advanced DX11 rendering engine

update: I created a (really REALLY crappy but quick) web page that has a few images of previous works, algorithms, and tests regarding some of the more advanced features listed above! I also have a demo that I will upload soon that shows real-time ambient occlusion in a dynamic scene (world space technique). It is unbelievably fast!

edit: I forgot the link lol (edit again for new site)

http://texcoord3.com/


-Mike-

So what are you exactly wishing to accomplish with the engine? Is it for developing your own games, a licensing solution similar to UE3 or CE3...any specific technical premise for the engine? Like how Frostbite's main aspect is the destructible environments, UE3 is about accessability, and CE3 is about realistic characters and environments.

Also if you are making it a native DX11 engine alot of big studios using it are now incorporating HW Tessellation and MLAA...not that it's required but not having those two features at some point will make the engine you are forming look quite dated, but I guess it all depends what you actually want to accomplish with the engine.

Just my two cents.

Quote:

Original post by BrioCyrain So what are you exactly wishing to accomplish with the engine? Is it for developing your own games, a licensing solution similar to UE3 or CE3...any specific technical premise for the engine? Like how Frostbite's main aspect is the destructible environments, UE3 is about accessability, and CE3 is about realistic characters and environments. Also if you are making it a native DX11 engine alot of big studios using it are now incorporating HW Tessellation and MLAA...not that it's required but not having those two features at some point will make the engine you are forming look quite dated, but I guess it all depends what you actually want to accomplish with the engine. Just my two cents.

So I have been commissioned by Dexsoft Multimedia to create an engine with these features (the rights remain mine/ours).

The specific thing going for this engine is superior support of hardware tessellation and displacement mapping and implementation of my real-time radiosity lighting. The end result will pull the engine's lighting effects way ahead of any game engine currently in existence (which is why I've been commissioned to do such an engine).

The engine will be the first rendering engine to feature my patent pending algorithm for computing global illumination for fully dynamic scenes of unlimited size using only traditional graphics hardware. Unlike other techniques, mine is capable of rendering high quality physically correct radiosity lighting of many bounces, specular reflections between all geometry dynamically, world space ambient occlusion and other sought after next-generation graphics solutions. I have composed research, papers, screen-shots, and technical demos to backup/prove my algorithms plausibility and authenticate these claims. Please contact me directly for further information.

-Mike-

Sent you an email...

To anyone interested in the progress of this:

I have created a site here:
http://texcoord3.wordpress.com/

thanks for your interest - let me know f you require anything from me =]

hey texcoord3, im a designer not a programmer but just wanted to say your work kicks ass. nice job man.

the API is about 80% done for the first revision.

Second revision (public release) *should* be available Q2/Q3 2012.

updates:

Hardware/OS support:
-I am also now adding dx11.1 and windows 8 support.
-The engine can also take advantage of AMD's GCN features such as improved compute performance and super textures as well as Nvidia's Kepler improved compute performance and hardware PhysX acceleration.
- It will also now support Crossfire/SLI.
-engine now uses scripting to manipulate objects and worlds (possibly ruby or python, atm it's a simple propriatary langauge)

Lighting:
-delaying voxel-rendering tecniques and replaced with analytical techniques or voxel-injection techniques.
-New type of culling - "Proximity Culling" will allow for 8-12 times more lights to be rendered iin a scene without significant performance penalty
-Hardware PCF for LOD lights and low settings.
-possible addition of static lighting techniques for lower-end systems/higher quality lighting in scenes that simply don't need 100% dynamic lighting
-real-time reflections of all surfaces can now follow the lighting equation correctly and be blurred

new lighting equation!!! takes color texture map, bump+height map, micro-surface roughness map, surface light-absorbtion ratio map

Physics:
-Nvidia PhysX with GPU physics acceleration

Tools:
-Mesh library for analytical evaluation and modification of meshes
-Online Scene/World editor (alpha)
-'Dragon' Debugger for run-time evaluation of engine and performance

i must say, you have pretty high goals.

unlimited lights?
and your epic algoritm cabable of doing real-time GI?

i want to see all this in a video befor i even belive you

just this two points are somthing everybody in the industry is struggeling with, and you, a solo man have apparently found a solution?
I just want to see this. im so excited!

and just to add, if you have only those two points, you could probably sell your light code for so much cash so you can pay every one intressted.

yeah nothing is better than an explicative video showing off your features in action

I actually have some videos and demos as well, email me if you want either (I suggest the demo is more fun).
The demo is only dx9 from over a year ago though. It was made as a demonstration of the technique for Wolfgang Engel's Shader Pro series.

And technically there aren't unlimited lights, just you could have a few million at once (NOT shadow mapping obviously).

