Synthetic images from the Raydiosity engine

(L) [2009/11/18] [albertoven] [Synthetic images from the Raydiosity engine] Wayback!

The knowledge and renders of the people at this site are awesome, I'm enjoying and learning so much here so let's share. These are some renders from my Raydiosity engine. It's a path tracer focused on procedural geometry and texture generation. Even the old castle image is completely generated with Raydiosity procedural API. So I put myself to work very carefully, stayed designing on paper and CASE tools the Raydiosity engine about 2 months before starting writing a single line of C++. Raydiosity engine is linked with no library except standard C++ basic libraries so everything on it is built from scratch. The engine is prepared to be executed on any number of computers at the same time and on each computer any number of instances of the engine could be launched simultaneously. Everything is controled with a front-end. Most of the renders were synthesised with a K7 (1 Raydiant instance) and a P4 (2 Raydiant instances). It took 6 months to get the engine working. I prefer equations and algorithms to hand-made mouse design so a great amount of effort has been put on the geometry and texture procedural API. Almost without exception all Raydiosity renders use exclusively procedural entities. Now I'm working on the sequel of the Raydiosity engine, it's called Raydiant engine and is a realtime no precalc true radiosity machine. Raydiant engine is already several times faster than Raydiosity and offers better quality. I'm working right now on a realtime interactive demo with Raydiant. I think Phantom is so right in forecasting realtime raytracing games will overcome raster ones at some point in time no so far away. There's information about this 2 engines, high resolution versions of these images and even some Z80 works from an early and enjoyable time at my blog [LINK http://albertoven.com/]. Wanted to translate to reality some images I had in my mind to share them with friends. All of them are made with the Raydiosity engine, here they are:
[IMG #1 Image] [IMG #2 Image] [IMG #3 Image] [IMG #4 Image] [IMG #5 Image] [IMG #6 Image] [IMG #7 Image] [IMG #8 Image] [IMG #9 Image] [IMG #10 Image] [IMG #11 Image] [IMG #12 Image]

[IMG #1]:

[IMG #2]:

[IMG #3]:Not scraped: https://web.archive.org/web/20101001220355im_/http://albertordmr.files.wordpress.com/2010/04/sweet_gypsum_flavour_512x512.png
[IMG #4]:

[IMG #5]:

[IMG #6]:

[IMG #7]:

[IMG #8]:

[IMG #9]:

[IMG #10]:

[IMG #11]:

[IMG #12]:

(L) [2009/11/19] [PabloDeHeras] [Synthetic images from the Raydiosity engine] Wayback!

Delicious visuals and very cleanly implemented tech from what I've read. I'm a C++ purist myself and see your program as another testament of what can be done without the use of dedicated hardware.
I have a question though. The images look like they have been generated with almost random geometry and fractals. How do you define Lipschitz for such surfaces (I assume you use a sphere tracer), or do you manually set a maximal gradient by trial and error?
Again, I want to eat these images! [SMILEY :-)]
Pablo

(L) [2009/11/21] [albertoven] [Synthetic images from the Raydiosity engine] Wayback!

>> thachisu wrote:Excellent work! These are the most artistic usage of procedural geometry/texture I have ever seen
albertoven wrote:Now I'm working on the sequel of the Raydiosity engine, it's called Raydiant engine and is a realtime no precalc true radiosity machine.
so... is this using radiosity or what? How did you render caustics with radiosity?
Thanks for the support. This is a path tracer so it naturally renders caustics and other effects without specific programing for them.

(L) [2009/11/21] [albertoven] [Synthetic images from the Raydiosity engine] Wayback!

>> homie wrote:any perf numbers on raydiant?
The Raydiant engine is still at a development stage. Perhaps I will soon open a thread to speak more about it. So the data available now is both sketchy and provisional. Perhaps the most significant data available now is this: using true real time no precalc pathtracing you can move around a scene with several thousand geometries (cylinders, spheres, triangle meshes, boxes and other things) at a resolution of 80x60 on an i7 (Raydiant uses as many threads as possible, in this case 8 are used). You get an average of about 6 frames per second. Each frame image is very noisy. This is somewhat slightly reduced because the rapid succession of frames, each one with its own noise pattern, tends to cancel a little bit the noise. The scene includes crystal, mirror, diffuse and translucent materials.

(L) [2009/11/21] [albertoven] [Synthetic images from the Raydiosity engine] Wayback!

