Hi guys,

I'm encountering a problem in my last render. I've enabled Z Depth in my auxiliary render data but the depth map came out like this.

I'm using BB's Environment Sphere and there are several objects in different distances.

It took hours to render so I gave up and I had to do the blurring by hand but I'd like to know why did this happen and how to avoid it.

Is there any way to get just the depth map without re rendering the whole thing using poser's native settings?

first, I guess the depth map can be generated from any render settings, even with SSS and raytracing (and hence IDL) OFF, and various render quality settings at a low value. But you do need some render process to make it.

In order to make sense out of the map, the issue is whether the regular 8 bit grayscale 0..255 is good enough. You might need to save in 16-bit format or EXR, and boost contrasts in Photoshop (or alike). So perhaps the info is there but you can't see it on screen, and it gets lost when saving to PNG or so.

Or we might just need more info to sort this one out, but then we need some more shots on the screen. The rendered result, a top-shot including cam position, things like that. 

have fun 

I followed your suggestion and rerendered in very low settings and got the same exact result. The strange part is that the window part is transparent on the psd, but when I saved it as a jpg it got a gradation. There is no information there except for the envsphere that is set emit light but not show in the rendered image. When I boosted the brightness the black section just started moving to the left and finally covered the whole image, no extra information there.

My render settings

Top View

Final render

It's the first time this happens, I usually have no trouble getting the depth map

okay, first: JPG does not support transparency while PSD does, so I'm not supprised by that effect. PNG will do transparency as well.

I'm not in the position to run tests and things, but my first impression is that something with a smooth Z-depth presence is standing in the way. And the only thing I can think of, is that environment sphere. Which is set to behave invisible to the rendering, but I'm not sure if it does so to to the z-depth determination. So, effectively, your camera is out of the sphere looking into the scene.

A solution might be either to move the camera inside, or (my favorite) to enlarge the environment sphere. For a test: just make it visible, is it blocking the render then? If so, we've got our culprit.

by the way, nice scene !!

Thank you aRtBee :)

I checked the sphere and it's very large and the camera is inside of it. I deleted the sphere and still get the same result.

When I deleted the whole room I finally got the right map. Is there a bypass for this? It's no use to me if I can't have a map of the further back room.

so the wall of the room closest to the camera is the culprit. You can take that section out (and in order to take some light from the dome out, build a new larger wall behind the camera).

The other option is to bring the camera into the room. Then you might need Camera Perspective to set the required focal length.  Perspective alters the Camera Scale too and enables you to make wide angle shots at a large distance or the other way around.

Just thinking out loud, perhaps someone else has a better idea. 

The wall is part of the room and can't be deleted, the fact that the wall is invisible doesn't coun? Next time I'll probably just put just three plains about where the room walls are.

Thank you for the help aRtBee, especially for the tip that I can render the image in low res just for the depth map. You saved me a lot of wasted render hours :)

okay, apparently Z-depth and invisible don't combine that well. But still you can do a render with wall, and a Z-depth without at a faster render pace. That's something. Usually you need the high image resolution either, for Z-depth. That's enother saver.

And the Camera Perspective can still be of use.

have fun ! 

Thank you aRtBee :)

how does it look if you turn off HSVexpTM, and incr. bounces and value of envsphere map?

The built in ZDepth from Poser doesn't take into account transparency. At all. All it considers is the geometry. This will be especially noticeable if you do a closeup on someone's head with transmapped hair: what looks like thin strands or wisps in the render will be ribbons of polygons or even a solid shell.

how does it look if you turn off HSVexpTM, and incr. bounces and value of envsphere map?

I turned off the tone mapping and still get the same result but I don't know how to increase the bounces of the environment sphere. Is it a certain value in the material room? Keith, I tried to render some hair to see if you're right and I got the strangest result

The back hair are the same polygons as in the front. It's really strange. In my original picture too, the window is the farthest back object and, even though it's also behind the wall, it has transparency.

That second image isn't just the the Z-depth map, is it? Because if it was the background should be black.

That second image isn't just the the Z-depth map, is it? Because if it was the background should be black.

It's just the depth map. I didn't put any background, just the girl and the hair. It turned white when I saved it as a jpg. It's weird isn't it?

     Try making the room invisible.  Get Z-depth render of the contents. In photo app, select and delete the white background.  Get a Z-depth render of just the room prop.  Overlay first render extract atop the second (room) render.

Just throwing out ideas;  don't know if it will help.

     I know it can be done;  I use it occasionally to create displacement maps for bas-relief carved panels.

.

     Try making the room invisible.  Get Z-depth render of the contents. In photo app, select and delete the white background.  Get a Z-depth render of just the room prop.  Overlay first render extract atop the second (room) render.

Just throwing out ideas;  don't know if it will help.

The problem with that is that the room's invisible wall gives the gradation so I can't get any zdepth with it in the render. I'guess it's easier to just place simple planes where the wall would be. It's a pain but at least it gives me a complete map :)

     Ah, I get you.  Well, if you have a modeler, export the room, select/delete the near side wall, and re-import. 

To be honest, right now I'm more interested in knowing why this happens. The zdepth map with the hair I posted has me questioning if the maps are accurate in general.

I use them for faking the DOF and I don't want it blurring in blocks rather that the transmapped objects.