正在加载视频...

视频加载失败

加载此视频时出现问题。这可能是由于临时网络问题，或视频可能不可用。

Working on sparse volumetric light maps for WebGPU. Got generation part sorted. The resolution is adaptive, refining near geometry and skipping through most of the empty space. The video is question uses a map that's only 10MB in size, including everything. That's less the amount of GPU memory that... show more

Alex Goldring

2,289 subscribers

18,651 次观看 • 4 个月前 •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

0 条评论

暂无评论

原始帖子的评论将显示在这里

相关视频

Working on Sparse Volumetric Light-maps. Thanks to CynicatPro🎃 for pointing me at Unreal's version. In a nutshell it's just another sparse voxel data structure. My implementation is, no doubt, different from Epic Games Store's own. I'm using 4x4x4 probe grid with intermediate nodes having very wide branching factor of 64 as well (4x4x4). I liked the parameters that Unreal is using, of limiting both total memory as well as the lowest level of detail, which is common in sparse grid implementations. Here's Bistro scene with just 1Mb limit. This is roughly equivalent to a 512x512 lightmap texture in 2d, except surface light maps require unique UVs and you typically get very little detail out of 512 resolution texture with a lot of light leaking. There is also no directional response. My implementation encodes second-order spherical harmonics for each probe (9 coefficients), encoding RGB channels as RGBE9995 (4 bytes). So far only worked on the structure, actual bake is yet to come. I've been eyeing sparse voxel structures for a while now, and have been studying them roughly since the GigaVoxel paper by Cyril Crassin but never really implemented anything for the GPU before. I was always the BVH-kind of guy. It's a fascinating topic. --- Stats for the scene: --- Total memory usage: 1.000 MB Node count: 609 Unique probe count: 24,025 Probe reuse: 38.36 % Unexpanded nodes: 15,714 --- Again, note that there is no GI going on here, only the structure of the probe tree and the algorithm for building it from a given scene.

Working on Sparse Volumetric Light-maps. Thanks to CynicatPro🎃 for pointing me at Unreal's version. In a nutshell it's just another sparse voxel data structure. My implementation is, no doubt, different from Epic Games Store's own. I'm using 4x4x4 probe grid with intermediate nodes having very wide branching factor of 64 as well (4x4x4). I liked the parameters that Unreal is using, of limiting both total memory as well as the lowest level of detail, which is common in sparse grid implementations. Here's Bistro scene with just 1Mb limit. This is roughly equivalent to a 512x512 lightmap texture in 2d, except surface light maps require unique UVs and you typically get very little detail out of 512 resolution texture with a lot of light leaking. There is also no directional response. My implementation encodes second-order spherical harmonics for each probe (9 coefficients), encoding RGB channels as RGBE9995 (4 bytes). So far only worked on the structure, actual bake is yet to come. I've been eyeing sparse voxel structures for a while now, and have been studying them roughly since the GigaVoxel paper by Cyril Crassin but never really implemented anything for the GPU before. I was always the BVH-kind of guy. It's a fascinating topic. --- Stats for the scene: --- Total memory usage: 1.000 MB Node count: 609 Unique probe count: 24,025 Probe reuse: 38.36 % Unexpanded nodes: 15,714 --- Again, note that there is no GI going on here, only the structure of the probe tree and the algorithm for building it from a given scene.

Alex Goldring

11,487 次观看 • 4 个月前

1 million unique objects in WebGPU. It appears to be the fashion to show off limits of one's engine. So here's "Shade" - my WebGPU engine running in the browser, rendering 1,000,000 unique meshes. That is - no instancing, and every mesh has a unique material, geometry and transform. We have full cascaded shadows here in the mix and post-processing stack. "Surely you can't have 1M instances, there's a trick, right?" - Nope. No tricks, just the state-of-the art Vizibility Buffer-based GPU-resident renderer. That is - on the CPU side we submit a fixed number of draw calls regardless of how many instances we have.

