What’s Next in Computer Graphics? | SIGGRAPH 2026 Technical Papers

Ajouté : 2026-05-31

[00:00:01]Hello, and welcome to the SIGGRAPH 2026 Technical Papers Trailer, where we showcase some of the incredible technical work being presented [music] this year at SIGGRAPH. Let's take a look at some of this amazing work.

[00:00:13]Simulate a range of materials with a new hybrid Lagrangian-Eulerian method that eliminates the need for particle-based quadrature.

[00:00:22]If you need a fine body dynamics instead, this paper lets you distribute your simulation across multiple GPUs and compute nodes.

[00:00:33]Clean your armadillos, then dispose of them with GPU-based Jacobi-like simulation [music] that leverages the smoothness of perturbation subspaces.

[00:00:42]Create beautiful works of art with sharp 2D fluid brushes using a progressive, resolution-independent, and GPU-based mixing system.

[00:00:51]Hello.

[00:00:54]Need virtual materials? This paper lets you create them using a video generation model that serves as a virtual gonio-reflectometer. [music] Let's talk about lighting. Simple PBR materials can't capture complex effects like fuzz or clear coat. This paper augments simple GGX [music] lighting and encodes these complex properties in a six-dimensional latent space.

[00:01:16]This new model for body relighting is trained on video relighting pairs captured in a large LED sphere using digital bi-pack technology.

[00:01:26]If you need to pre-bake complex light scattering [music] in 3D assets, this paper lets you encode it as neural eight-dimensional light transport functions.

[00:01:36]These novel shape-changing persistence of vision displays use centrifugal force to form tropospheres from LED strips.

[00:01:46]Have simulated fun at the beach with this physically-based framework for simulating mixtures of sand, [music] air, and water.

[00:01:57]Then dry yourself off with these towels, which are simulated by generalizing the Douglas-Rachford splitting [music] scheme to solve complicated non-convex couplings between yarns.

[00:02:07]This new variant of the mixed [music] point material method supports everything from sand and snow to elastic solids.

[00:02:12]>> [music] >> It can also be coupled to traditional rigid body solvers.

[00:02:17]If you need to simulate fluids, this efficient boundary conforming fluid simulation network [music] has everything you need.

[00:02:28]Texture Lava and Paint are two novel wave-based procedural texturing algorithms, which can produce natural-looking textures [music] even under extreme deformations.

[00:02:39]See the latest advances in generative AI, like this pixel-aligned method for 3D generation from images.

[00:02:49]Explore novel surface representations, like this neural surface representation that encodes surfaces [music] as a combination of a Catmull-Clark base surface and a neural offset map.

[00:03:00]This paper reformulates generative modeling as learning cumulative flow maps, [music] which can be used to generate anything from SDFs to sketches.

[00:03:09]These generated puzzles can only be disassembled by simultaneously moving [music] parts along distinct paths.

[00:03:17]Why won't this come apart?

[00:03:20]If 3D printing is your game, this fused filament fabrication method generates non-planar layers and continuous [music] tool paths aligned with user-defined fields.

[00:03:30]If you need to move furniture, enhance collaborative human-object interaction animation by generating hand and finger motion [music] and refining noisy data.

[00:03:40]This real-time generative animation framework delivers high-quality zero-shot motion synthesis at [music] 15,000 frames per second. Robots go bananas for it.

[00:03:51]If you want to transfer motion capture to a robot, this reinforcement learning paper lets [music] you do it without artifacts like foot sliding, hovering, or self penetration.

[00:04:01]This new kinematic kit bashing method lets you build all kinds of articulated assemblies ranging from kitchen cabinets to whatever this is.

[00:04:13]Let's finish with a little music courtesy of a new data-driven [music] framework for physics-based muscle-driven dexterous control targeting musical scores.

[00:04:25]These are just a few of the amazing technical papers being presented [music] at SIGGRAPH 2026. Join us this July in Los Angeles to be part of the future of computer graphics and interactive [music] techniques.