I have many practical techniques for GI now actually. The reason why I have no deployed competition (none that I consider at least) is because everyone else is concerned with consoles and mobile devices. I am not. Thus progress is made towards high end graphics for the PC. To be fair LPVs are similar to SGI although without question inferior (and came after ironically), so some kudos to Crytek (and SSAO was a fantastic inovation). The Enlighten stuff hardly works...at least as implemented as I have seen in game (in theory it should be pretty powerful, but I guess they had to tune it down). Not hating on BF3 nor Crysis2 btw lol. At the moment, static global illumination still seems to take the cake imo (at least until I have time to deploy my code), especially the algorithms presented by Valve.

I actually have some videos and demos as well, email me if you want either (I suggest the demo is more fun).
The demo is only dx9 from over a year ago though. It was made as a demonstration of the technique for Wolfgang Engel's Shader Pro series.

And technically there aren't unlimited lights, just you could have a few million at once (NOT shadow mapping obviously).

I have many practical techniques for GI now actually. The reason why I have no deployed competition (none that I consider at least) is because everyone else is concerned with consoles and mobile devices. I am not. Thus progress is made towards high end graphics for the PC. To be fair LPVs are similar to SGI although without question inferior (and came after ironically), so some kudos to Crytek (and SSAO was a fantastic inovation). The Enlighten stuff hardly works...at least as implemented as I have seen in game (in theory it should be pretty powerful, but I guess they had to tune it down). Not hating on BF3 nor Crysis2 btw lol. At the moment, static global illumination still seems to take the cake imo (at least until I have time to deploy my code), especially the algorithms presented by Valve.

"The reason why I have no deployed competition (none that I consider at least) is because everyone else is concerned with consoles and mobile devices."
-No, i dont belive they are concerned with consoles, the genreation of consoles we have now are outdated and the new ones are coming pretty soon.
Beside of that fact that just having the tech cabable of doing Real-Time GI on a pc Envierment is a enormus succes! I have seen some videos on some real-time GI in some engines, and they have 2-3 bounces.

"

The Enlighten stuff hardly work"

[color=#282828][font=helvetica, arial, verdana, tahoma, sans-serif]-Please, Do explain more.[/font]

[color=#282828][font=helvetica, arial, verdana, tahoma, sans-serif]"[/font][color=#282828][font=helvetica, arial, verdana, tahoma, sans-serif]static global illumination still seems to take the cake imo (at least until I have time to deploy my code)"[/font]

[color=#282828][font=helvetica, arial, verdana, tahoma, sans-serif]-Untill you have something to actualy prove this with (like a video).[/font]

[color=#282828][font=helvetica, arial, verdana, tahoma, sans-serif]I´m not trying to start a fight, im just arguing and questioning you![/font]

[color=#282828][font=helvetica, arial, verdana, tahoma, sans-serif]it´s my nature [/font]

EDIT, this text editor SUCKS! it formats like shit.

the fact remains that virtually none of those techniques will work in a practical sense. Anyone can produce global illumination effects in fixed or small scene. The problem is implementing a solution that works for diverse and potentially unknown situations. I'm concerned with illuminating hundreds of kilometers of geometry, not the Cornell Box.

Enlighten hardly works... Do I really need to explain? Just watch a video of BF3 and you tell me if I'm wrong. Most of the time, especially outdoors, I can't even notice any GI effects besides SSAO...even if I look at screenshots. The qualiy is even worse on consoles where, less a few cases, I could swear it isn't even activated at all.

...I don't even understand your third comment... Are you asking for a demo? I already said that you can email me and I will send you some =]

Also I'm a little confused as to why you are questioning the work. I've already published papers on it and I've worked with various Industry leading programmers with great recomendation... I thought my name had a bit more weight to it but I suppose not. Nevertheless feel free to read my papers or check out my work online. You can find everything either on my site or through google.

Also about the upoming 4 consoles;
None of them have enough power to significantly alter graphics as we know it. Yes, tesselation and other 11.1 features are there, but they all lack the raw processing power needed to really get things done. To give you an idea my current engine runs on mimimum settings on a gtx560ti... there is no way in hell I can cram all 88 base rendering stages onto current gen hardware... and next gen hardware would already be a stretch for anything beyond minimum or low settings. I do a lot of research and the only viable techniques published online are of screen-space heritage, or related to SGI/LPV/Voxel type algorithms, crude ray-tracing/casting/rasterization techniques, or various static techniques. Of these the ray-tracing/casting/rasterization techniques are out for **large** worlds due to current and next-gen GPU architecture (although I'm tweaking WSAO). So you are only left with baiscally static, SGI/LPV, or some version of SSAO.... which is exactly what I stated above.

