i think i remember once someone mentioning that grouping objects uses more memory than if your objects are ungrouped. i tend to group alot of things when placing then. should i ungroup everything in the scene before rendering? or is this memory thing a myth?
no thoughts on this? im just wondering if im shooting myself in the foot (or the ram) by grouping a bunch of object to possition them. then adding that group to another group to possition the whole set. then grouping that with other groups to have everything as a whole. etc.
just a thought da_elf, but why don´t you make a simple test with that. I myself make a lot of groups too, it would be interesting to know if it really consumes more memory.
regards
Mark
i’ve noticed that its faster to render a 5000 polygon mesh then a 100 * 50 polygons objects , the prepering objects is slower,
so if you can attach the models wherever you can…
definitely not a myth…grouping bad…selection sets good
I had a scene ages ago where there were heaps of nested groups (groups within groups) like 5-8 levels of them !!! the file went from like 170 meg to 30 or something after I cleaned it up, ever scince I’ve kept an eye on them.
Im using selection sets now for the most part, my files seem easier to navigate in and possibly a little more stable. I was a grouping junkie but I ahave been able to slowly train myself to use the selection sets and now prefer them to grouping.
Just a thought,
-dave
wow. ive gotta get out of grouping for sure then
interesting,
I think I must quit grouping too then. thanks for the tip
Nested groups are a problem.. grouping in general, can help.
Each level of a group adds another hierarchy to the set, which entails another set of coordinates, and positions/rotations/scales for each.
Its not much until you start layering them on top of each other with large numbers of objects..
I personally find it easier to set up more by Overall Area of model, and then by material, more fond of attaching than of grouping to reduce object clutter.
but to each his own.
one other problem I just think of, when not grouping, it is much more difficult to merge certain objects from one scene into another. Does anyone have a solution for that?
Attaching is mostly not an option for me, since I work with npower CAD objects.
For merging I’ve always saved a copy of my scene with the elements that I’m wanting to merge grouped (with a name like “to be merged”), then ungroup once they’re in the new scene. I don’t think grouping temporarily like that would leave any extraneous data behind for Max to consume.
The problem with grouping comes about when raytracing (like Vray and other GI renderers do).
When the ray being traced hits an object, to find out the information about that object it needs to open the group, the subgroup, (.. etc.. etc.. depending on how many groups deep you are) before it can get to the object.
It may not seem like much, but as you havem more objects in groups it becomes a lot of extra overhead to sort through the groups, and remember, it has to do this for every ray, so its being done thousands (if not millions) of times in any given frame rendering. It adds up, quick.
Is that true? the scene “Objects” array is not bothered by groups except that the groups are included afaik
I’ve always understood that you can do as you wish as far as workflow goes until render time. Then you should ungroup everything and collapse to editable mesh to conserve memory and make the scene as lean as possible. However this may be crap now because I remember being told this when first learning Max 2.5. Can anyone confirm or quash this?
I’ve never attached objects. Is this a good way to go. Currently I am using selection sets more and more as it is easy and doesn’t seem to become corrupted, at least so far, like I have found in the past with groups.
Of course the corrupting may just be me! :lol:
just stumbled accross this thread, I would agree about the grouping and rendering being slow. I try to have everything in selection sets and everything seems to work a lot better.
Has anyone done any proper tests to prove the theory?
I’ve also just stumbled across this thread. It seems to have fizzled out before any conclusions where drawn…
Did anyone look into this further, and if so would they mind sharing what they found?
Cheers,
Olly
Yes, it would be nice to get some solid answers from those in the know. I only know what I posted in my earlier thread. Further information would be nice! 
hmmm. seems like this has resurfaced wow. this is kinda old hehe. but its still important info. and here is exactly what i found in my research…opps. i believe i forgot to do research, ive been quite busy hehe
My theory is that the convenience far out weighs the render time hit.
Of course that is just a theory because I have no idea what if any the render time hit is.
Now, groups without group heads and groups without any children are just plain evil, that I can attest to. :twisted: