PDA

View Full Version : when to use compositing? when just rendering in 3d


Raptor235
02-18-2002, 06:09 AM
Hi there I was just wondering if I could get some feed back on when I should settle for rendering everything in a 3d package or when I should break it all up and then put it back together in a comp package such as shake. I know you get a lot more control over each layer if you break it up but I dunno I just need some imput on when and how people break things up and just some tricks of the trade if anyone has any to share.

DragomanJK
02-20-2002, 12:36 AM
When you are doing an animation and the background stays the same it is easier to composite the layers sepretely then rendering the whole thing a bunch of times. If it is going to take a long time to the render the whole scene out, it is sometimes easier to cut it into layers. That way you have computer on for 3 hours for background, then you do stuff, then you render foreground for another 3 hours...

i-d
02-20-2002, 01:56 AM
Rarely you do everything in 3d package.
Even if you need a 3d space, compositing is your friend.
You don't really need to break things up :)
layering is a good aproach.

Raptor235
02-20-2002, 02:00 AM
what's a good way to layer stuff in maya I know there is render layers I'm gonna try that.

mestela
02-20-2002, 10:51 AM
Maya's render layers are great when they work, but I've found quite often they don't... :mad:

My current tack is when I'm ready for rendering, I create a new scene, load the finished scene as a reference, isolate what I want for a pass, save the scene as layer1.ma, isolate the next layer elements, save as layer2.ma etc.

The advantage is that changes to animation mean little to no changes for rerendering, and the file sizes for each layer tend to be really small, maybe 1 or 2 k.

The disadvantage is that you're limited to simple adjustments per layer; you can toggle visibility, some shader parameters, and lights, but thats about it. You can't reference particles, it's flakey with changing or parenting references, and referencing references (!) sometimes falls apart.

But it's still more reliable for me than render layers.

-matt

Relic9
02-21-2002, 07:09 AM
Well there are several advantages for compositing rather than rendering it out at once.

1)At first there's the independant lighting issue, it's easier to oversee a lighting setup for indepently lighted elements.
Alot times you can just light elements more beautifull when doing it seperatly from eachother.(not always, but look at FF spririts within, it gave a nice touch to the movie)
If you'd just stuff a huge scene with many lights, you might not just get that extra touch when lighting some things/characters seperatly.

2)Secondly, some scenes are so huge to handle at once that even the best SGI systems are having a hard time providing a reasonable framerate environment for artists to work in.
Offcourse you can hide alot of stuff when animating and animate with everything in a low polygone count, but even this has it's restrictions.

3)Rendertimes, i think that's self-explainetary, try rendering true raytraced shaders in combination with particle system with combustionlike shaders and under the influence of complex dynamics.
Even the studio in Honolulu splitted up scenes like these for FF spririts within, while having one of the largest renderfarms in the world.
It's just plain faster and therefore saves them money.

4)Esthetical reasons and package advantages.
For example, for Final Fantasy X the game, a studio responsible for the highpoly animatics used Afterburn(3ds max) from Digimation for dust effects and cloud/explosion effects.
While the main package they worked in was maya, that way they were able to use the famous maya dynamics with a 3dsmax plugin.
They rendered out the dust in max with afterburn and composited it into maya animations.
Combining the best features of several packages.
I also read that in FF the movie they used the maya standard renderer for some(not many) objects in the movie, just because it looked better than with other renderers.

5)Easy independant layer by layer color correction.
You know how bad many movies look and how amaturistic before being color corrected?
Imagine you made a huge scene(all in 1, no compositing layers used), just imagine how a pain in the ass it would be color correcting this, the people from post production would have to mask alot of objects in order to be able to color correct them 1 by one, continuesly tracking the objects in order to having them look the same throughout the movie.
Because some objects need to be darker, some blood needs to be more red, some stars need to be more brighter.(for example)
Now when you have layers, you just color correct the entire layer and it stays that way through the entire shot/scene and you save yourself alot of work.

That's about of what i could remember from the 1st test i ever had on the academy, i'm not sure if i mentioned all points.
But do feel free to add stuff :)

CGTalk Moderation
01-13-2006, 01:00 AM
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.