Is there such a thing as “native” 3D in Stereo 3D?

A subject that I often stumble upon in various articles, conferences and comments is the question of content being “native 3D” and this is often used in opposition to converted content.

First, let’s clarify the question of “native 3D”. I think that the usage in this context is abusive and that it is a bias form to try to qualify non-converted content. In fact the only native 3D is the reality, the reality is in 3D, the representation of 3D by effects or tricks to your eyes and brain is not native 3D.

Stereoscopic 3D in television or theater is an effect and it plays on parts of how you perceive 3D; the one where your eyes and brain uses stereopsis to “see” depth. Those who can’t see stereoscopic images will not see any depth even if in the real world they would (with monocular depth cues).

So Stereoscopic 3D is an effect and therefore you can’t really talk about native 3D.

As an effect, the perception of 3D can be easily distorted, minimized or exaggerated. Production of Stereo 3D images is not trivial and requires a lot of care, planning and expertise.

When people discuss about “native 3D” they are generally referring to:

  • content that is “camera acquired” with pairs of camera specially calibrated to capture stereoscopic pictures (often with special rigs);
  • content that is “camera acquired” with a camera equipped with a pair of lens calibrated at a fixed width;
  • content that is “virtual” and created with 3D models and viewed by a “pair” of virtual camera that creates the two views (mainly 3D animation like the Pixar movies or virtual scenes in Avatar).

The “virtual” type of 3D is the one that is easier to handle and gives the most predictable result. Since the virtual environment is well under control the experience of the stereo 3D version can be quite predictable and quite good.

Both types of “camera acquired” have their advantages and inconvenient, I won’t go in details today about it but if it is not handle with care and expertise, the result can be quite unpredictable. In fact, you can take a look at this list from Director James Jim Cameron to get an impression on how tedious it can be.

As an example of the complexity and care that you must take, when you shoot with two camera, you have the ability to not only vary the interocular distance (which your eye can’t do, but your eyes can’t zoom either) but you can also converge at different distance (like your eyes do)… however, if you do that, you must make sure that your focus and depth of field are appropriate to the converged distance; if not, people may look at the background and see aberrations in it due to the closer convergence (people will not notice if the background is fussy and that’s what you need to be careful about).

On the other hand, with a stereoscopic camera with two lens at a fixed interocular distance, you can’t vary the distance between the lens and you can’t converge. The result of that is that only the average distance will look somewhat realistic. If it’s too far and you zoom in the 3D effect will be less than what you would expect in reality and if it’s too close the object will look somewhat distorted as you can’t converge on it.

There is also a lot of conditions that can affect how realistic a 3D picture look. For instance, I saw an american football match in 3D and it was raining. The fact that the rain was located mostly in front of the screen (negative space) and that the TV didn’t fill my whole field of view created a strange feeling where the rain drops were in the middle of the room appearing and disappearing from nowhere! That was bringing my attention to that and I was not enjoying any other aspect of the 3D experience.

Don’t get me wrong, “camera acquired” content can be quite good but it is something that requires expertise and means to achieve a good result. It is too easy to put cameras in anybody’s hands and to create a sub-standard 3D experience that lower the reputation of 3D.

On the other hand, conversion can be quite good and sometimes superior to “camera acquired” content. This requires also expertise and means to give great results. I saw 18 minutes of the upcoming result of the conversion in 3D of “Titanic” and it was really really good. This look as good as most top quality “camera acquired” content and I think that it shows that it is as much a legitimate technique than “camera acquired” or “”virtual” content to produce 3D.

Again stereo 3D is an effect and treated as such and with care it can be good regardless of the source or it can be quite bad regardless of the source is not treated correctly.

I’ll touch more on conversion in another blog post!

Leave a Reply