Enzi
Community Member

Posts: 67
|
 |
« Reply #100 on: September 11, 2008, 11:16:37 AM » |
|
Some new screens: ssao+ibl+ no tex 
ssao+ibl 


edit: Just for your information, no Texture except the HDR map was used in the city scene.
|
|
|
|
« Last Edit: October 17, 2008, 06:43:03 PM by Enzi »
|
Logged
|
|
|
|
AriusEso
Customers
Community Member
    
Posts: 940
Esoteric
|
 |
« Reply #101 on: September 11, 2008, 11:23:23 AM » |
|
Very cool man. 
|
|
|
|
|
Logged
|
-...-  
|
|
|
Mietze
Community Member

Posts: 415
Pleeease, don't let it crash!
|
 |
« Reply #102 on: September 11, 2008, 11:57:03 AM » |
|
Very cool! This is a good combination! Hope you enjoy some parts of my code 
|
|
|
|
|
Logged
|
|
|
|
Enzi
Community Member

Posts: 67
|
 |
« Reply #103 on: September 11, 2008, 12:03:31 PM » |
|
Actually I had the whole implementation before I found your project. I looked at it and it would have saved me a lot of time. But nevertheless, I learned a lot. edit: Oh, and I'm using a perspective projection. I haven't found the sense of using orthographic yet. 
|
|
|
|
« Last Edit: September 11, 2008, 12:07:19 PM by Enzi »
|
Logged
|
|
|
|
Mietze
Community Member

Posts: 415
Pleeease, don't let it crash!
|
 |
« Reply #104 on: September 11, 2008, 12:17:33 PM » |
|
edit: Oh, and I'm using a perspective projection. I haven't found the sense of using orthographic yet.  The good thing with using an orhtographic one is, that the different splits don't overlap as that much 
|
|
|
|
|
Logged
|
|
|
|
amerthebest
Community Member

Posts: 2
|
 |
« Reply #105 on: December 09, 2008, 05:17:28 AM » |
|
Could you share you source code with us please?
Thx
|
|
|
|
|
Logged
|
|
|
|
lovetv3d
Community Member

Posts: 139
|
 |
« Reply #106 on: December 11, 2008, 03:13:29 AM » |
|
|
|
|
|
|
Logged
|
Sorry for poor English
|
|
|
Enzi
Community Member

Posts: 67
|
 |
« Reply #107 on: December 11, 2008, 09:45:08 AM » |
|
What's up with that image. I know it's from Leadwerks and that his approach was more an outline shader than ssao.
|
|
|
|
|
Logged
|
|
|
|
Enzi
Community Member

Posts: 67
|
 |
« Reply #108 on: January 16, 2009, 10:37:32 PM » |
|
After some time I disliked my ssao and today I rewrote it completely. After 12 hours in Rendermonkey, here are the results. Bigger radius  Smaller radius, perfect for humanoids  There are still things to improve. Scaling the radius with the current depth a little better would be the first thing that comes to my mind. Oh, and the shader is super fast. 
|
|
|
|
|
Logged
|
|
|
|
|
Zaknafein
|
 |
« Reply #109 on: January 17, 2009, 01:01:44 PM » |
|
Looks awesome. Are you going to releaase that?
|
|
|
|
|
Logged
|
|
|
|
|
sybixsus
|
 |
« Reply #110 on: January 18, 2009, 06:25:10 PM » |
|
I agree with Zak, awesome. The new version has a little more noise, but the overall effect is much better. I second the question whether you're going to release it.
|
|
|
|
|
Logged
|
|
|
|
AriusEso
Customers
Community Member
    
Posts: 940
Esoteric
|
 |
« Reply #111 on: January 18, 2009, 06:55:44 PM » |
|
Yea, looks great.  I have been thinking about mine and maybe including some IL in there too.
|
|
|
|
|
Logged
|
-...-  
|
|
|
Enzi
Community Member

