• Create Account

## Shadow mapping and high-up objects

Old topic!

Guest, the last post of this topic is over 60 days old and at this point you may not reply in this topic. If you wish to continue this conversation start a new topic.

16 replies to this topic

### #1GuyWithBeard  Members

1650
Like
1Likes
Like

Posted 19 February 2013 - 06:40 AM

Hi,

I have cascaded shadow maps setup and working pretty well in my engine. I have large outdoor scenes and currently only one light (the sun) casting shadows. The problem is that when objects are high up, eg. right above the camera, it is outside the closest, highest detail cascade, and thus does not cast any shadows. My camera usually looks a bit down, since it's a third-person game I am making. How should I go about fixing this? Should I just make the cascades big enough to contain all objects in the scene? This, of course, would mean that the resolution of the cascades goes down.

The high-up objects are mostly static. Should I look into doing some sort of offline precalculated shadow data for these objects?

### #2GuyWithBeard  Members

1650
Like
0Likes
Like

Posted 19 February 2013 - 06:51 AM

The same problem also occurs if, say, the camera is looking horizontally forward, the sun is shining on the back of the character and there is a bird flying between the sun and the camera. In that case the bird is outside all of the cascades.

### #3L. Spiro  Members

24826
Like
0Likes
Like

Posted 19 February 2013 - 07:00 AM

The cascades are extruded infinitely towards the sun so anything between what is directly in front of you and the sun should cast a shadow.

So the question is why isn’t the bird getting inside the bounding volume cast by the sun?  Your sun-cast bounding volume is probably capped too low (are you capping it just above what the is in the player’s view?).  It shouldn’t be capped in the direction of the sun at all (no vertical limit) specifically because of this situation.

L. Spiro

### #4GuyWithBeard  Members

1650
Like
0Likes
Like

Posted 19 February 2013 - 07:04 AM

Yes, I am making the cascades as tight as possible (so they cover just what's in the frustum). I have a tweakable value for scaling the cascade near and far clips, but the shadow quality decreases noticeably if I scale them up, presumable because of limited depth buffer resolution, am I right?

EDIT: It's quite possible that I haven't done the CSM correctly, but this is what I am seeing with my current setup. If having infinite cascades is the norm, how do you deal with depth buffer resolution issues?

Edited by GuyWithBeard, 19 February 2013 - 07:06 AM.

### #5L. Spiro  Members

24826
Like
1Likes
Like

Posted 19 February 2013 - 07:23 AM

Yes, I am making the cascades as tight as possible (so they cover just what's in the frustum). I have a tweakable value for scaling the cascade near and far clips, but the shadow quality decreases noticeably if I scale them up, presumable because of limited depth buffer resolution, am I right?

EDIT: It's quite possible that I haven't done the CSM correctly, but this is what I am seeing with my current setup. If having infinite cascades is the norm, how do you deal with depth buffer resolution issues?

You are confusing “far” with “towards the sun”.
I don’t mean you should adjust how far away from the viewer each cascade is, I mean the bounding volume for each cascade should not had a limit in the direction of the sun.

Assume the sun is directly above you. The first cascade is 5 meters out, the second is 15, etc.
The distance out is not a problem. It is the distance above you that is the problem. There should be no limit vertically towards the sun.

L. Spiro

### #6GuyWithBeard  Members

1650
Like
0Likes
Like

Posted 19 February 2013 - 07:30 AM

The "near plane" and "far plane" in my comment above refer to the near and far plane distances of the orthographic projection of the cascade cameras. They have nothing to do with the player camera or where the cascades or shadow casters are.

I can indeed make the cascade "infinite" by moving the camera back by infinity/2 and moving the far plane forward by infinity/2. It's just that earlier when doing this my shadow quality went way down, and I thought this had something to do with depth buffer resolution, but maybe I was wrong.

### #7L. Spiro  Members

24826
Like
3Likes
Like

Posted 19 February 2013 - 07:53 AM

The process is that you start with no near plane from the sun’s point of view (but do have a far plane), gather all the objects in the volume created by that box (its frustum), and the decide on the sun’s near and far planes to tightly fit the objects that were gathered into that frustum, creating your final orthographic projection based on the nearest and farthest object you gathered.

You don’t have to worry much about distances being crazy and causing floating-point errors unless the scale of your world is huge, in which case you would need a general strategy for this anyway.  Typically a simple offset when checking the distance in the depth texture will do.

