Parallax Occlusion Mapping

Oogst · Post by **Oogst** » Sun Jul 06, 2008 9:22 am

Potentially, it could be: cone step mapping allows for big steps through the height map without missing any details, so there should be no artefacts with cone step mapping. You do have to post a link to be sure, though.

nullsquared · Post by **nullsquared** » Sun Jul 06, 2008 12:36 pm

http://www.gamedev.net/community/forums ... _id=372739

(also, there appears to be some patent issues due to the discovery of this method in the past or something like that, I believe this can't be used commercially or something, can anyone confirm?)

PolyVox · Post by **PolyVox** » Sun Jul 06, 2008 2:33 pm

I haven't had time to properly read that thread, but have a look at this one too:

http://www.gamedev.net/community/forums ... _id=403699

It's by the same guy and is a similar but different approach to work around the patent issues.

nullsquared · Post by **nullsquared** » Sun Jul 06, 2008 2:34 pm

PolyVox wrote:I haven't had time to properly read that thread, but have a look at this one too:

http://www.gamedev.net/community/forums ... _id=403699

It's by the same guy and is a similar but different approach to work around the patent issues.

I saw that, too, but patents or not, RSM is nowhere near the quality & speed of CSM, especially for very irregular and deep surfaces.

vilgeits · Post by **vilgeits** » Sun Jul 06, 2008 7:54 pm

I'm a bit lost whats better, whats faster? CSM or POM?

nullsquared · Post by **nullsquared** » Sun Jul 06, 2008 8:36 pm

vilgeits wrote:I'm a bit lost whats better, whats faster? CSM or POM?

CSM is both faster and of higher quality than POM (check out the demo at the link I posted). However, it requires a separate pre-processed "step map", unlike POM which relies on solely a height map.

PolyVox · Post by **PolyVox** » Sun Jul 06, 2008 9:20 pm

The GPU Gems 3 article actually talks about relaxed CSM. The standard CSM stops the ray before it hits the surface, then you use bilinear search to find the intersection point. Relaxed CSM lets the ray intersect the surface once, and then backtracks to find the intersection point. This means the cones corresponding to each pixel can be bigger, and hence bigger steps can be taken with each iteration. This leads to faster convergence.

Implementing these kind of approaches is quite high on my list of things to do due to the low detail in my project. I'm interested in whatever people come up with!

Oogst · Post by **Oogst** » Mon Jul 07, 2008 9:13 am

Cone step mapping (CSM) is an optimisation of parallax occlusion mapping (POM). But what is that RSM term that is being used? Is that just Relaxed Cone Step Mapping (that would be RCSM, right?), or something else?

xadhoom · Post by **xadhoom** » Mon Jul 07, 2008 11:53 am

Kojack wrote:
The length of a cross is 1 if both vectors have the length 1.
No it isn't.
The length of a cross product is affected by the angle between the 2 vectors. The result is only length 1 if the 2 vectors are unit length and at 90 degrees to each other.

If the normal and tangent are always going to be at right angles, then it's fine. But that's a specific case.

Yes you are right. It has to be ensured, that the angle between normal and tangent is 90 deg. But this really should be the case in the shader, because we take both as
rotation matrix axis. If not all three vectors are orthogonal to each other, this matrix will lead to wrong results regardless of which length the vectors are.

vilgeits · Post by **vilgeits** » Mon Jul 07, 2008 12:21 pm

Is there any usable/easy to convert example of CSM or RCSM?

PolyVox · Post by **PolyVox** » Mon Jul 07, 2008 5:04 pm

Oogst wrote:But what is that RSM term that is being used?

It's Radial Step Mapping, which it appears was devised to work round the patent issues in CSM. I think anyway, I'm probably giving the impression that I know a lot more about this than I do!

Oogst · Post by **Oogst** » Mon Jul 07, 2008 8:45 pm

PolyVox wrote:
Oogst wrote:But what is that RSM term that is being used?
It's Radial Step Mapping, which it appears was devised to work round the patent issues in CSM. I think anyway, I'm probably giving the impression that I know a lot more about this than I do!

It is a bit sucky that the link does not actually explain how it works.

Faxnico · Post by **Faxnico** » Fri Sep 12, 2008 4:26 pm

Hi!

