The goal of this tutorial is to fill out the dark parts of Maya docs about the relationship between the camera and the Image Plane and how to do the render to match (both Maya and PRMAN) the background image at compositing stage just as we saw it in the viewport while we were working in Maya.
This isn't rocket science, is just that the documentation limits to describe the attributes, and the default configuration of a new Camera isn't proper.
This is a common problem when we load an image sequence in the Image Plane of a Camera in order to track a 3D model.
As you have already noticed, the Camera and the Image Plane have several attributes of Film Back and Placement that affect the way we see the images inside the Resolution Gate, so the question is: Which of all these damm configurations is the the one that will do my render to match in the compositing just as we saw it in the viewport when we did the 3D tracking? Why the client complains about the final compositing because it doesn't look as the playblast I show to him? Aha!!! So this already has happened to you?
Even worse, if you change this attributes and you didn't pick the right combination, the render you do with Maya won't match the PRMan's render (a very frustrating situation).
In order to begin let's assume something, your project is for television and you will work at D1 NTSC resolution (To have a specific example, once you understand it you can adequate it to whatever you want), i.e. 720x486
The device Aspect Ratio is 4:3 (horizontal to vertical proportion), but the NTSC specification uses frames of 720x486 pixels to convert them in analogic signal. If you divide 720 by 486 equals 1.481481481481 etc. etc. that isn't equal to 4 divided by 3 (1.3333333333), I mean, the frames are enlarged 1.11111111 in the vertical axis at the moment they are converted to analogic signal, that's why you have to set the Pixel Aspect Ratio to 0.9 at render time. Your rendered images will look shrinked in the vertical axis to compensate.
In other words, it would be as if you do your renders at 720x540 (it does have 4:3 proportion) and then scale them to 720x486.
This could be too basic and obvious to some, but it is necesary for better understanding.
First create a Camera and its own Image Plane in the attribute editor.
Before you begin tracking you may want to configure the focal distance (of course, if you have available that info). There is a parameter that is intimate related to focal distance that affects angle of vision; In Maya the cameras were conceived as cinematographic real world cameras, no as video cameras for TV, so we can change the Aperture. In a real camera the Aperture is the physical size in width and height of the film plane, and the focal distance is the distance in milimeters from the lens plane to its focus (the point where it concentrates the light), if the film area is small the angle of vision will be narrow.
As the size of the camera aperture increases, a longer focal length is required to achieve "normal" perspective. For example, a 35mm camera uses a 50mm lens as a normal lens. On a 16mm camera, the same 50mm lens appears telephoto in nature. A 25mm lens is required to achieve "normal" perspective on a 16mm camera.
And you may ask: Why should I care about this if my project is for TV?, well, it is convenient to make the Resolution Gate match with the Resolution Gate, in other words: What you see in the Film Gate is exactly what you're going to get in your render. And as you maybe have already noticed when you create a new Camera the Aperture is 1.417 horizontal and 0.945 vertical, this is a 1.5 Film Aspect Ratio. I really don't know what is the equivalent camera aperture for a video camera, but for shure it will depend upon the physical size of the CCD, what I'm sure is that the Film Aspect Ratio must be 1.333333, and if you son't know what combination pick you can use the Film Gate presets and chose 35mm TV projection.
Warning: If your exported your camera from Boujou the Film Aspect Ratio will already at 1.333333 with a software calculated aperture.
What's next is to set the Film Fit to Vertical (the default is Fill)
I don't have a scientific explanation for this, as I understood the docs Fill should be OK but I don't know neither in which way MTOR uses this parameters for the camera RIB generation, so believe me, after exahustive tests this works.
The camera is ready, what's left if to configure the way we see the reference image in the Resolution Gate. When we work in Maya we do it with 1:1 proportions, we don't see vertical shrinked geometry in the viewport I mean. But if the sequence we are using is 720x486 it means it is shrinked 1:0.9 and we have to compensate this, that's why we have to configure the Placement of the Image Plane and set the Fit attribute "to size" and to click the Fit to Film Gate button.
Now you can track your scene, because just as you are seeing the image plane and the objects is what you're going to get in your Maya and RenderMan render.
- Armando Ricalde (Doc)