L. Spiro

### #8osmanb  Members

2081
Like
7Likes
Like

Posted 19 February 2013 - 08:06 AM

POPULAR

This is a real problem, though. One (good) solution usually goes by the name "pancaking". Basically, you still set your near and far planes (for the light's ortho projection) to tightly bound the camera frustum. But you make sure not to cull anything behind the near plane (depending on how you do your frustum culling, this is usually pretty easy). Then, in your shadow-mapping vertex shader, after you do your projection, you clamp the Z value:

projPos = mul(matrix_mvp, In.Pos);
projPos.z = max(projPos.z, 0.0f);


This has the effect of flattening everything that's off-screen (from the camera's perspective) onto the near plane of the light's projection. (Because you only really care about the silhouette of those objects). There is the potential for minor artifacts when you have large triangles (they're warped if one vertex is behind the near the plane and another one isn't), but if you just give yourself a small buffer zone (make sure to pull the near plane back a short distance), it should work fine. We're using this technique now, and haven't had any problems with it.

Edited by osmanb, 19 February 2013 - 08:07 AM.

### #9GuyWithBeard  Members

1650
Like
0Likes
Like

Posted 19 February 2013 - 08:08 AM

Hmm, very interesting. I need to look into the pancaking technique. Thank you both.

### #10CC Ricers  Members

1491
Like
0Likes
Like

Posted 19 February 2013 - 10:24 AM

Should I just make the cascades big enough to contain all objects in the scene? This, of course, would mean that the resolution of the cascades goes down.

This would be an intuitive assumption but it's not completely correct. I, too, at first thought my shadows will appear more pixelated if I increase the Z extents of the light frustum, but this only affects the depth resolution. As you do not change the X and Y dimensions of the orthographic projection, the texel-to-pixel ratio of the shadows should stay the same. You simply will be making the "tube" of the frustum longer from one end.

If you are using a big enough bit precision in your texture format (at least 24 or 32 bpp), arbitrarily extending the depth usually won't pose any problems.

Edited by CC Ricers, 19 February 2013 - 10:24 AM.

New game in progress: Project SeedWorld

My development blog: Electronic Meteor

### #11GuyWithBeard  Members

1650
Like
0Likes
Like

Posted 19 February 2013 - 12:43 PM

Wow this is really awesome. The pancaking solves all of my problems in one go. And it took 5 minutes to implement. Thank you very much for the tip!

### #12MJP  Moderators

18194
Like
6Likes
Like

Posted 19 February 2013 - 03:03 PM

POPULAR

If you're using DX10/DX11 you can implement "pancaking" very easily by setting "DepthClipEnable" to false in the rasterizer state (which also avoids rasterization issues caused by mucking with the projected vertex position).

Edited by MJP, 19 February 2013 - 03:04 PM.

### #13L. Spiro  Members

24826
Like
0Likes
Like

Posted 19 February 2013 - 04:37 PM

At work, we use the term “pancaking” to describe what is going to happen to one of our poor coworkers, who lives in a rigidity old shack, the next time there is the slightest breeze through Tokyo.

I have never heard of this technique and it is a good idea.

L. Spiro

Edited by L. Spiro, 19 February 2013 - 06:25 PM.

### #14osmanb  Members

2081
Like
0Likes
Like

Posted 19 February 2013 - 06:23 PM

I have never heard of this technique and it is a good idea.

Yes. I first heard about it from an older (2009?) Advances in Real Time Rendering presentation, given by Bungie. It was only mentioned briefly, and it was described as though other people had implemented the technique. Very few people have ever posted about it, though, so I don't think it's actually very common.

### #15GuyWithBeard  Members

1650
Like
0Likes
Like

Posted 19 February 2013 - 11:21 PM

Thanks MJP, this just gets better and better.

### #16Tournicoti  Prime Members

704
Like
0Likes
Like

Posted 23 February 2013 - 07:56 AM

Personally, I cap the light's viewing frustum with the upper bound of the world's aabb so that it's impossible to miss any shadow caster.

(I follow -1.0f*sunDirection until the wanted plane is above the upper face of the world's aabb)

Edited by Tournicoti, 23 February 2013 - 08:31 AM.

### #17Tasche  Members

222
Like
0Likes
Like

Posted 23 February 2013 - 05:40 PM

oh sweet i actually had this problem too, but didn't want to post because i could actually live with it...

but seeing this is really only 5 min to implement you have another happy customer =)

Old topic!

Guest, the last post of this topic is over 60 days old and at this point you may not reply in this topic. If you wish to continue this conversation start a new topic.