>> toxie wrote:i want more info on how these wonderful pics were created (tech, algorithms, whatever)!
They are all 100% procedural geometry and mostly also procedural textures. So C++ code generates every one of them. The Raydiosit procedural API failitates this. Some concrete examples: the back wall of the first one (called 'hemisfear') is a extruded procedural texture I've called Alberto's Maze (at [LINK http://albertoven.com/] there are some videos of this texture in its 2D form). The crystal jar (from the 'before it flows' picture) is made with the revolution of a single crystal tube generated with a trigonometric function. The purple 3D fractal ('blueberry hiper snowflake') is a natural 3D extension to 2D snow flakes. The white ball with something red inside ('sweet gypsum flavor') is a red crystal ball covered by a maze generated parametrically (distance between beans, grow preferences, ...). The green maze ('the king of the potato people wont let me') is a combinacion of a procedural maze and texture with some hand located geometries. Beacuse Raidiosity engine has a bit matrix geometry it very efficiently handles boxes arrays (like a maze).

(L) [2009/11/21] [albertoven] [Synthetic images from the Raydiosity engine] Wayback!

>> PabloDeHeras wrote:Delicious visuals and very cleanly implemented tech from what I've read. I'm a C++ purist myself and see your program as another testament of what can be done without the use of dedicated hardware.
I have a question though. The images look like they have been generated with almost random geometry and fractals. How do you define Lipschitz for such surfaces (I assume you use a sphere tracer), or do you manually set a maximal gradient by trial and error?
Again, I want to eat these images!
Pablo
Thanks Pablo, I appreciate your interest specially since your work at Visualbiotech really looks impressive. The Raydiant engine is a path tracer, it is specifically designed to grow, currently it can represent triangles meshes, spheres, cylinders, bit matrices and boxes. But any geometry can be added as needed. Any new geometry must implement a set of common methods. Montecarlo method is used to sample the renders. Do you use Lipschitz at all geometries?, what geometries traces the Visualbiotech engine?

(L) [2009/11/21] [PabloDeHeras] [Synthetic images from the Raydiosity engine] Wayback!

>> albertoven wrote:PabloDeHeras wrote:Delicious visuals and very cleanly implemented tech from what I've read. I'm a C++ purist myself and see your program as another testament of what can be done without the use of dedicated hardware.
I have a question though. The images look like they have been generated with almost random geometry and fractals. How do you define Lipschitz for such surfaces (I assume you use a sphere tracer), or do you manually set a maximal gradient by trial and error?
Again, I want to eat these images!
Pablo
Thanks Pablo, I appreciate your interest specially since your work at Visualbiotech really looks impressive. The Raydiant engine is a path tracer, it is specifically designed to grow, currently it can represent triangles meshes, spheres, cylinders, bit matrices and boxes. But any geometry can be added as needed. Any new geometry must implement a set of common methods. Montecarlo method is used to sample the renders. Do you use Lipschitz at all geometries?, what geometries traces the Visualbiotech engine?
Thanks Alberto,
I am pretty impressed by the simplicity of your approach, from your images the complexity that you achieve is quite stunning.
Our engine BioInspire is under NDAs so unfortunately I can't tell you what it does.
Keep up the excellent work,
Pablo

(L) [2009/11/22] [albertoven] [Synthetic images from the Raydiosity engine] Wayback!

>> PabloDeHeras wrote:
Thanks Alberto,
I am pretty impressed by the simplicity of your approach, from your images the complexity that you achieve is quite stunning.
Our engine BioInspire is under NDAs so unfortunately I can't tell you what it does.
Keep up the excellent work,
Pablo
Thanks!, glad you get it. What a shame you can not speak about that engine because the NDA, perhaps you have some personal project showing your view?, I'm always interested in learning some more!

(L) [2009/11/22] [homie] [Synthetic images from the Raydiosity engine] Wayback!

>> albertoven wrote:homie wrote:any perf numbers on raydiant?
The Raydiant engine is still at a development stage. Perhaps I will soon open a thread to speak more about it. So the data available now is both sketchy and provisional. Perhaps the most significant data available now is this: using true real time no precalc pathtracing you can move around a scene with several thousand geometries (cylinders, spheres, triangle meshes, boxes and other things) at a resolution of 80x60 on an i7 (Raydiant uses as many threads as possible, in this case 8 are used). You get an average of about 6 frames per second. Each frame image is very noisy. This is somewhat slightly reduced because the rapid succession of frames, each one with its own noise pattern, tends to cancel a little bit the noise. The scene includes crystal, mirror, diffuse and translucent materials.

is that 800x600 or 80x60?
thanks

(L) [2009/11/22] [castano] [Synthetic images from the Raydiosity engine] Wayback!

