Update on Futuristic Buildings

Update on my previous post: I brought back my beloved car character from Unbiased Truck Soccer. Some images (rendered blazingly fast at 16 spp per frame):





One 720p image (16 spp, almost no noise to be found):

I still have to create a tight fitting bounding box around the car to increase the path tracing performance (although the performance is still quite good without it). The source code of Simplex Paternitas also provides examples of rotation matrices and scripted camera animation sequences, so there are lots of cool things left to explore. The car will eventually be user controllable with a camera following the car and in contrast to my previous demos featuring the car (Unbiased Truck Soccer etc.), it will also be able to rotate instead of just translate.


UPDATE: The rotation of the car is now working. Below is a video that was rendered on my 8600M GT with only 4 samples per pixel per frame (framerate is about 5 fps at 640x480 resolution).

There still isn't an acceleration structure for the car (in this case an axis aligned bounding box), which degrades the path tracing performance a little, but the code for should be done soon. After that I'm going to make the car user controllable. There's also a new 'futuristic building' which degrades performance even more, so it will need a hitbox as well:

"Futuristic Buildings" a new real-time path traced demo

I've decided to put my freshly learned C++ skills to the test, and what better way is there than modifying the source code of an already awesome program, in this case the recently released real-time path tracer "Simplex Paternitas" by Jacco Bikker. For "Futuristic Buildings" I modified some of the animation from Simplex Paternitas, added some new animations and also a new "skyscraper" building with moving spheres.

The purpose of this demo was to emphasize the extraordinary capabilities of this path tracer, in particular the ultra-high-quality global illumination with color bleeding and dynamic soft shadows at real-time framerates. The demo renders almost noise-free images at 16 samples per pixel and should render them very fast when you have a high end GPU (>30 fps I guess, I only tested it with a 8600 M GT).

Some images:









Download "Futuristic Buildings" at

http://code.google.com/p/tokap-the-once-known-as-pong/downloads/list

A CUDA enabled GPU is required as the path tracing happens entirely on the GPU.

This is work in progress of course, I'm planning to add another building and a user-controllable character (the truck from Unbiased Truck Soccer :) . A video should follow soon.

Simplex Paternitas, a real-time path traced animation by Jacco Bikker

Jacco Bikker (developer of the Brigade path tracer and Arauna ray tracer) has released a cool video and executable demo of a real-time path traced animation called "Simplex Paternitas" created for iGathering 2011 (a demoscene festival organised at IGAD):

http://www.youtube.com/watch?v=PaTz9tJ7_KY

The demo in the video runs smoothly on just one GTX 470. Links to the executable demo and source code can be found in the video description. I've tried the 32 spp demo on my laptop with 8600M GT and got around 2 fps! :)


Soft shadows, refraction and caustics


Accurate color bleeding from the blue and red walls is clearly visible in the shadows of the characters and reflective sphere

The animation reminds me of vintage CG from the eighties (like this one) which was rendered with either rasterization or Whitted-style ray tracing and needed several CPU hours per frame, while this is path traced in just a few milliseconds per frame.

UPDATE: I tinkered a bit with the scene parameters (different colors and materials and brighter sky). The image is taken at 28 samples per pixel. There is some slight greenish color bleeding from the ground onto the bottom half of the spheres and the character (not Kirby :) I will upload a video very soon. Even at 16 spp, there is almost no noise in the outdoor scene except for refractive spheres.

Trailer for Reflect, the first real-time path traced game

A video of the Portal-inspired game "Reflect", which uses the Brigade path tracer has surfaced on Youtube (sassy soundtrack included):

http://www.youtube.com/watch?v=qntMz2QPA_E

The game is running on 2 Xeon (quad core) CPUs and 2 GTX 460s in this video. The number of indirect light bounces seems rather low, but the mere fact that path tracing can be done at playable framerates and reasonable noise levels is quite a breakthrough. Real-time path tracing for games is finally becoming reality, and this is just the beginning. Congratz to the team behind Reflect!