1 million unique objects in WebGPU. It appears to be the fashion to show off limits of one's engine. So here's "Shade" - my WebGPU engine running in the browser, rendering 1,000,000 unique meshes. That is - no instancing, and every mesh has a unique material, geometry and transform. We have full cascaded shadows here in the mix and post-processing stack. "Surely you can't have 1M instances, there's a trick, right?" - Nope. No tricks, just the state-of-the art Vizibility Buffer-based GPU-resident renderer. That is - on the CPU side we submit a fixed number of draw calls regardless of how many instances we have.

Alex Goldring

86,497 次观看 • 5 个月前

Chaos and the double pendulum: the configuration space of the double pendulum is a torus (a circle for each angle coordinate) so we can plot the evolution of such a pendulum in two ways: either as a "real pendulum" in physical space, or as a point in configuration space :)

Chaos and the double pendulum: the configuration space of the double pendulum is a torus (a circle for each angle coordinate) so we can plot the evolution of such a pendulum in two ways: either as a "real pendulum" in physical space, or as a point in configuration space :)

Steve Trettel

149,856 次观看 • 2 年前

Sparse Volumetric Light maps (WebGPU). Worked on ensuring C0 continuity across the entire map. For now this is achieved by purging incomplete node levels, works well enough even if it's a bit of a blunt tool. Actual locations for probes during the bake are now going through an optimization phase, which allows me to push probes behind surfaces out into the open resulting in much fewer light leakage artifacts. This video here has only 28,376 probes in the map, and takes 1.8Mb of VRAM

Sparse Volumetric Light maps (WebGPU). Worked on ensuring C0 continuity across the entire map. For now this is achieved by purging incomplete node levels, works well enough even if it's a bit of a blunt tool. Actual locations for probes during the bake are now going through an optimization phase, which allows me to push probes behind surfaces out into the open resulting in much fewer light leakage artifacts. This video here has only 28,376 probes in the map, and takes 1.8Mb of VRAM

Alex Goldring

15,814 次观看 • 4 个月前

[NeurIPS '24] DreamMesh4D: Video-to-4D Generation with Sparse-Controlled Gaussian-Mesh Hybrid Representation Abstract (excerpt) We introduce DreamMesh4D, a novel framework that combines mesh representation with sparse-controlled deformation technique to generate high-quality 4D object from a monocular video. To overcome the limitation of classical texture representation, we bind Gaussian splats to the surface of the triangular mesh for differentiable optimization of both the texture and mesh vertices. In particular, DreamMesh4D begins with a coarse mesh provided by a single image based 3D generation method. Sparse points are then uniformly sampled across the surface of the mesh, and are used to build a deformation graph to drive the motion of the 3D object for the sake of computational efficiency and providing additional constraint. For each step, transformations of sparse control points are predicted using a deformation network, and the mesh vertices as well as the bound surface Gaussians are deformed via a geometric skinning algorithm. The skinning algorithm is a hybrid approach combining LBS (linear blending skinning) and DQS (dual-quaternion skinning), mitigating drawbacks associated with both approaches. The static surface Gaussians and mesh vertices as well as the dynamic deformation network are learned via reference view photometric loss, score distillation loss as well as other regularization losses in a two-stage manner. Extensive experiments demonstrate that our method outperforms prior video-to-4D generation methods in terms of rendering quality and spatial-temporal consistency.

