Resource difference between rendering passes

Become a member of the CGSociety

Connect, Share, and Learn with our Large Growing CG Art Community. It's Free!

THREAD CLOSED
 
Thread Tools Search this Thread Display Modes
Old 02 February 2013   #1
Resource difference between rendering passes

Forgive my ignorance if some terminology is off not to mention general understanding of composition. I'm curious to understand what kind of difference there is in rendering time between the separate "main" passes excluding special effects such as fur etc.

Say if I was to render Diffuse, Specular, Reflections, Shadows, Depth/Z, what kind of differences I would see in rendering times. I suppose it is quite situation specific. But some kind of rule of thumb or relative estimate would be appreciated if specific difference can not be determined.
 
Old 02 February 2013   #2
Originally Posted by Cromfel: Forgive my ignorance if some terminology is off not to mention general understanding of composition. I'm curious to understand what kind of difference there is in rendering time between the separate "main" passes excluding special effects such as fur etc.

Say if I was to render Diffuse, Specular, Reflections, Shadows, Depth/Z, what kind of differences I would see in rendering times. I suppose it is quite situation specific. But some kind of rule of thumb or relative estimate would be appreciated if specific difference can not be determined.


You usually render it at once, and then it's broken down into passes. Most will take the same as would take in the time of rendering: blurry reflections, blurry refractions, depth-of-field and motion blur. Also fur, depending on your renderer.
 
Old 02 February 2013   #3
Originally Posted by mister3d: You usually render it at once, and then it's broken down into passes. Most will take the same as would take in the time of rendering: blurry reflections, blurry refractions, depth-of-field and motion blur. Also fur, depending on your renderer.


Can you elaborate on the "render it at once, and then it's broken down into passes" as it sounds quite counter intuitive. Clearly I'm not understanding the pipeline at large since that feels opposite of what I thought.

If I had a still image what I would want to composite with multiple passes just to split down the effects for later fine tuning. This I assume to be the main benefit of rendering in passes, that the final image can be split into logical layers based on their visual information and individual layers can be corrected or altered for the purpose of not needing to re-render all other layers. Hence one needs to have logical categories such as diffuse, specular etc that are being used to make the final image. First you render 2 separate images and finally you combine them to the final image, not the opposite from what I understood from your snippet of text.

So if one needs to render 2 images, for the sake of example, one containing reflections and one containing diffuse. These are then rendered separately to be finally combined to create the final image as the work is done. This thing is sort of secondary for my question. Im interested in getting better understanding of situtation where 1 frame can be rendered in 1 time units with just diffuse, and 2 time units for reflections for example using fancy raytracing. This is my assumption that diffuse takes less time than reflections. Now, thats just an assumption and needs some flesh around it from someone who is actually doing multipass composition.

Now that all may sound just pointless and obvious. But the thing what I'm after is getting some kind of understanding of the different visual information that is being split to passes. That is there differences on rendering as in what type of visual information is currently fast to process and what is not by render engines.

Damn this isn't so easy to articulate :P
 
Old 02 February 2013   #4
There are basically two completely separate processes involved in rendering in "passes". One is something called "AOVs" or "outputs" or "layers" or what have you -- every software seems to have a different name for it. Most render engines are able to output things like reflection shading, Z-depth and material IDs in a single render pass. This does generally not have any impact on render times, but it is also quite limited in its scope. This is typically used to tweak the final look, but can also be used to make it easier to comp stuff into your scene. For example, you could render out diffuse shading and reflections separately to make it easier to comp in the shadows cast by a filmed actor.

Beyond that, there is something usually called "render passes", "scene states" or "takes" (again, different names in different software). This is something that allows you to change basically any property in the scene, and then render again. It can be used to toggle the visibility of different items, for example, so you can render your character in one pass, and then the hair in a different pass, and the shadows cast by the hair onto the character in a third. The capabilities and specifics of render pass systems vary from software to software. Typically, this means rendering the scene several times. But by breaking out different elements, each render pass can be quicker than it would be if you rendered everything in one go.
 
Old 02 February 2013   #5
Originally Posted by CaptainObvious: There are basically two completely separate processes involved in rendering in "passes". One is something called "AOVs" or "outputs" or "layers" or what have you -- every software seems to have a different name for it. Most render engines are able to output things like reflection shading, Z-depth and material IDs in a single render pass. This does generally not have any impact on render times, but it is also quite limited in its scope. This is typically used to tweak the final look, but can also be used to make it easier to comp stuff into your scene. For example, you could render out diffuse shading and reflections separately to make it easier to comp in the shadows cast by a filmed actor.

Beyond that, there is something usually called "render passes", "scene states" or "takes" (again, different names in different software). This is something that allows you to change basically any property in the scene, and then render again. It can be used to toggle the visibility of different items, for example, so you can render your character in one pass, and then the hair in a different pass, and the shadows cast by the hair onto the character in a third. The capabilities and specifics of render pass systems vary from software to software. Typically, this means rendering the scene several times. But by breaking out different elements, each render pass can be quicker than it would be if you rendered everything in one go.


Ok, now that starts to make much more sense. Thanks for the clarification.

"Most render engines are able to output things like reflection shading, Z-depth and material IDs in a single render pass."

So ultimately the end-user is not able to see if there indeed was a difference between these layers. It is just managed by the software on the fly and you get the final result as is. So that you dont go through manually by disabling this setting or that setting, rendering the pass and then proceeding to render say the inverted selection of settings. Maybe I try to experiment a bit as this doesnt seem to fall under the workflow what really happens.
 
Old 02 February 2013   #6
Quote: So ultimately the end-user is not able to see if there indeed was a difference between these layers. It is just managed by the software on the fly and you get the final result as is. So that you dont go through manually by disabling this setting or that setting, rendering the pass and then proceeding to render say the inverted selection of settings.
Actually you can set up single passes with stripped down material/light settings instead of rendering in one go. Not sure about any increase in speed.
 
Old 02 February 2013   #7
Originally Posted by Cromfel: So ultimately the end-user is not able to see if there indeed was a difference between these layers. It is just managed by the software on the fly and you get the final result as is. So that you dont go through manually by disabling this setting or that setting, rendering the pass and then proceeding to render say the inverted selection of settings. Maybe I try to experiment a bit as this doesnt seem to fall under the workflow what really happens.

I'm not quite sure what you mean by this.
 
Old 02 February 2013   #8
Originally Posted by CaptainObvious: I'm not quite sure what you mean by this.


I mean that if this is not really part of any workflow then there is diminishing possibility for someone to have paid attention to the rendering times between the different passes. Making my assumption for getting quick answer not so good. So I rather just set up the test scenario and start fiddling around with it by my self instead of hoping that someone has the answer for me.
 
Old 02 February 2013   #9
Thread automatically closed

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.
__________________
CGTalk Policy/Legalities
Note that as CGTalk Members, you agree to the terms and conditions of using this website.
 
Thread Closed share thread



Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

vB code is On
Smilies are On
[IMG] code is On
HTML code is Off
CGSociety
Society of Digital Artists
www.cgsociety.org

Powered by vBulletin
Copyright 2000 - 2006,
Jelsoft Enterprises Ltd.
Minimize Ads
Forum Jump
Miscellaneous

All times are GMT. The time now is 02:36 AM.


Powered by vBulletin
Copyright ©2000 - 2017, Jelsoft Enterprises Ltd.