UPDATE: there's another video of Reflect which is slightly less noisy: http://www.youtube.com/watch?v=z27EpDJprQg

EGSR 2011 paper on real-time noise-free GPU path tracing

Just read an interesting paper called "Guided Image Filtering for Interactive High-quality Global Illumination" by Bauszat, Eisemann and Magnor, which was accepted at EGSR 2011. The paper describes a novel method for edge-aware filtering of noisy path traced images using the normal and depth buffers instead of the noise image itself. The indirect lighting is filtered separately from the direct lighting and the whole pipeline (path tracing + filtering) is implemented on the GPU using CUDA (the numbers in the paper are obtained with a GTX 285). The new method is compared to cross bilateral filtering and the edge-avoiding à-trous wavelet filter described in a HPG paper from last year and the results seem very promising.

I very much agree with the introduction to the paper:

"...many approximations exist to reproduce certain effects of the real global illumination and try to incorporate it into a classic rasterizer. Unfortunately, in order to speed up the computation they are either aiming at simulating only a single effect, like soft shadows [ED08] or ambient occlusion [RGS09] or they are are limited to certain constraints, requiring low-frequency lighting environments [SKS02], static scenes [LSK07] or precise parameter adjustment for varying scenes [RGK08]. Code complexity increases tremendously for systems that simulate global illumination effects using rasterization because multiple approximations need to be combined to achieve a sufficient visual quality. On the other hand path tracing naturally incorporates all these effects with a very simple rendering algorithm.

Another drawback of approximative algorithms is that they not necessarily converge to the correct result. Therefore, they will only become faster with better hardware but do not necessarily produce a higher quality result. Instead of trying to incorporate global illumination effects into a rendering pipeline which is not capable of physically correct rendering, it is a much more promising design to take a physically valid rendering approach and implement approximative algorithms on top to speed up the rendering. Designing them in an adjustable way results in a very future-oriented rendering paradigm."

Research in fast interactive GPU path tracing with minimal noise has made enormous advancements over the past year and I'm confident that real-time high-quality path traced global illumination for games will be feasible soon.

Stunning animation rendered with Octane Render

Just saw this very impressive video on the Octane forum today and thought it deserved its own post (the level of realism is otherworldly):

http://vimeo.com/25686881

The whole scene is 3D. Rendertime was two minutes per frame on average on a single GTX480, so eight of these GPUs rendering simultaneously would reduce the rendertime to just 15 seconds per frame (rendertimes in Octane scale almost linearly with the number of GPUs), which is just completely nuts considering the quality.

Update: there's a new video on youtube showing that Octane can produce a real-time rendered low resolution preview animation of an outdoor scene with a rendertime of only 55 milliseconds per frame!

Accelerating Path Tracing by Eye Path Reprojection

Just stumbled upon this upcoming paper about interactive GPU path tracing from Niklas Henrich. The abstract sounds very promising:

"Recently, path tracing has gained interest for real-time global illumination since it allows to simulate an unbiased result of the rendering equation. However, path tracing is still too slow for real-time applications and shows noise when displayed at interactive frame rates. The radiance of a pixel is computed by tracing a path, starting from the eye and connecting each point on the path with the light source. While conventional path tracing uses the information of a path for a single pixel only, we demonstrate how to distribute the intermediate results along the path to other pixels in the image. We show that this reprojection of an eye path can be implemented efficiently on graphics hardware with only a small overhead. This results in an overall improvement of the whole image since the number of paths per pixel increases. The method is especially useful for many indirections, which is often circumvented to save computation time. Furthermore, instead of improving the quality of the rendering, our method is able to increase the rendering speed by reusing the reprojected paths instead of tracing new paths while maintaining the same quality."

Hopefully, the results from "Accelerating path tracing by re-using paths" (a paper from 2002 by Bekaert et al. who reported a 9x speed-up for CPU path tracing) can be replicated on the GPU.

Sparse voxel octree with real-time global illumination and dynamic geometry

Cyril Crassin has posted a very nice video on his blog at http://blog.icare3d.org/2011/06/interactive-indirect-illumination-and.html showing the latest developments in his sparse voxel octree research which will be presented at Siggraph 2011 in Vancouver.