[NeurIPS '24] DreamMesh4D: Video-to-4D Generation with Sparse-Controlled Gaussian-Mesh Hybrid Representation Abstract (excerpt) We introduce DreamMesh4D, a novel framework that combines mesh representation with sparse-controlled deformation technique to generate high-quality 4D object from a monocular video. To overcome the limitation of classical texture representation, we bind Gaussian splats to the surface of the triangular mesh for differentiable optimization of both the texture and mesh vertices. In particular, DreamMesh4D begins with a coarse mesh provided by a single image based 3D generation method. Sparse points are then uniformly sampled across the surface of the mesh, and are used to build a deformation graph to drive the motion of the 3D object for the sake of computational efficiency and providing additional constraint. For each step, transformations of sparse control points are predicted using a deformation network, and the mesh vertices as well as the bound surface Gaussians are deformed via a geometric skinning algorithm. The skinning algorithm is a hybrid approach combining LBS (linear blending skinning) and DQS (dual-quaternion skinning), mitigating drawbacks associated with both approaches. The static surface Gaussians and mesh vertices as well as the dynamic deformation network are learned via reference view photometric loss, score distillation loss as well as other regularization losses in a two-stage manner. Extensive experiments demonstrate that our method outperforms prior video-to-4D generation methods in terms of rendering quality and spatial-temporal consistency.

MrNeRF

12,323 次观看 • 1 年前

Virtual Textures in the Browser. This is a WebGL 2.0 implementation I did in 2023. The car model is from Sketchfab by Pavel Matoušek. It's a worst-case for virtual textures because the texture is generated from photogrametry software and has pathologically bad UV layout. The texture is 16,384 x 16,384 pixels, which is 180 Mb of PNG data. Loading this in Blender would crash Blender for me back then. The GPU memory requirement for this texture is 1 Gb. The VT displays the model after loading just 34 Kb (one tile). The other tiles are fetched in via a smart priority queue in the background as you move camera around the scene. The physical page texture is 2k, which is 16 times smaller than the source. The solution supports multiple separate virtual textures, and has GLTF integration. Texture filtering as well as anisotropy are fully supported as well. You can try the demo yourself here: Here's a writeup I did while working on this: The video recorded on a very crummy internet connection and I intentionally reload the page in the middle. The delay comes from the geometry download, actual texture part is near-instant.

Virtual Textures in the Browser. This is a WebGL 2.0 implementation I did in 2023. The car model is from Sketchfab by Pavel Matoušek. It's a worst-case for virtual textures because the texture is generated from photogrametry software and has pathologically bad UV layout. The texture is 16,384 x 16,384 pixels, which is 180 Mb of PNG data. Loading this in Blender would crash Blender for me back then. The GPU memory requirement for this texture is 1 Gb. The VT displays the model after loading just 34 Kb (one tile). The other tiles are fetched in via a smart priority queue in the background as you move camera around the scene. The physical page texture is 2k, which is 16 times smaller than the source. The solution supports multiple separate virtual textures, and has GLTF integration. Texture filtering as well as anisotropy are fully supported as well. You can try the demo yourself here: Here's a writeup I did while working on this: The video recorded on a very crummy internet connection and I intentionally reload the page in the middle. The delay comes from the geometry download, actual texture part is near-instant.

Alex Goldring

19,368 次观看 • 5 个月前

Was working on CSM (Cascaded Shadow Maps) blending. My CSM implementation is a little unusual, I switch cascades as late as possible, using the highest available resolution cascade. Most CSM implementations choose cascades based on view depth, the problem with that is - you waste a lot of high resolution shadow texels, in my experience easily upwards of 50%. I figure - GPU already spent the effort to compute those shadow texels - why not use them, and get perceptually about 20-50% resolution increase in your shadows. This is not new, and MJP showed this in his code too (see below). The problem is in blending, if you use cascade projection matrix for choosing a cascade - blending becomes non-trivial. So I spent a lot of time working on that, and finally cracked it. Video shows blending of 5 cascades, the blend margin is exaggerated for demonstration purposes and cascades are pushed to the near plane for the same reason. The result is a perfect forward-only blend at the cost of a bit of ALU. Matt Pettineo is pretty much an authority on the subject nowadays, and I found his amazing repo very useful in the past:

Was working on CSM (Cascaded Shadow Maps) blending. My CSM implementation is a little unusual, I switch cascades as late as possible, using the highest available resolution cascade. Most CSM implementations choose cascades based on view depth, the problem with that is - you waste a lot of high resolution shadow texels, in my experience easily upwards of 50%. I figure - GPU already spent the effort to compute those shadow texels - why not use them, and get perceptually about 20-50% resolution increase in your shadows. This is not new, and MJP showed this in his code too (see below). The problem is in blending, if you use cascade projection matrix for choosing a cascade - blending becomes non-trivial. So I spent a lot of time working on that, and finally cracked it. Video shows blending of 5 cascades, the blend margin is exaggerated for demonstration purposes and cascades are pushed to the near plane for the same reason. The result is a perfect forward-only blend at the cost of a bit of ALU. Matt Pettineo is pretty much an authority on the subject nowadays, and I found his amazing repo very useful in the past:

Alex Goldring

18,775 次观看 • 5 个月前

Vehicle-combat-game-dev-day 484: GPU instanced grass. Only one mesh for now, but the rendering system is robust and supports a LOT of instances and is fully procedural, generating grass on the fly! The game's levels are fully RNG & procedural #indiegamedev #solodev

Vehicle-combat-game-dev-day 484: GPU instanced grass. Only one mesh for now, but the rendering system is robust and supports a LOT of instances and is fully procedural, generating grass on the fly! The game's levels are fully RNG & procedural #indiegamedev #solodev

Jussi Kemppainen | Drivers of the Apocalypse

11,854 次观看 • 4 个月前

Solo Through America's Wild West Coastline 🇺🇸 California's coastline is one of the most beautiful in the world. There is a section where the coastal HWY 1 goes inland and leaves a part of this rugged coastline and communities on their own. Join me today as we road trip through one of the most beautiful parts of the world and meet up with some fascinating locals who call it home.

Solo Through America's Wild West Coastline 🇺🇸 California's coastline is one of the most beautiful in the world. There is a section where the coastal HWY 1 goes inland and leaves a part of this rugged coastline and communities on their own. Join me today as we road trip through one of the most beautiful parts of the world and meet up with some fascinating locals who call it home.

Peter Santenello

11,671 次观看 • 1 年前

$This is only the beginning. 🔊 Sound On What does it mean to explore a house built from memory — one that feels familiar, yet shifts each time you return? That question is at the heart of Memory House, a project that unfolds through space, silence, and surreal interruption. Grateful to the World Labs team for making this experiment possible — their latest tool Marble opens up a new way of building. With Gaussian splatting, a single 2D image becomes a 3D space you can step inside. From there, entire environments can grow by stitching rooms together, layering ambience, proximity, and subtle triggers for immersion. The teaser captures just the start of an exploration in narrative: waking up in bed, following the quiet sounds of home, and stepping into a kitchen where reality begins to fracture. But that’s only one fragment of a project now expanding into 30+ connected rooms. With Worldlabs, alongside tools like Veo3 and ElevenLabs, Memory House is evolving into a new form of storytelling — where memory isn’t told on a page, but experienced through space.$

This is only the beginning. 🔊 Sound On What does it mean to explore a house built from memory — one that feels familiar, yet shifts each time you return? That question is at the heart of Memory House, a project that unfolds through space, silence, and surreal interruption. Grateful to the World Labs team for making this experiment possible — their latest tool Marble opens up a new way of building. With Gaussian splatting, a single 2D image becomes a 3D space you can step inside. From there, entire environments can grow by stitching rooms together, layering ambience, proximity, and subtle triggers for immersion. The teaser captures just the start of an exploration in narrative: waking up in bed, following the quiet sounds of home, and stepping into a kitchen where reality begins to fracture. But that’s only one fragment of a project now expanding into 30+ connected rooms. With Worldlabs, alongside tools like Veo3 and ElevenLabs, Memory House is evolving into a new form of storytelling — where memory isn’t told on a page, but experienced through space.

The Artist's Journey

19,085 次观看 • 9 个月前

Hannatu: The fact that there is no arena that produces Afrobeats is quite a shame. That there is only one 5-star hotel in the capital of Nigeria, Abuja, the Hilltop Hotel, is also quite a shame. Abuja should have 5 or 6 world-class hotels, and we are working on about 3 of them in Abuja and 2 in Lagos.

