Sounds like challenging but sane, animation and Vfx is such a tough industry. In 2010 I was directing a kids animated series, this was our own project with my wife and we had funds but I burned out by pouring my soul in it…

The Vfx and animation industry isn’t going very well, I’m happy I moved to tech 3 years ago

There’s also these things called museums or even just art galleries, maybe they should try one day.

I’d buy a print

Great article, thank you!

I get that the news cycle is packed right now, but I just heard from a colleague at the Smithsonian that this is fully a GIANT SQUID BEING EATEN BY A SPERM WHALE and it’s possibly the first ever confirmed video according to a friend at NOAA

10 YEAR OLD ME IS LOSING HER MIND (a thread 🧵)

In my multithreaded graph, data is stored contiguously in a buffer and accessed with offset+size. As you remove or add nodes the buffer becomes fragmented and my sorting/compacting functions were buggy so I added little #dearimgui widget to help me out. Now it works perfectly

The Offset Geometric Contact (OGC) method @ankachen.bsky.social will present at #SIGGRAPH2025 allows guaranteed penetration-free simulation of codimensional objects with minimal computational overhead.

Project page: graphics.cs.utah.edu/research/pro...

That, plus I don’t buy the non invasive mind transfer, and the way a ship that big, gently crashes without half burning/breaking.
Also surprise surprise no alert system at all.

Meanwhile life in California 🌴
#JoshuaTree #SaltonSea #SequoiaNationalPark #KingsCanyon #MountWilsonIObservatory #HuntingtonGarden #MonsterTruck #LosAngeles #SantaMonicaMountains #California

Meanwhile life in California 🌴
#JoshuaTree #SaltonSea #SequoiaNationalPark #KingsCanyon #MountWilsonIObservatory #HuntingtonGarden #MonsterTruck #LosAngeles #SantaMonicaMountains #California

Meanwhile life in California 🌴
#JoshuaTree #SaltonSea #SequoiaNationalPark #KingsCanyon #MountWilsonIObservatory #HuntingtonGarden #MonsterTruck #LosAngeles #SantaMonicaMountains #California

Faster GPU BVH! 🎉

A jewel in Paris

Just found a Flashback mosaic while visiting Paris ❤️

It brings memories, I started my internship with an SGI Indy station and Softimage 3.5!

I added computation to the index version and It's also 100x faster to evaluate 🤯

Unity's vfx graph is like that, and that's something that I'd like to explore at some point.

I knew I'd have to make such a comparison at some point, but it was pretty far in the list and going OO was for me the easy and fast way to just get things working.
Following your updates made me curious and inspired me.
So thank you!

6.14 seconds for 10K nodes...
I mocked up something index based and it was 0.0036 seconds. I guess it time to refactor V_V.

Another reasoning behind going with pointers was the assumption that removing nodes and connections on very large graphs (10000 or 100000 nodes) without pointers would be slow, but I haven't tried so I can't really compare.

I guess it's pretty similar to your's or a DCC, an heterogeneous mix of POD, transforms, meshes, and linear algebraic types

It's also a way of experimenting/learning and improving my c++ but I still consider myself as a noob

I was inspired by Dreamworks LibEE 2 paper and wanted to see what it would take to make my own evaluation engine the same way.
So it's mostly plumbing and managing a graph on top of TBB.
I'm also using #dearimgui for the UI.

I went this way at first as it was less data management with the idea of going to a more cache and alignment friendly version later....which hasn't happened yet even if I went into a few revisions.
That's why I'm interested by seeing your iterations.

I'm not criticizing at all, just curious. In my own node graph ports are owning the values and connected by pointers so....

So a hash map? I was wondering as adding/removing input port values of different types in a std::vector<char> every time you add and remove a node seems tedious.
Or maybe I'm completely wrong?

So are you storing the values in an external array and accessing with an offset+byte length?
Your graph data model deserve a blog post!

Indeed

That's how it is in Bifrost