Major improvements on previously published results (http://artis.imag.fr/Membres/Cyril.Crassin/) include real-time indirect lighting (for diffuse and glossy materials) and support for fully dynamic voxel objects (by fast mesh voxelization and updating the voxel octree in real-time). There has been research in animated sparse voxel octrees before (using rasterization, see http://bautembach.de/wordpress/?page_id=7, http://www.youtube.com/watch?v=Tl6PE_n6zTk, http://www.youtube.com/watch?v=Hnvr0hxyDvk and http://www.youtube.com/watch?v=gNZtx3ijjpo, which doesn't look as detailed as Jon Olick's raycasted static sparse voxel octree tech from Siggraph 08), but this is the first time it's being done with ray casting (cone tracing actually).

The global illumination algorithm resembles photon mapping: instead of photon tracing, the scene is rasterized from the perspective of the light source and radiance is stored in the octree, followed by filtering in screen-space and a final gathering step using approximate cone tracing. More details: http://artis.imag.fr/Publications/2011/CNSGE11a/GIVoxels_Siggraph_Talk.pdf

HPG paper and video of "Improving SIMD Efficiency for Parallel Monte Carlo Light Transport on the GPU"!

Another post on the work of Dietger van Antwerpen, but that's because his research efforts for GPU path tracing are so damn amazing and groundbreaking. I can't stop imagining what the possibilities could be when implemented in a real-time path tracer for games. His HPG2011 paper "Improving SIMD Efficiency for Parallel Monte Carlo Light Transport on the GPU" and an accompanying video are available here: http://graphics.tudelft.nl/~dietger/HPG2011/index.html.



The video shows a side-by-side comparison between standard path tracing (PT), bidirectional path tracing (BDPT) and Metropolis light transport (MLT) in a number of scenes with complex lighting. The difference in efficiency of the MLT and BDPT algorithms compared to regular path tracing is huge: for example, the kitchen scene (indirect lighting, actually it's lit by light passing through a lens, essentially a caustic) converges extremely slowly with PT and is still mostly pitch black (because the probability of a random path hitting a small light source behind a lens is very low), while BDPT (paths starting simultaneously from the camera and the light source, "meeting each other half way") and MLT (paths starting randomly, but once an important light contributing path is found, nearby paths are explored) both do a much better job and show a recognizable scene in a matter of milliseconds. All these algorithms are running completely on the GPU with almost zero CPU load. The end of the video shows a flooded Cornell box scene rendered with MLT at interactive rates, showing that the caustic light pattern on the floor converges very fast (something that would take a very long time for a regular path tracer). The last page of the HPG 2011 paper contains a very interesting comparison between a fully converged reference image and a PT, BDPT and MLT image after just 30 seconds on a GTX 480.

An often recurring criticism of using the GPU for rendering is that for scenes with complex lighting and materials, the CPU outperforms the GPU because the CPU is able to use smarter and more efficient algorithms (like MLT and BDPT) while the GPU is only efficient at regular, "dumb" path tracing. This is no longer true as proven by this paper: not only can GPUs use more efficient rendering algorithms like BDPT and MLT, they can also do it an order of magnitude faster than the CPU! The table comparing performance between GPU rendering and CPU rendering (GTX 480 vs Core i7 920) demonstrates that, depending on the scene, PT is 10-18x, BDPT 8-15x and MLT 9-15x faster on the GPU than on the CPU!!

It's amazing to see that the still very young field of physically based GPU rendering has made such tremendous advancements in just a couple of months (thanks to guys like Dietger among others). I can't wait to see where this field is going to head during the next year with better hardware (Nvidia Kepler, AMD Graphics Core Next) and even more optimized and efficient algorithms. It really is mind-boggling...

Dietger van Antwerpen's thesis "Unbiased physically based rendering on the GPU" available!