Hannatu: The fact that there is no arena that produces Afrobeats is quite a shame. That there is only one 5-star hotel in the capital of Nigeria, Abuja, the Hilltop Hotel, is also quite a shame. Abuja should have 5 or 6 world-class hotels, and we are working on about 3 of them in Abuja and 2 in Lagos.

Imran Muhammad

195,409 次观看 • 4 个月前

[SIGGRAPH '25] TeGA: Texture Space Gaussian Avatars for High-Resolution Dynamic Head Modeling Note: On the left that's a 3DGS rendering! Contributions: 1. We propose a simple approach for rigging 3D Gaussians within the continuous tangent space of 3DMM face models, allowing Gaussians to move freely across mesh triangles. 2. We propose a novel CNN-based deformation model that is agnostic to the number of 3D Gaussians, naturally enabling adaptively densification of the representation to improve detail where most needed, with expression-dependent shading. 3. We show significant improvements over baseline SOTA methods and demonstrate the ability to render even extreme close-up images at high quality.

[SIGGRAPH '25] TeGA: Texture Space Gaussian Avatars for High-Resolution Dynamic Head Modeling Note: On the left that's a 3DGS rendering! Contributions: 1. We propose a simple approach for rigging 3D Gaussians within the continuous tangent space of 3DMM face models, allowing Gaussians to move freely across mesh triangles. 2. We propose a novel CNN-based deformation model that is agnostic to the number of 3D Gaussians, naturally enabling adaptively densification of the representation to improve detail where most needed, with expression-dependent shading. 3. We show significant improvements over baseline SOTA methods and demonstrate the ability to render even extreme close-up images at high quality.

MrNeRF

28,434 次观看 • 1 年前

"The Promise" was written and recorded in a small garden shed turned into a studio at the home of vocalist Clive Farrington's father. The space was so tiny that they had to attach the keyboards vertically to the walls to fit everything! Released in 1988 as the debut single for When In Rome, it became #1 on the American Dance Chart, #1 on the Billboard Hot 100 and one of the biggest one-hit wonders of the 80s.

"The Promise" was written and recorded in a small garden shed turned into a studio at the home of vocalist Clive Farrington's father. The space was so tiny that they had to attach the keyboards vertically to the walls to fit everything! Released in 1988 as the debut single for When In Rome, it became #1 on the American Dance Chart, #1 on the Billboard Hot 100 and one of the biggest one-hit wonders of the 80s.

Ladytron Fan Account

128,755 次观看 • 1 个月前

🚨 NOW: President Trump takes huge VICTORY LAP on the success of Space Force — which some Democrats MOCKED Space Force is now vastly improving the accuracy of our weapons on the ground. And we're leading China and Russia, easily 🔥 "I'm very proud of the Space Force because that was a branch that a new one that came up, and we were losing to Russia and to China in space." "Space Force is now leading the other countries, China, Russia, and there is no fourth, really, and leading them by, we saw numbers the other day, leading, we're leading in space by so much it's, we're never going to be caught. So I'm very proud of Space Force!" "I think it's going to be one of our most important as time goes by. A lot of the things that you see, a lot of the accuracy of the weapons is caused because of what we created with Space Force, so we're very proud of that."

🚨 NOW: President Trump takes huge VICTORY LAP on the success of Space Force — which some Democrats MOCKED Space Force is now vastly improving the accuracy of our weapons on the ground. And we're leading China and Russia, easily 🔥 "I'm very proud of the Space Force because that was a branch that a new one that came up, and we were losing to Russia and to China in space." "Space Force is now leading the other countries, China, Russia, and there is no fourth, really, and leading them by, we saw numbers the other day, leading, we're leading in space by so much it's, we're never going to be caught. So I'm very proud of Space Force!" "I think it's going to be one of our most important as time goes by. A lot of the things that you see, a lot of the accuracy of the weapons is caused because of what we created with Space Force, so we're very proud of that."