>> albertoven wrote:A question: will you want to play a game where the resolution is extremely low (like 80x60), the render is very noisy, the images have real time no precalc global illumination (so there can be any number of moving lights), getting around 6 fps and with crystal, mirror, translucent and other exotic materials?
I would say that 80x60 is way too small to be playable, but on the other side the sampling frequency of the global illumination does not have to be as high as the frequency of the primary rays and direct shadows. 6 fps is still too slow, though. I'd consider 30 fps to be the bare minimum for a fps-style game.

(L) [2009/11/23] [albertoven] [Synthetic images from the Raydiosity engine] Wayback!

>> castano wrote:albertoven wrote:I would say that 80x60 is way too small to be playable, but on the other side the sampling frequency of the global illumination does not have to be as high as the frequency of the primary rays and direct shadows. 6 fps is still too slow, though. I'd consider 30 fps to be the bare minimum for a fps-style game.
There are extremely high quality games out there already. They are really awesome. Anything less than true global illumination wouldn't make a real difference. So, the moment lights are classified as primary and secondary some of the power of global illumination is diluted. What about 10 primary lights? and 1000?. What if the room is dark and some kind of phosphorescent fungus is growing over everything (this would be a procedural texture)?. Raydiosity and Raydiant engines don't make that distinction. I say: lets optimize the general case, don't concentrate on the cases where there are few primary lights with specific shapes/volumes. The powerful hardware needed for this is closing on us... lets prepare!
Of course a fps is probably one of the worst game types to start using this technology. I totally agree 30fps is the minimum for such a game. What about a graphic adventure or puzzle game?. Would you like to explore a life like scenario?. You could get decent resolution snap shoots in seconds. Also the i7 with 8 kernels (=16 threads) is just around the corner. Even better: the Fermi platform would make it playable at real time and decent resolution.
Any way size matters. I agree with you 80x60 is too small. But this is a concept demo. Want a try?

(L) [2009/11/24] [7th Guest] [Synthetic images from the Raydiosity engine] Wayback!

>> albertoven wrote: Of course a fps is probably one of the worst game types to start using this technology. I totally agree 30fps is the minimum for such a game. What about a graphic adventure or puzzle game?. Would you like to explore a life like scenario?. You could get decent resolution snap shoots in seconds. Also the i7 with 8 kernels (=16 threads) is just around the corner. Even better: the Fermi platform would make it playable at real time and decent resolution.
Any way size matters. I agree with you 80x60 is too small. But this is a concept demo. Want a try?
Pathtraced 7th Guest @320X240 please!

(L) [2009/11/27] [Guest] [Synthetic images from the Raydiosity engine] Wayback!

>> albertoven wrote:7th Guest wrote:albertoven wrote:
Pathtraced 7th Guest @320X240 please!
ok, that's just a perfect choice for the test. A game that doesn't necesarily need high frame rate but will enormously benefit from life like appearance. Only one problem thou. It really scared me the firts time I played it. I'm not going to sleep well if it looks even more real (=more scary)
Albertoven, if you want to quickly test your pathtracer and you have not much time to build a game level, you could use the data set from the Arauna based Outbound game, available for download at [LINK http://igad.nhtv.nl/~bikker/downloads.htm]
But a 7th Guest remake would be awesome as well [SMILEY :D]

(L) [2009/12/05] [albertoven] [Synthetic images from the Raydiosity engine] Wayback!

>> beason wrote:These are fantastic! Keep up the good work. Also consider posting to deviantart.com if you haven't already. If you have, what's your user id? Some forum users have accounts there. The ones I know about are lyc and greenhybrid, plus my own (beason).
to beason and green hybrid: You are so kind, after so much work your appreciation is highly welcome!. Thanks for the advice about deviantart. You are so right. I'm now in the process of migrating my favourites images from Raydiosity to Raydiant engine so extremely high resolution instances can be obtained and posted there. I'll send you a notice at the very moment I have an account. By the way... any hints about your favourite ones?, you can browse high res versions off some of them at [LINK http://albertoven.com/].

Synthetic images from the Raydiosity engine back

Board: Board index ‹ Ray tracing ‹ Visuals