The thesis from Dietger van Antwerpen, co-developer of the Brigade path tracer who pioneered bidirectional path tracing (BDPT), energy redistribution ray tracing (ERPT) and Metropolis light transport (MLT) in CUDA is finally available (since yesterday) at http://repository.tudelft.nl/view/ir/uuid%3A4a5be464-dc52-4bd0-9ede-faefdaff8be6/

It's 180 pages and this is just a master thesis, nuts! :)

Update on CentiLeo, the out-of-core GPU ray tracer

CentiLeo, the awesome out-of-core interactive GPU path tracer for massive models (like the 400 million polygon Boeing 777 model) that I've blogged about in this post, is going to be presented this summer at Siggraph:

http://www.siggraph.org/s2011/content/out-core-gpu-ray-tracing-complex-scenes

According to the video on youtube "CentiLeo implementation uses CUDA and is based on Kirill Garanzha's PhD research in Keldysh Institute of Applied Mathematics, Russian Academy of Sciences."

As mentioned in the last post, Kirill Garanzha will also present a paper entitled "Simpler and Faster HLBVH with Work Queues" at High Performance Graphics 2011, which aims to make real-time ray tracing of highly dynamic scenes possible by fully rebuilding (instead of refitting which significantly degrades ray traversal performance) the BVH from scratch in real-time. Combined with the out-of-core path tracing tech from CentiLeo, this could be very compelling and make photoreal animations of highly dynamic multi-million polygon scenes render very fast on the GPU.

Update: This week, Garanzha also presented a paper at the Computer Graphics International 2011 conference on a novel way to build high quality BVH's on the GPU (quality measured by GPU path tracing on a GTX 480) with the title "Grid-based SAH BVH construction on a GPU" (behind a paywall). The BVH refitting is faster than the original HLBVH implementation.

HPG 2011 paper list online

The Real-Time Rendering blog just blogged that Kesen Huang has put the list of papers for the High Performance Graphics 2011 symposium online at http://kesen.realtimerendering.com/hpg2011Papers.htm. HPG 2011 takes place in Vancouver (home of the Canucks ;) on Aug 5-7.

These are some of the more juicy titles on the subject of ray tracing and path tracing that immediately catch the attention:

Simpler and Faster HLBVH with Work Queues (Kirill Garanzha, Jacopo Pantaleoni, David McAllister) I've been waiting an eternity for Pantaleoni's HLBVH code to be open sourced, so I hope the code from this new paper will get published eventually.

Improving SIMD Efficiency for Parallel Monte Carlo Light Transport on the GPU (Dietger van Antwerpen) This paper is from the co-developer of the real-time path tracer called Brigade, who also implemented Metropolis light transport and ERPT on the GPU, so it should be really interesting.

Active Thread Compaction for GPU Path Tracing (Ingo Wald) From the guy that pioneered real-time CPU ray tracing.

Some other attention-grabbing titles:

Real-Time Diffuse Global Illumination Using Radiance Hints (Georgios Papaioannou)

VoxelPipe: A Programmable Pipeline for 3D Voxelization (Jacopo Pantaleoni)

High-Performance Software Rasterization on GPUs (Samuli Laine, Tero Karras)

MSBVH: An Efficient Acceleration Data Structure for Ray Traced Motion Blur (Leonhard, Gruenschloss, Martin Stich, Sehera Nawaz, Alexander Keller)

The program looks very promising so far, especially for real-time and interactive ray tracing and path tracing.

UPDATE:

The paper by McGuire et al. on the Alchemy Screen Space Ambient Obscurance Algorithm contains a pretty neat idea, which could prove to be very useful for real-time path tracing. The idea is to adapt the number of samples per pixel with distance from the camera, essentially some kind of LOD scheme for the spp number as shown in this picture (taken from the paper):

CBox Unbiased Truck

I've modified the scene in the Kajiya path tracer a bit more: it now consists of a Cornell Box out of axis aligned boxes with the (in)famous truck from Unbiased Truck Soccer:

Color bleeding from the red and green wall:

The screenshots were rendered with 8600M GT (6 fps default view). On a GTS 450, the demo runs at 70 fps in default view. It should run at >200 fps on a GTX 580 with 8 samples per pixel. This new path tracer is just incredible fun, I can't stop messing with it.

