too as it’s just two sphere textures

Mascot horrors gotten real weird lately huh

After this semester I’m going to see if I can improve on the Gaia team’s MCMC method with a direct MC and try to find a better fit. I’d love for this to be actually accessible at this scale

Warning: the entire Gaia dr3 dataset is 700 *gigabytes.* the culled data (position, distance, luminosity/flux, temperature, and “fog”) is ~15gb. Filtering it took around a day on my old pc.

But!! Parts are visible in each chunk, ~250mb each. It’s worth it!!

I wonder how useful modern diffusion model papers would be at this problem. I’ve found a lot of success taking neural methods and ripping the neural network out for nearby domain problems, like this one. Fourier gradient descent is a fun one that’s similar to non-neural CNNs as an example

If you can find a used one the entire 3000 series is wonderful. My heavily bottlenecked 3090 has been a dream despite the decade old cpu it’s bolted to

Wow my tablet takes *awful* screenshots. I’ll be sure to do this on my desktop from now on 😵‍💫

I wonder how it’d look if you weighed the lifetime by 1 - abs(dot(velocity direction, surface normal))? My thought here is it’d concentrate toward the edges and almost look like a shock.

It almost feels like your slowly constructing an abstract rollercoaster game lmao

I’ve been experimenting with a method of losslessly compressing using a BVH *as* the compressor and my early test showed a 40% reduction before a gzip, so it’s promising!

Gaia dr3 is a catalogue of around 1.8 billion stars. This chunk is ~250mb, and there are ~3300 chunks for a total of ~700gb.

It’s got a ton of useful scientific data that isn’t great for rendering, so my next task is figuring out how to compress it to less than 24 so it’ll fit into my gpu!

Oh!! This is all WebGPU?? Very impressive

Do you use any special software for this, or do you just use a timer? My Gaia rendering project may need a good profiler and I’m curious what people actually use besides time_ns.