Posts: 67
|
 |
« Reply #112 on: January 18, 2009, 09:28:16 PM » |
|
Indirect lighting? In terms of color bleeding? There is a quite nice approach on gamedev and in the ogre forum from nullsquared. The results are pretty good but it uses a lightmap. There has to be another way to get this right. I was thinking of a map where the geometry is just output with no lighting at all. Just the albedo. To increase the emissive you just scale the needed object. I don't know if it's a dumb idea. I really need to sit down and try it out cause it's bothering me a lot. Maybe it's working for simple objects like spheres and cubes. And don't ask me why I don't want to use a lightmap. I really can't tell you why.  @Release: It's too soon to tell if I'm going to release it. I'm sorry. I can only tell you that the way the algorithm works is the same as the one that is floating around in the internet, the one from Arius, Unreal or Crysis. They all are doing the same. Taking samples via a random normal texture and compare it to the actual depth. @Noise: Unforunately it's something you can't get rid of. Unless you take much more samples. In my previous picture (the one with the dinosaur) I used 16 samples. Now I only take 8. I think it's a great jump in quality but I really have to see it in a real scene. If the noise is seeable I have to increase the sample rate. But I guess, because of the low frequency of the noise it's hard to see in a fully lit scene. At least, I hope so.
|
|
|
|
|
Logged
|
|
|
|
|
|
Aurora
Community Member

Posts: 36
|
 |
« Reply #114 on: January 19, 2009, 04:18:06 AM » |
|
@Release: It's too soon to tell if I'm going to release it. I'm sorry. I can only tell you that the way the algorithm works is the same as the one that is floating around in the internet, the one from Arius, Unreal or Crysis. They all are doing the same. Taking samples via a random normal texture and compare it to the actual depth.
You should release the shader when you are done, ofcourse you did use alot of time creating it perhaps, but not much reason for 20 people to make the same thing 20 times right  You can always sell it for like 1$ with excample on how to use it or something to make a bit of loose cash, then people have the options at least. Just a thought 
|
|
|
|
|
Logged
|
ASUS P5E AiLifestyle /w Intel Core 2 Quad Q9550 @ 3.6 Ghz 4GB Mushkin PC-8500, Asus Gforce Asus GTX295 1792MB, 1.95 TB HDD 28" ViewSonic VX2835wm 3ms on Windows Vista Home Premium x64. Ninnea Online - http://www.Ninnea-Online.com (In progress)
|
|
|
Enzi
Community Member

Posts: 67
|
 |
« Reply #115 on: January 23, 2009, 04:35:17 PM » |
|
Again, I rewrote the algorithm. It's working now with the delta of the actual ViewSpacePosition of the random point instead of the currently processed depth and some other tweaks of course. It looks cool with both. Here are some new screens: 

64k triangles, performing with 120fps on a 8800GTS 512
|
|
|
|
|
Logged
|
|
|
|
|
Raine
|
 |
« Reply #116 on: January 23, 2009, 04:55:23 PM » |
|
One of the best implementations I've seen lately, kind of like the one I've seen on gamedev.net (glsl code tho)
|
|
|
|
|
Logged
|
|
|
|
|
Zaknafein
|
 |
« Reply #117 on: January 23, 2009, 06:17:37 PM » |
|
Amazing. The white background makes a huge difference, it almost looks like the model scatters light.
|
|
|
|
|
Logged
|
|
|
|
AriusEso
Customers
Community Member
    
Posts: 940
Esoteric
|
 |
« Reply #118 on: January 23, 2009, 09:02:23 PM » |
|
Yea, that's exceptional. 
|
|
|
|
|
Logged
|
-...-  
|
|
|
uncasid
Community Member

Posts: 109
|
 |
« Reply #119 on: October 22, 2009, 04:39:23 PM » |
|
Any plans on releasing that shader?
|
|
|
|
|
Logged
|
|
|
|
|