Executable and source code at http://code.google.com/p/tokap-the-once-known-as-pong/downloads/list

UPDATE: a more challenging lighting set up with an open box only illuminated by the sky:

The truck seen from behind, indirectly lit by skylight bounced off the back and side walls. As expected with standard path tracing, the noise is a lot worse in this scenario. Bidirectional path tracing should converge faster using fewer samples.

Arnold Render impressions from FMX 2011

http://www.3dworldmag.com/2011/05/04/sony-pictures-imageworks-unbiased-raytracing-in-feature-films-now-viable/

More details by Marcos Fajardo: http://www.cgchannel.com/2011/05/fmx-2011-tuesdays-highlights/:

Having showed a 400-million-polygon asset from G-Force as proof that even the heaviest scenes can be raytraced in production, Fajardo singled out volumetrics as a particular area of benefit.

“[With a traditional hybrid Reyes renderer] you’re losing the interaction of the raytrace effects with the volumes so even simple things like shadows become a problem,” he said. “[With Arnold] it just works. There’s no ‘gotcha’.”

The move to fully ray traced rendering in the movie industry (f.e. Sony Pictures Imageworks and Blue Sky) has started and it's happening really fast. Now it's up to the game industry to do the same.

Real-time path traced Unbiased Truck Soccer moved to Kajiya-engine

It has been a while since the last update on Unbiased Truck Soccer, the real-time path traced game featuring a truck made of spheres. Two major changes happened since last time: I've upgraded from my laptop with 8600M GT to a desktop with GTS 450, not a high end card but a fair bit better than the ultra-low end 8600M GT (I'm still using the latter for developing and testing new ideas, because if it looks and runs OK on crappy hardware, you can be sure that it will look great on a real GPU). I've also started porting the Unbiased Truck game to the brand new and recently released Kajiya scene path tracer from Jacco Bikker and Jeroen van Schijndel. This decision was made for several reasons:

- when comparing path tracing performance on the 8600M GT and the GTS450, TOKAP runs only 5x faster while the Kajiya path tracer runs 11x faster. It's also optimized for Fermi cards and uses CUDA 3.2, while tokaspt (TOKAP's framework) was developed for CUDA 2.0 and pre-Fermi cards. The path tracing performance of tokaspt was already extremely fast, but this new path tracer is just in a league of its own.

- it's going to be fairly easy to port this to OpenCL, since the CUDA-specific code is very small and there's also an OpenCL port of the Kajiya demo available at http://code.google.com/p/kajiya-gpu/ Still waiting for working OpenCL 1.1 drivers from Nvidia.

- more primitives to choose from: not only spheres (as in Tokap), but also cubes (axis aligned bounding boxes actually) and ellipsoids, so there will be less restrictions on game ideas. An example of a physics game: drive the truck into a building made of boxes and make the whole structure collapse like in this video. This would require additional support for oriented bounding boxes.

- the Kajiya path tracer uses "hitboxes", a kind of acceleration structure, improving performance. Tokaspt doesn't have built-in acceleration structures, hence the path tracing performance went down very quickly when adding more spheres (or a second truck).

- proper support for skydome and sunlight with much better looking shadows and much reduced noise levels

Some pictures and videos (screenshots rendered with 8600M GT, videos with GTS450):

The next two images show all the advantages of rendering with path tracing: reflection, refraction, soft shadows, ambient occlusion and colour bleeding. Surpassing Pixar quality graphics in real-time! :D

Nice refraction + color bleeding and gradual light fall-off on the car's rear:


An "overcast" sky (no sunlight):



Night scene, only illuminated with a spotlight. Notice the indirect light bounced from the floor onto the cuboids:

Convergence in this simple scene is extremely fast. While moving, frames are displayed at 8 spp and the scene runs with great quality at 50-60 fps (512x512 resolution) on my GTS 450. When the camera is stationary, 32 spp are accumulated.

I hope I will soon be able to animate the truck and incorporate some CPU physics.