Finally please note where I am stating fact and where I am giving my opinion. Syntax makes a world of difference.

I actually have some videos and demos as well, email me if you want either (I suggest the demo is more fun).

I would like to see your demos of global illumination, ambient occlusion and unlimited lights.
I would post my review/impression here, if you don't mind.

you could send me up to 25MB to profl at 9y.com

the fact remains that virtually none of those techniques will work in a practical sense. Anyone can produce global illumination effects in fixed or small scene. The problem is implementing a solution that works for diverse and potentially unknown situations. I'm concerned with illuminating hundreds of kilometers of geometry, not the Cornell Box.

Enlighten hardly works... Do I really need to explain? Just watch a video of BF3 and you tell me if I'm wrong. Most of the time, especially outdoors, I can't even notice any GI effects besides SSAO...even if I look at screenshots. The qualiy is even worse on consoles where, less a few cases, I could swear it isn't even activated at all.

...I don't even understand your third comment... Are you asking for a demo? I already said that you can email me and I will send you some =]

Also I'm a little confused as to why you are questioning the work. I've already published papers on it and I've worked with various Industry leading programmers with great recomendation... I thought my name had a bit more weight to it but I suppose not. Nevertheless feel free to read my papers or check out my work online. You can find everything either on my site or through google.

Also about the upoming 4 consoles;
None of them have enough power to significantly alter graphics as we know it. Yes, tesselation and other 11.1 features are there, but they all lack the raw processing power needed to really get things done. To give you an idea my current engine runs on mimimum settings on a gtx560ti... there is no way in hell I can cram all 88 base rendering stages onto current gen hardware... and next gen hardware would already be a stretch for anything beyond minimum or low settings. I do a lot of research and the only viable techniques published online are of screen-space heritage, or related to SGI/LPV/Voxel type algorithms, crude ray-tracing/casting/rasterization techniques, or various static techniques. Of these the ray-tracing/casting/rasterization techniques are out for **large** worlds due to current and next-gen GPU architecture (although I'm tweaking WSAO). So you are only left with baiscally static, SGI/LPV, or some version of SSAO.... which is exactly what I stated above.

Finally please note where I am stating fact and where I am giving my opinion. Syntax makes a world of difference.

Where can i find theese papers?
Can you link stuff here on this webpage instead of just saying there is "stuff"?

"

there is no way in hell I can cram all 88 base rendering stages onto current gen hardware. "

[color=#282828][font=helvetica, arial, verdana, tahoma, sans-serif]

So you dont have what you tell us? a real-time render engine?[/font]

If you want the demo please send me an email, I don't know how to send it to you otherwise.

"
there is no way in hell I can cram all 88 base rendering stages onto current gen hardware. "

[color=#282828][font=helvetica, arial, verdana, tahoma, sans-serif]

So you dont have what you tell us? a real-time render engine?"[/font]

[color=#282828][font=helvetica,arial,verdana,tahoma,sans-serif]

Sorry it was very late and I meant on current gen *console* hardware. It runs fine on cards from 2009 and later. Note that there are compatibility issues with it running on ATI/AMD cards for some unkown reason.[/font]


http://www.coniserver.net/wiki/index.php/Screen_Space_Ambient_Occlusion_Shader

check my website for some Papers and Demos - I am adding new sections and uploading them from google documents
http://texcoord3.com/papers/

the reason why I cannot simply distribute the demo on my site is because I was stupid and did not encrypt the art. So by contract I cannot publicly release the demo as it is (and I don't have time to code encryption for an obsolete demo). I'm uploading an old video of it though. Also if you email me I can send the demo since I can track who has it and thus you will be held responsible if it is leaked to the public.

Check my youtube channel in like 10 minutes. http://www.youtube.com/user/texcoord3/videos

the reason why I cannot simply distribute the demo on my site is because I was stupid and did not encrypt the art. So by contract I cannot publicly release the demo as it is (and I don't have time to code encryption for an obsolete demo). I'm uploading an old video of it though. Also if you email me I can send the demo since I can track who has it and thus you will be held responsible if it is leaked to the public.

Check my youtube channel in like 10 minutes. http://www.youtube.c...excoord3/videos

Dude, this is as public as it can be!

the video? I don't care about the video just the zip files that run the actual demos.

If you redistribute the demo files then I simply won't share them in the future.

From seeing your demo, you dont have half of what you speak about?

There is just a big bloom filter applied to your entire screen, making it impossible to even see any GI, or SSAO?
Besides that it looks like you dont clamp your postprocessing since you get ghost artifacts like crazy.

May i post a picture to show you what i mean?