Is it possible to block out unneeded rendering ?

[masonspappy]

Attached image has taken over 23 hours to render and will probably require another 12 to complete.
What I'm noticing is that areas of water that will never be seen by the viewer are still being rendered, and this takes significant time to do. After the water is rendered, the rock outfacing is rendered over top the water.
Would it be possible to tell Terragen to not render water that will be hidden by rock outfacing from the viewer? 

[Matt]

In the Beta you can try the new "Defer all shading" option on the render node. It will completely avoid this overdraw, and will also improve the quality of the water. However, it doesn't always decrease render time, and sometimes increases it. I don't know how it will play out with this scene. Also, occasionally it can ruin renders by splashing primary colours over them during the anti-aliasing pass and/or the post process. Consider it a beta feature

When using Defer All Shading I recommend changing the soft shadow sampled on your sunlight to no more than 4 (to reduce render times).

Matt

[Dune]

There's a few things you can do:
1. set sorting bias in the lake object to -1000000 (makes the renderer do other things first)
2. make a mask where there's no water (check out my clipfile), taken from the height of the terrain.
3. make a mask where you don't see the water by rendering a small fast render, changing it in photoshop to a black/white mask where there is water/no water. Import this in an image map shader, projected by same render camera and use that as mask for the water shader (through the default shader, like in my clip file).

[Kadri]

In the Beta you can try the new "Defer all shading" option on the render node. It will completely avoid this overdraw, and will also improve the quality of the water. However, it doesn't always decrease render time, and sometimes increases it. I don't know how it will play out with this scene. Also, occasionally it can ruin renders by splashing primary colours over them during the anti-aliasing pass and/or the post process. Consider it a beta feature
...

Matt can this be tested with a basic test with say a 200x100 picture?
If this looks OK and rendered without the unnecessary background can we assume that a HD render will be the same 100% for example?
Or is there a Beta nature or uncertainty regarding small image versus big image regarding the above?

[Matt]

Matt can this be tested with a basic test with say a 200x100 picture?
If this looks OK and rendered without the unnecessary background can we assume that a HD render will be the same 100% for example?
Or is there a Beta nature or uncertainty regarding small image versus big image regarding the above?

I think it should be an OK predictor, but I'm not sure. There are so many variables. Use the same AA in the low res test and the HD render. AA affects quality and render time with deferred shading.

Matt

[Kadri]

Matt can this be tested with a basic test with say a 200x100 picture?
If this looks OK and rendered without the unnecessary background can we assume that a HD render will be the same 100% for example?
Or is there a Beta nature or uncertainty regarding small image versus big image regarding the above?

I think it should be an OK predictor, but I'm not sure. There are so many variables. Use the same AA in the low res test and the HD render. AA affects quality and render time with deferred shading.

Matt

Looks like worth to try. Thanks Matt.

[masonspappy]

Everyone, thanks for the feedback.  The final full-size render ( Detail 5, AA .5, mx samples , GI Cache detail/quality: 5/5) took 34 hours to render.  Using 34 hours as a baseline it will be interesting to see if I can implement any of these suggestions to speed up render time without loosing quality.  Thanks again for your comments!

[fleetwood]

..... ( Detail 5, AA .5, mx samples , GI Cache detail/quality: 5/5) took 34 hours to render. .....

I hope you mis-typed the above    Detail 0.5  AA 5  maybe ?

[Oshyan]

The "splotchy color bug" is semi-random unfortunately. So doing a low resolution test is no guarantee that the issue won't show up in high resolution. If my recollection is correct it's a bad interaction with GISD though, so if you don't use GISD you can (I think) be pretty sure the messed up color issue won't occur.

- Oshyan

[Matt]

If my recollection is correct it's a bad interaction with GISD though, so if you don't use GISD you can (I think) be pretty sure the messed up color issue won't occur.

That used to be true, but I think the new post effects (bloom, starburst etc.) can also trigger it.

Matt

[masonspappy]

..... ( Detail 5, AA .5, mx samples , GI Cache detail/quality: 5/5) took 34 hours to render. .....

I hope you mis-typed the above    Detail 0.5  AA 5  maybe ?

I like that even better. Thanks for catching !!