Download executable (Win32) and source code at http://code.google.com/p/tokap-the-once-known-as-pong/downloads/list

-

Minecraft rendered with Octane

After the impressive and awe-inspiring videos of Minecraft rendered in real-time with the Brigade path tracer (1, 2, 3 and 4), here's a pretty neat day/night cycle animation of a Minecraft level rendered with Octane, which looks completely real and kind of resembles a LEGO world: http://www.youtube.com/watch?v=vCC6Bczi2x0

With a couple of ultrahigh-end GPUs, I think the daylight parts in this animation can be path traced in real-time with very good quality, even though all surfaces are glossy, because almost all of the lighting comes from a skylight. The following video demonstrates the extreme speed with which a completely glossy benchmark scene (the Death Star trench from Star Wars) converges in Octane on 2x GTX 460 under different lighting conditions: http://www.youtube.com/watch?v=uHY90epFf2g. After having messed around with the Brigade path tracer for countless hours on a laptop with 8600M GT and still getting pretty good results in real-time, I'm convinced that Minecraft levels will soon be playable in photoreal quality on high-end hardware. Once that goal is reached, it's pretty easy to imagine what's going to come next.

Cycles, a new unbiased CPU/GPU renderer for Blender

The GPU path tracing virus keeps spreading at an incredible rate: Blender is soon going to have a fully integrated physically based renderer, codenamed 'Cycles', which can use the CPU or the GPU for rendering (GPU path tracing starts at 1:18).

http://code.blender.org/index.php/2011/04/modernizing-shading-and-rendering/

Changing materials, moving objects around and adding/deleting new objects can be done in real-time and the Cycles rendering engine can also rebuild the acceleration structure of the Suzanne model at lightning fast speeds. The source code will be available in about two weeks! It's going to be really cool when you can modify your model and immediately see the changes rendered in real-time in photoreal quality. Speaking of real-time editing and rendering, one of Luxrender's developers has made an utterly amazing video showing live editing of a physics simulation inside Blender and rendering it in real-time with Luxrender's PathGPU2 (a 100% OpenCL path tracer and part of SmallLuxGPU v1.8 with a heavy focus on animation rendering):

http://www.youtube.com/watch?v=bSQoJW9ajmU (real-time part starts at 1:23)

The video was rendered on a Core i7 920 + 2x ATI HD4890's, certainly not a bad system, but the HD4000 series was not exactly conceived with OpenCL computing in mind. I would love to see a GTX580 rendering this scene though. According to the graph below from Anandtech depicting the performance of SmallLuxGPU 1.7, I can imagine that card would just fly in this scene and make the video that much more awesome!

Update: SmallLuxGPU 1.8beta (which incorporates Metropolis light transport in OpenCL) is now available for download from the Luxrender forum

Update2: A lot of useful information and implementation details on the Cycles renderer can be found on http://wiki.blender.org/index.php/Dev:2.5/Source/Cycles under the header "Design"

Incredible real-time GPU path tracer using WebGL

Yesterday I came across another awesome GPU renderer (made by Evan Wallace), it's a very neat and extremely fast path tracer using GLSL shaders running in the browser:

http://madebyevan.com/webgl-path-tracing/

A WebGL-enabled browser is required to run this (I highly recommend the latest Chrome build from http://www.khronos.org/webg/wiki/Getting_a_WebGL_Implementation#Chrome.2FChromium (the Chrome Canary build works great if you're on Windows, Firefox 4.0 crashes and other Chrome versions didn't work for me, it just keeps loading).

It's great fun to edit the scene: move the light or objects and see soft shadows being cast in real-time (60 fps) on the walls and other objects. You can also change materials and add extra objects. It renders blazingly fast even on very low end GPUs. Very impressive!

Source code for the path tracer:

view-source:http://madebyevan.com/webgl-path-tracing/

There's also a GLSL version (Mac OS X):

http://www.madebyevan.com/glsl-path-tracing/

With this technology, a path traced physics simulation like the one in this video is perfectly possible in real-time on a current GPU. Or maybe a real-time path traced version of Tetris ;)

CentiLeo: a brand new interactive out-of-core CUDA path tracer for massive models

Wow, the real-time GPU path tracing goodness just keeps on coming! I've barely recovered from the fantastic real-time path traced Kajiya scene by Jacco Bikker and Jeroen van Schijndel, and now I stumbled upon another hugely impressive interactive GPU path tracer, called CentiLeo, which is able to render massive models (such as a stripped down Boeing 777 model containing 370 million polygons) interactively with global illumination on just 1 GTX 480 using out-of-core GPU path tracing (3-10 frames per second at 1024x768, 1 sample per pixel with 1 GI bounce from an HDR image).