I'm experiencing some troubles setting up this POM shader with Caelum.

When the sunlight reaches its maximum intesity, the following texture...

http://img360.imageshack.us/img360/1664 ... s01mm5.jpg

... appears "a bit" too white as you can see from picture below.

While it looks fine when it's dark.

Is there a way to work easily around this?

Thank you and sorry for posting many screens.
NicolÃ²

Oogst · Post by **Oogst** » Fri Sep 12, 2008 9:15 pm

Looks like very bright specular to me. POM should not result in overbright lighting: it just delivers certain normals and colours. After that, lighting is the standard thing. So what could go wrong is that the normals might not be normalised somehow (seems unlogical, because they probably should come from a normal map), or that the lighting is just too bright.

Caphalor · Post by **Caphalor** » Sat Sep 13, 2008 9:45 am

Caelum's sun is really bright by Default. Try something like:

Code: Select all

	mCaelumSystem->getSun ()->setDiffuseMultiplier (Ogre::ColourValue (1.5, 1.5, 1.3));
	mCaelumSystem->getSun ()->setSpecularMultiplier (Ogre::ColourValue (1, 1, 1));

But even in the dark the texture doesn't look optimal in my opinion. Are you sure that you use a correct normal map with a height map in its alpha channel?

Faxnico · Post by **Faxnico** » Sat Sep 13, 2008 3:25 pm

First of all, thanks for your replies.

That was the reason indeed...

Caphalor, I'm almost new to game programming and I've absolutly no confidence with those techniques. Could you suggest me, for this first time, some realistic NVidia Photoshop Plugin's or CrazyBump's settings for that texture + Parallax Occlusion Mapping, please?

Thank you all again.
NicolÃ²

Oogst · Post by **Oogst** » Sat Sep 13, 2008 10:00 pm

I do not know if the Nvidia plugin or Crazy Bump can save in exactly the format you want, but it is easy with the Nividia plugin nevertheless.

If you use that plugin, then I suppose you already have a depth map. Make a copy of the depth map and then apply the Nvidia normal map plugin to the original. Now add a channel (in the channel window, click the little arrow in the top right and then choose 'new channel'). Then copy/paste the depth map that you just copied to the new channel. If you now save this file as a 32 bit TGA, you have a texture with a normal map in the RGB and a height map in the A.

tera4d · Post by **tera4d** » Tue Sep 23, 2008 11:11 am

Wow that looks really good.
This may sound a bit stupid hehe

but how do i use these materials and scripts?
I have never used the material system before because im fairly new to ogre3d.
I have placed the files in the right directory's where they belong but when i load them into the project and compile it then the model is just plain white :S but the lights do move on it like a glow on white paper.
Can someone please help me?

Caphalor · Post by **Caphalor** » Tue Sep 23, 2008 1:20 pm

Do you use Ogre 1.6 (Shoggoth)? The provided materials only work with Ogre 1.6.

tera4d · Post by **tera4d** » Tue Sep 23, 2008 4:10 pm

Yes i do use Ogre 1.6 but i got it to work after i updated my driver hehe

It looks very good and runs smoothly too.

lukeneorm · Post by **lukeneorm** » Thu Jun 04, 2009 3:46 pm

I can't download the rar from the link in the first post.
Is it broken?

rafa.gdev · Post by **rafa.gdev** » Thu Jun 04, 2009 4:26 pm

Yes, the link is broken

But this thread is from 2008 hehe

lukeneorm · Post by **lukeneorm** » Sun Jun 07, 2009 10:50 pm

rafa.gdev wrote:Yes, the link is broken
But this thread is from 2008 hehe

Yes, but it is still an interesting techniques

Anyone can help me to download the files?

Post by **jacmoe** » Sun Jun 07, 2009 10:59 pm

Caphalor was here yesterday, maybe you're lucky and he still have the code.

yashiz · Post by **yashiz** » Tue Jun 09, 2009 8:05 am

request the code too...

Ogre Forums

Parallax Occlusion Mapping

Re: Parallax Occlusion Mapping

Re: Parallax Occlusion Mapping

Re: Parallax Occlusion Mapping

Re: Parallax Occlusion Mapping

Re: Parallax Occlusion Mapping