shade maps.. am i missing something?

[super gnu]

the workflow for shademaps, if i understand correctly, involves rendering the whole image first, to generate the shade map, then rendering the stereoscopic view using the shade map to accelerate rendering of the second view.

but since we are essentially rendering 3 images with 1 accelerated, as opposed to 2 un accelerated images, it doesnt seem to save any time whatsoever? its much quicker just to render the stereoscopic view first time without the shade map..?

maybe i understood wrong.

maybe since you can re-use the shade map in certain circumstances it might be beneficial, but since i have no idea what is stored in the shade map, i dont know what can/cant be changed before a recalculation of the shade map is necessary..

[vlado]

The shade map helps to accelerate both the left and right views. So you are rendering one full image, and two accelerated. For many cases, this is faster than rendering two full views. Some stereo systems require you to render not two, but up to nine images, in which case the savings are much greater.

As to what can be changed... not much; only DOF values, I think.

Best regards,
Vlado

[super gnu]

would it be possible to change the shademap system so it calculates it on the left view and uses it on the right? that way you dont need to pre-render a whole image? you do one unaccelerated and one accelerated render.. or can i do this already?

[vlado]

It won't quite work since the shade map must be computed with settings which may not be quite suitable for production rendering - e.g. DOF/moblur/AA filtering must be disabled.

Best regards,
Vlado

[iancamarillo]

Hi,
Im rendering several hd animations with no GI or dof. should I make the shade map in this situation or will there be no performance gain using shade maps with my set up? Thanks!

[vlado]

Is there motion blur? Is it a stereo animation?

Best regards,
Vlado

[iancamarillo]

no motion blur but i was thinking of making velocity passes. yes, full 1080p stereo render.

[vlado]

In that case, it depends. If you use the shademap-based renders as one of the eye renders, then it may be faster if the other eye goes through the shade map. You'll have to do a few tests though - it might not be worth the extra hassle.

Best regards,
Vlado

[iancamarillo]

Ok. The YouTube video looks cool (the lamp) but it doesn’t go through each step. For instance, if I don’t have to use the vray frame buffer, I won’t. But are there any benefits/conveniences to the vray frame buffer? My workflow is this:

1. Create animated scene with standard cameras.
2. Add stereoscopic helper to scene and adjust eye distance (I use RT to check these values - which is super convenient!)
3. Make output size 3840X1080 and render

My scenes aren't super fancy right now (no gi/dof/moblur) so I just want to get some fast high quality feedback to check the overall feel of the 3d effect. Is this workflow ideal for that?