3D movies are hitting the movie theaters like tidal waves! The characters and objects come reaching out of the traditional flat screen, to provide an even more realistic in-movie experience. Plus IMAX 3D and growing popularity of digital projectors, the traditional film projectors are gradually a thing of the past.
But what’s “true 3D” and what’s “mock 3D”, and how about those “2 and half D”? Currently the industry is still experimenting with different techniques to create the “perfect” look… with different style cameras (IMAX, HD), different setups, even ingenious innovations by film such as “Avatar”. The main obstacle is still the budget. The high cost of having two cameras per unit prevents smaller studio to attempt for the true 3D setup. Not only that, production cost of dealing with stereoscopic footages is also significantly higher, due to each shot has to generate both the left eye & the right eye footage, which is twice as expensive.
So what are the difference between all the 3D movies on the market? First, the 3D animated films, such as “Toy Story 3″, renders the film with 2 in scene camera, creating an accurate Z depth information (X & Y are for the horizontal and the vertical, while Z is for the depth value into space). Unless there are many post special effects / motion blur adding to the raw render during compositing stage, which might encounter some stereoscopic issues, or else it is a great representation of a true 3D movie.
As for live footage films, different production has their own unique setup to achieve the look. But in general:
1) Two separate cameras shooting at the same time: Expensive, but this will provide one of the most accurate depth interpretation. Although most likely you will have to compensate for the difference between two cameras – polarization, coloration, or vertical alignment issues… etc. It would also run into the same issues when adding special effects, such as explosion, rain, smoke, blood… etc, because each effect has to have z-depth build in, in order to merge in with the footage seamlessly.
2) One camera with a mirror build in for image offset (the a negative offset on Y-axis). This is a slightly inexpensive way of dealing with shooting 3D footages. Still effective, but the difference between both eyes are much greater, especially the polarization issues from the (mirror) reflection needs to be addressed separately. Also if someone on set has bumped into the camera and cause the mirror to rotate or changing focus, you will most likely hear screams half way across the globe.
3) 2D to 3D conversion. By generating 3D models for each shot, to determine the overall depth of each key objects/characters. Then use the generated Z depth to create clean plates. Let the left eye being the fixed eye, an offset right eye information will be created with occlusion areas, to be later on patch in by the clean plate. This method can generate convincing 3D effects, but many limitations such as motion blur area, or ghosting images are common occurrences during this style of conversion.
4)Similar to the 2D to 3D conversion, but not as emphasize on the overall depth interpolation. The whole movie looks like cut out puppets inside a boxed-in set, which for some earlier animated movies, could be an interesting effect. But overall it has a very cheap look, and not offering an immersed 3D experience. (Some industry people would much prefer to see the film in 2D then “suffering” through this style of under developed 3D).
5) 3D generation through software! This is insane! how is this even possible but a gimmick from the TV and the software companies. Sorry, but there is just NO WAY to generate an accurate 3D simply by offsetting footages using perspective warp and corner pins. Seriously, if it is so simple, then all the stereoscopic film studios would already invest in the technology, why even hire artists with 3D stereoscopic experience (hitting my head against the wall)!
Hope this breakdown would offer whoever that’s reading this blog some more insight into the stereoscopic world. It is truly amazing to work with true 3D footages, and the end result is extremely rewarding! If you have any other discoveries, feel free to share.
signing off…
-e