A twelve-minute video in HD of this amazing renderer in action:

http://www.youtube.com/watch?v=mxx9dyPO0js

Some screengrabs of the video with more info:

The planned features sound extremely exciting and seem to be aimed at production rendering with among other things bidirectional path tracing, Metropolis light transport, hair and displacement mapping:

All these features will be CUDA based according to the author of the video (see youtube comments):

"the target is CUDA implementation because it is fast and flexible enough. BDPT, MLT, tesselation, displacement mapping, hair/fur, texture filtering - sure, will be CUDA-based.

Programmable materials can be done in CUDA. But an additional idea would be to port some existing CPU/C++ material shaders to work with our GPU compute building blocks. Voxels are not yet planned. But they can be potentially implemented.

Our goal is to finish the TODO feature list by September 2011. The concepts are pretty simple, but all the devil is in details. The future work is engineering. Almost all research was already done.

The website will be created this or next week. This demo shows what we can do now."

The holy grail for game graphics:

This tech would be an ideal fit for rendering huge voxel data sets stored in a sparse voxel octree at interactive rates with high-quality global illumination. Definitely something to keep an eye on in the future.

Scene from Kajiya's paper 'The rendering equation' can now be path traced in real-time on the GPU! UPDATE: exe available

This is just incredible, another milestone in the history of rendering! Jacco Bikker, the main developer behind the real-time path tracer Brigade, and Jeroen van Schijndel, an IGAD research assistant, have made a new simple but superb path tracer (similar to tokaspt) which can render the classic path tracing scene from the 1986 paper "The rendering equation" by Jim Kajiya in real-time on just one GTX 470! There is very little noise overall, there's just some in the glass spheres and in the shadowed caustics.

Youtube video: http://www.youtube.com/watch?v=OHWxsUUataw

Some stats:

- 512x512 rendering resolution

- 8 bounces

- 64 samples per pixel

- 12 frames/second on 1 GTX470

The amazing path tracing speed is partly due to the fact that there are no triangle meshes but only geometric primitives (spheres and boxes) in this scene, which are computationally much cheaper to intersect than triangles. The scene in the video is not 100 % identical to the original one (the structure consisting of the revolved parabola with the oblate spheroid, 'mushroom' for short ;-), is missing), but they're working on it (UPDATE 2: the mushroom is finished, see link at the end of this post). This is the original scene from the 1986 paper:

The above is an off-screen photograph (published in the paper). This is the direct feed image:

It's really mind-boggling when you realize that Kajiya needed 1221 minutes (20.35 hours) to render this image on a supercomputer from 1986 (an IBM 3081 mainframe) and 25 years later it can be computed at the same resolution in 36 milliseconds on a GTX580! A speed up of 2 million times!! Sounds like a great 25th anniversary :D !

(gotta love that '80ies font :-)

I would love to see some animation in this scene, for example an animated light casting moving shadows or a collapse of the pile of green spheres, which would greatly accentuate the "real-timeness" of the path tracing.

UPDATE: Executable and source code for this demo are now available at the links in this thread on the ompf forum. It's awesome, I'm getting a frametime of 1900ms in the 64 spp version on my poor 8600M GT, which is 49x slower than a GTX580 in this demo (kajiya-perf, default view at 64 spp/frame needs 1759 ms/frame on my 8600M GT and only 36 ms/frame on a GTX580)! Time to upgrade :)

UPDATE 2: Like Jacco Bikker has promised in the comments, the mushroom-like structure is now done! Visit http://ompf.org/forum/viewtopic.php?f=6&t=3174 for a screenshot of the updated scene and for more info on this project.