movie modeling question

when modeling a scene would you model the whole set or would you model your scene like hollywood builds movie sets? for example would your set be modeled open to one or more sides like a stage set? or would you just model the set and place your camera inside the room that is your set?

also what is the easiest way to generate greebles and applying them to models?

Arrange the set based on what will only be on camera on a shot by shot basis.

This could mean only a backdrop if the shot is a fairly static shot or with a limited amount of panning. In this situation, you would only shoot a still of the set. And in that still, you’d only put what will be in the shot - what is in front of the camera.

This way you can render the character only, the BG image will not take much time at all to render. Simply match the lighting on the character to the BG scene.

If the camera is to have more wider motion or tracking or dolly/boom shot type thing, then arrange the set accordingly.

In general, also think in layers.

Camera motion and background

Character Animation

Physics such as Cloth.

Ideally work from a storyboard - if only a simplistic sketch of the shot.

Then include only what will be in the shot.

Animate according to what will be in that shot. You could then shoot the background as an animation in one pass complete with camera movement without the character.

Then use this BG as an image sequence. You can then tweak the character animation and rendering as a composite over the BG. A much faster way to work.

In the case of the character you would ideally animate the character in an empty set with only place holders for any practical props or things such as stairs etc.

Once the character animation is done, then add the cloth objects and simulation of other things such as hair. Don’t try to do all those things at once.

Once the character is complete then render it over a BG animation or include it in the set if that seems the best for the shot.

And in general - cheat. It is what appears to be that is what is important. And it is only what the camera actually appears to see that is what you are after. Do this on a shot by shot basis. You can cheat objects in or out of frame. It is the composition of the shot that is most important - not what is actually in the “real” set. And this goes for any other method as well. Any compositing of BG or special effects or anything can be done in layers or cheated any way you can think of as long as it saves time and looks right - in the camera.

particle painting: http://vimeo.com/3841078

Yeah, go take the Universal Studios tour and you’ll see a “city street” where everything on that street has nothing but a front. “Those guys were on to something.”

Take Richard’s admonitions to heart.

The way I do it is like this:

  • Design the set, paying particular attention to scale. Design the props. Put everything into its own separate (“library”) file. From the very start you’ll do everything by linking with relative-paths. At this point don’t worry about materials, textures, etc.

  • If you’ve got a large and complex set, design it first and place all the parts, then separate it into major objects. Put the objects into appropriate groups. Place an “empty,” well-named, precisely at the center of each object. (Put all of the empties into another group.) Now you have “set pieces” (no pun intended…) that you can link to separately, and you have a master reference (the empties…) of precisely where each one fits relative to one another. If you’re shooting in only one area, just grab the groups you need. If you need several groups, link to the empties-group also so that you can always be sure of placement.

  • Create a separate file for each sequence of movement (what I call a “skit”), linking in the base-scene to everything else. Place plenty of cameras to cover the action well. Now spin-off scenes linked to the base scene. In each scene select one camera, specify the output file name settings. Then Ctrl+click the “preview” button at the bottom right side of the 3D window. Presto… instant animations, just low res.

  • Edit this. Build the movie now. From this you’ll know what footage you actually need to build, and what frame-numbers of each.

  • Break each shot down as Richard described. Design and build final renderings such that everything is broken out into separate tracks of data. (My staples: color, reflection, specular, shadow.)

  • You’ll build up the final shot by compositing. The idea is that if you just have to “tweak” something you can probably avoid re-rendering, and if you do have to re-render you can re-render exactly (and only) what you need, and thus keep on track.

Sure, it’s a bit of a PITA to keep track of all that material but it pays-off big in the end, because you find something and say, “I need to fix that,” and you find that you can ‘fix that’ in a very reasonable amount of time.

Try the Discombobulator script that comes with blender. Open a script window and it is under Scripts / Mesh
http://www.nccn.net/~w_rosky/evan/evan/programs/discombobulator/tutorial.html

RIchard

That’s a very cool organization scheme. I like it.

Yeah and Richard brought up a very very good point that I only touched on.

That is the rendering itself can be done in layers. This can be done with render buffers.

There is a great tutorial on this subject. It is for LightWave but it applies to Blender as well.

http://www.spinquad.com/article/story/8

great tip here Culver, Richard,sundialsvc4 and The Happy Friar. this will help me tremendously, and the information you have shared is by far the best reply i have ever received in this community. I do have a couple of projects i have bouncing between but i think i will put the other one on hold and flesh out my other one. once i do i will post what i have and again thanks to all of you.