Hi guys, im sure this is a pretty noob question, but im trying to figure it out,
what type of comps require compositing with 2d passed, and which ones are better to approach in 3d space. Im sure that depends on a shot by shot basis but which criteria u guys use to pick either approach?
CG output contains many channels of information in addition to R, G, and B. Two of the most important channels are: [ul]
[li] Alpha: indicates transparency. (1.0 = solid, 0.5 = ghost, 0.0 = invisible.) [*] Depth: indicates the distance of the object from the camera. [/ul]
[/li]What you call “2D” compositing programs, therefore, can do “3D-aware” work because of the presence of these (and other) channels, all of which are normally embedded in a single input file.
Compositors, these days, are almost-universally node based, which means that you can actually construct a sort of “assembly line” by selecting various kinds of filters and plugging them together in different ways. The data “flows” through all of the various “filters and pipes” to finally produce (one or more…) finished outputs.
Incidentally: renderers, textures, materials, and so-on are these days constructed using exactly the same principles.
Naturally, since we’re talking about digital computers, this notion of “merging” can be extrapolated in many different ways. It’s really a matter of using one mathematical function (or output thereof) to somehow influence another, thereby eventually affecting the final-image in some way.
So, you’ve actually got an amazing amount of creative control available to you. This is a very, very, very deep pool. (Ommmm-m-m-m-m… :wise: )
You’ve probably heard that “movies are shot by the director, but they’re made by the editor.” Whether or not you agree with that notion, I think that the same general idea definitely applies to 3D work. The renderer is used to generate components, but “post production” steps are used to fine-tune and assemble them into a completed image. By and large, those steps are working with “bit-mapped images,” accompanied by depth information, assembling layer upon layer of information. But, these processes are quick.
Thanks sundialsvc4 vbmenu_register(“postmenu_6054880”, true); for replying, what u said is great but here’s my situation, i’m currently migrating from combustion to nuke, and i’m already in love with node based compositing, but in nuke i can bring .obj and recreate my 3d scene, so i was used to exporting rpf’s and exr’s with their usual channels, and i’m trying to figure out when one or the other way of comping is needed.
thnks
I think compositing in 2d gives you more control of the overall look of the final shot, like CC, Fog, Snow etc etc… having to design the plates more accurately before compositing like your camera moves and the placement of the elements,
and compositing in 3d with cards and objects you have more freedom in general.
So
in 2d there’s more planning involved before the compositing
and in 3d there’s more planning involved in the compositing??
Am i hitting on anything here?
i’m liking the 3d more than 2d so should i try to compose all my shots in 3d?
Using a combination of both is ideal, depending on the situation. For simpler comps and for shots with little to no camera move, a 2d solution is fine. A 3D solution will come into effect when you have a moving camera, a more elaborate set design, or just need the data that a 3d solution provides (depth, 3d motionblur, etc). The other thing to remember is that you can’t really light and texture appropriately within Nuke at this time (maybe with the newest release, haven’t tried it yet), so you will still have to rely on rpfs and exrs for your 3d rendered files. All the information by sundialsvc4, while informative, is probably not the direct answer you were looking for?
The other thing to remember is that you can’t really light and texture appropriately within Nuke at this time (maybe with the newest release, haven’t tried it yet),
I am not sure I totally agree with that. There are some pretty elegent relighting and retexturing solution out there at the moment. Sure they do not allow for compex transeprancy and odd shaders, but people are doing IBL, SSS and all the usual shader stuff live in Nuke.
What about shadows, shadow casting, GI and AO? Particle systems? I should have mentioned that in my previous post. Sure Nuke has a lot of nice scanline tools, but you’ll still need a dedicated 3d package to do some of the more fancy items. There’s still the speed hit as well. You’ll need to render your things out before you start comping them, just to keep your speed up. It does help the compositor in production, no doubt about it, but there are some items which must be relegated to the lighters and fx artists. Again, I haven’t spent too much time with the latest iteration of Nuke (5.2) so there are some things which may have been added since the Benjamin Button days.
[QUOTE=Aruna]What about shadows, shadow casting, GI and AO? Particle systems? QUOTE]
This is my main issue, and a was confused on how to aproach it, i guess is not much confussed but inexperienced in 3d comping, I’m trying to wrap my head on it.
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.