(The image above is NOT from the original S3D version of Avatar which had correct 3D. Click for larger)
What is Pseudo Stereo 3D:
The simple way to answer this question is to watch any 3D video with the glasses reversed. Making sure the left lens of the glasses is over the right eye. Most likely it will lead to visual confusion for many. But quite a large number of people do not know that they are watching reversed or inverted 3D. To make matters worse, depending on the actual scene there might not be anything “obvious” that is wrong, fooling even experienced stereographers. (This is from witnessing it first hand while on location. More on that later).
Pseudo Stereo Explained:
Click on the first image above to enlarge. View this image with red-cyan anaglyph glasses. Notice anything odd besides the shortcomings of anaglyph color technology? Chances are not many will find it odd, and in-fact it looks like a real 3D still image from the movie Avatar. However this is a 2D to 3D conversion done in under 30 seconds from a 2D version of the movie.
So where is the pseudo stereo in it?
Look carefully at the small Helicopter near the left side of the Big Mountain. It is situated at wrong depth from the rest of the image. Infact there is something quite disconcerting when looking at the image.
Now, swap the anaglyph glasses around. you will see the Helicopter seems to be nicely floating out of the screen looking correct. But look at the clouds now, cutting through the mountain. THIS is what pseudo stereoscopic 3D imagery is (previously the helicopter exhibited pseudo stereo). The other mountains also now look wrong with back mountains cutting through the ones in front.
Pseudo Stereo 3D Can hurt Audiences:
The eyes supply the two images to the Brain which then fuses the images together and gives us the “illusion” of true depth when watching stereoscopic imagery. However, when presented with pseudo stereo, in the background the processing pipeline between the brain and eyes goes like this:
Eyes: Here’s the two images as we see it, go ahead and fuse them
Brain: Ok but this one bit is not making sense, can you focus/converge on the helicopter
Eyes: It looks fine.
Brain: hmmm ok try one more time…
Eyes: done. Nothing bad. it’s in focus, there’s even good parallax so our muscles arent straining to converge/diverge
Brain: hmmm, ok then…moving on.
As you can guess, if this is a feature length production, quite a bit of strain will happen in the background, that may lead to senses strain in many people, and definitely in all people once they realize how to identify Pseudo Stereoscopic imagery.
Another un-intended side effect can be, that it detracts from the main story unfolding as audiences’ eyes will lock onto these anomalies, while the brain in the background tries to make sense of the false depth cues being presented to it.
(click for larger)
Pseudo Stereo 3D ON AIR:
With the lack of true stereoscopic 3D content, it is tempting for both Broadcasters and Content producers to resort to using today’s breed of so-called professional conversion solutions to convert 2D to 3D. At the very least, rather than produce pseudo stereo 3D content by using the “analysis” settings on these boxes, it is our sincere request to simply refrain from that option, and instead use the boxes in output modes such as seen in the image above. This may look flat and simply “mapped” 3D, but at least it won’t cause harm to audiences, as wrong depth cues can.
The way forward to 3DTV success could be in a form of a hybrid 2D signal with 3D embedded.
Real time 2D to 3D converters and the Danger of Pseudo 3D:
The image above is from live screen grabs taken during last years FIFA 3D opening ceremony, where a 2D camera feed was passed through (I believe) a JVC realtime 2D to 3D converter box. The image above does not have any Pseudo Stereo 3D, which is a good thing, because thankfully the settings seem to have been set to simply “paste” the FLAT 2D video on some combination of geometry and offset it into positive parallax (into the screen effect). As can be seen the images look 3D but there is no differentiation of depth between say the hands in front of the face, or the foreground player and the backgrounds.
Meanwhile I was invited on another occasion recently along with TV broadcast engineers to witness the other capabilities of the box, namely live analysis of a 2D signal and processing of the same signal in realtime to convert the signal into stereoscopic 3D output.
This is where in my opinion the device failed…rather a lot! On every scene I could point out quite a few Pseudo stereo effects. At first the Broadcast TV engineers did not pick up any of these false depth cues, until I physically walked to the 3DTV and pointed them out. As the engineers then learnt to identify Pseudo stereo, it was clear to see that this “mode” of the converter box would not be acceptable to convert 2D content to 3D, and if at all the box was to be used, it could only be used for doing “3D ish” effects such as the one in the image above.
It is not our intention to single out the JVC box or target it for Pseudo 3D. I’ve seen similar examples with ALL 3DTVs on the market today that offer this “pseudo” 3D option. One other professional solution that I witnessed in the R&D labs of a manufacturer was also still-not-there yet.
To put this into perspective, James Cameron is spending a year and approx $18 million to convert Titanic, if a box could do it without pseudo stereo, it would have been done in a week ( I say a week, because promoters of these boxes say that it “aids” post production and can do up to 80% of conversion in automated fashion)
In truth I don’t think it can do more than 5% in real time, and when it does true Stereoscopic 3D conversion, it’s mostly because the scenes lend themselves to time-multiplex stereo 3D in the first place. Some depth-map based software do better, but they are not real time and usually require manual cleanup.
Two Types of Pseudeo Stereo 3D:
From the comments below the article, I have to clarify that there are Two types of Pseudo Stereo 3D occurances:
1) The easy to correct version: In this version of Pseudo stereo, it’s a matter of simply swapping the image pair being presented or projected, or swapping the glasses around to fix it
2) The Dangerous Variety: This is what the article is all about. In this version of Pseudo Stereo 3D, the actual left/right image pair are presented correctly. but “elements” within the scene are at wrong depth. This comes about largely due to automated 2D to 3D conversion boxes, and at times due to post production Compositor mistakes. Usually it’s because not all people in the post production department can readily identify pseudo stereo 3D mistakes.
Pseudo Stereo 3D advice for Stereographers On-Location:
As mentioned in the beginning of this article, I was on assignment providing a client with Stereo 3D QC services for a shoot that already had a dedicated 3D camera crew and Stereographer. A shot was being set-up and ready for take, when at video village I put on the 3D glasses and noticed something abnormal. On turning the glasses inside out, it was clear that the shot was inverted and in pseudo stereo.
The nature of the shot was such that it was not obvious, one reason was there was a foreground layer with no in-between depth cues, just a far background.
Correcting such a shot is not a simple matter of swapping the images in Post! Why? because depth decisions, parallax, interaxial etc were all being made on-location while watching an inverted 3D live signal. The only pseudo stereo cue was a bit of clothing on the main talent showing at wrong depth.
Advice to any Stereographer:
Whether experienced or otherwise… is, as best practice, to always look at a scene with the glasses swapped, regardless if something in the scene looks abnormal or not. This is also good advice to the Post production department, both in VFX and Edit.