Page 1 of 1
"Render Instance" of a "Render Instance"
Posted: Thu May 28, 2015 2:11 pm
by mentor972
I tried to see if there was another post about this, but I can't find it. This may technically be the first time this has been asked..
Let's say I have a carton full of boxes that have been textured in a store. When I create a set of instance objects for those boxes and check "render instances" Octane renders just fine.
But say I want to create an instance of that instance and clone the carton of boxes 20 times to fill some shelves. Once I check the "Render Instances" box the new instances disappear. If "render instances" is not checked everything is fine, but it would be nice to optimize my scene in this manor. Or maybe with Octane it doesn't matter. This may be a limitation, but I don't know. I haven't heard anyone else talk about it. Any thoughts?
"Yo dawg, I heard you like instances."

Re: "Render Instance" of a "Render Instance"
Posted: Thu May 28, 2015 2:15 pm
by aoktar
it's from first words of plugin specifications. It's a limitation. You can't embed a render instance to other render instance. You should re-arrange your object structure for this.
Re: "Render Instance" of a "Render Instance"
Posted: Tue Jun 02, 2015 6:40 pm
by mentor972
I did. I figured a couple things out since then. It appears from some tests that if you do a "render instance" of an object that any subsequent instances of that instance actually do render an instance anyway. Even though the checkbox isn't checked. In other words, only the first instance in the chain needs the box checked. The tri count will still behave as if everything under it in the chain is checked.
Re: "Render Instance" of a "Render Instance"
Posted: Wed Jun 03, 2015 2:11 pm
by ralf_breninek
Aoktar correct me if I am wrong:
Instanciated instances do render, but will not be as memory efficient as "real" instances. Therefore you have the benefit of Master - Child behaviour (changing the master object also changes the child) but without the positive VRAM and viewport effects. At least thats the way I would explain it in simple words

Re: "Render Instance" of a "Render Instance"
Posted: Wed Jun 03, 2015 2:35 pm
by aoktar
ralf_breninek wrote:Aoktar correct me if I am wrong:
Instanciated instances do render, but will not be as memory efficient as "real" instances. Therefore you have the benefit of Master - Child behaviour (changing the master object also changes the child) but without the positive VRAM and viewport effects. At least thats the way I would explain it in simple words

Hi! That's wrong! Render instances will consume very little footprint on VRAM and renderer. It behaves exactly same as C4D's render instances. Master-Child relationship is same also
Re: "Render Instance" of a "Render Instance"
Posted: Wed Jun 03, 2015 2:53 pm
by ralf_breninek
What I meant was that an instance of an instance with turned off "render instance" will behave that way.
A single instance with turned on "render instance" will for sure work as expected and save a lot on memory.
Re: "Render Instance" of a "Render Instance"
Posted: Fri Jun 05, 2015 4:24 pm
by mentor972
From the initial tests I can see, an instance of an instance behaves as if the "render instances" IS checked. The polygon count remans the same. A cube with 12 polys remains 12 polys when I instance a "render instance."