Eric Daugherty

15,034 次观看 • 1 个月前

I am very bullish on the Luxury space in India. The space is going to quadruple in the next 5 years and is one of the fastest growing spaces in India. I talk about a few plays in the space in this video.

I am very bullish on the Luxury space in India. The space is going to quadruple in the next 5 years and is one of the fastest growing spaces in India. I talk about a few plays in the space in this video.

Rajarshi Shome

16,926 次观看 • 5 个月前

🚨BREAKING: Trump honors Sylvester Stallone and uses the time to brag about himself: “Sly is one of the biggest names on the Hollywood Walk Of Fame. In fact, the only one that's a bigger name on the Hollywood Walk Of Fame, they say, is a guy named Donald Trump.” What a sad loser.

🚨BREAKING: Trump honors Sylvester Stallone and uses the time to brag about himself: “Sly is one of the biggest names on the Hollywood Walk Of Fame. In fact, the only one that's a bigger name on the Hollywood Walk Of Fame, they say, is a guy named Donald Trump.” What a sad loser.

CALL TO ACTIVISM

552,170 次观看 • 10 个月前

"You really want to get your mind blown? Look. Type up in Google or any search engine the words 'pale blue dot.'" "When we sent out the Voyager spacecraft on one of its last commands, we told it to turn around and take a picture of home—of Earth. Everything you have ever known, you've ever learned, everything you ever loved, everything you ever feared, every person that ever lived in history, every person that ever died, ever—is all on that tiny little blue dot hurtling through the vacuum and inky blackness of space." "Now, that's uncomfortable for some people to think about. And then when you think about the size of the universe…" Lue Elizondo

"You really want to get your mind blown? Look. Type up in Google or any search engine the words 'pale blue dot.'" "When we sent out the Voyager spacecraft on one of its last commands, we told it to turn around and take a picture of home—of Earth. Everything you have ever known, you've ever learned, everything you ever loved, everything you ever feared, every person that ever lived in history, every person that ever died, ever—is all on that tiny little blue dot hurtling through the vacuum and inky blackness of space." "Now, that's uncomfortable for some people to think about. And then when you think about the size of the universe…" Lue Elizondo

Shawn Ryan

740,320 次观看 • 1 年前

Kaikai: The current political class is in a very difficult space. We are now dealing with a younger generation that is more educated than the generation that is in charge of the political responsibilities. The youth view both the opposition and government as one bunch that doesn’t understand them. Therefore, the call for dialogue will be a tough one #CitizenNewsGang Linus Kaikai

Kaikai: The current political class is in a very difficult space. We are now dealing with a younger generation that is more educated than the generation that is in charge of the political responsibilities. The youth view both the opposition and government as one bunch that doesn’t understand them. Therefore, the call for dialogue will be a tough one #CitizenNewsGang Linus Kaikai

Citizen TV Kenya

20,369 次观看 • 11 个月前

Moving on is the hardest part. This is about losing a loved one and dreaming of them after they're gone. But memories are subjective, distorted mirrors of reality. That's what's in the latent space. The Hardest Part by Washed Out Made with Sora

Moving on is the hardest part. This is about losing a loved one and dreaming of them after they're gone. But memories are subjective, distorted mirrors of reality. That's what's in the latent space. The Hardest Part by Washed Out Made with Sora

Paul Trillo

65,207 次观看 • 2 年前

A rendered timelapse of thousands of photos processed into a single ~30 sec video. The light play on the scene was unreal and the clouds ended up parting quite nicely to reveal the massive snow caps. One of my favorite timelapses of all time

A rendered timelapse of thousands of photos processed into a single ~30 sec video. The light play on the scene was unreal and the clouds ended up parting quite nicely to reveal the massive snow caps. One of my favorite timelapses of all time

Brandon Yoshizawa

16,472 次观看 • 3 个月前