Details, 3D details

In comments Gregory suggested that a proper test for resolution of 3D systems might involve such things as the perceptibility of really small text. I agree entirely.

And apparently such tests have been done. These are recounted to some extent in the decision by the National Advertising Division of the US Council of Better Business Bureaus.

You can read about one of the small text tests at Dr Soniera’s website. This is based on examining some text within a Blu-ray 3D presentation, which is more or less the way I tend to do things. Dr Soneira’s evidence was adduced in the NAD’s decision to the effect that passive 3D TVs provide full resolution.

The other test was by a ‘clinical research study’ by Drs Martin Banks and Joohwan Kim entitled ‘Spatial Resolution of Temporally and Spatially Interlaced Televisions’. Now it turns out that Dr Banks runs something called the Visual Space Perception Laboratory at the University of California, Berkely. Perhaps I’m misinterpreting, but it’s hard to see Dr Banks’ qualifications as anything other that impressive for conducting this kind of study.

What follows is taken from the NAD’s decision, and is thus filtered through the NAD’s paraphrasing. Here’s what Samsung claimed:

Samsung provided nine separate reports showing that the images delivered by passive 3D televisions were measurably and visibly less detailed than those provided by active 3D televisions. In addition to the Samsung Korea, Intertek and Underwriters Labs (“U.L.”) reports, Samsung provided the report of a clinical research study by Drs. Martin Banks and Joohwan Kim (“Banks report”), and a comparison of Samsung and LG 3D display resolution by Joe Kane (“Kane report”). Samsung also provided published, third-party reports from expert reviewers at Consumers Union, HDTVTest.com, CNET.com, and TrustedReviews.com. All of these reports and reviews, without exception, state that the evaluated passive 3D televisions delivered less detailed 3D images than active 3D televisions (i.e., delivered only a maximum of 540 vertical scan lines to each eye when in 3D mode) but that active 3D delivers 1920x1080p to each eye in 3D mode. All of them attributed the loss of detail with passive 3D to the spatial interlacing method used by passive 3D televisions to display the respective 2D images to each eye, which inarguably cut the source resolution in half.

Focusing on the Banks report:

Samsung’s expert, Dr. Martin Banks, Professor of Vision Science at the University of California at Berkeley, explained that the two half-resolution images created by passive 3D technology are not recombined in the brain to create a full resolution image, as claimed by Dr. Soneira. Dr. Banks’ research demonstrates that the effective human visual resolution of the Samsung active shutter system is equal to the full resolution of the television because both eyes receive information from all the pixels in the television. Dr. Banks’ study shows that LG’s passive 3D approach achieves only half resolution to each eye and that there is no way the binocular image could have more than half resolution vertically. Thus, Dr. Banks concludes, the “effective spatial resolution is higher with the temporally interlaced Samsung television than with the spatially interlaced LG television except at long viewing distance (six times picture height) where the effective resolutions are the same.”

Soneira had a go at Banks’ methodology:

Insofar as Dr. Soneira criticized Dr. Banks’ research by arguing that it utilized “artifical” 3D content, Dr. Banks responded that the “letter acuity test” he used in his study is a worldwide standard for optometrists, ophthalmologists and motor vehicle departments to assess visual acuity. Indeed, he stated that it is much more accurate to use a letter-acuity test, with high contrast lettering of a standard font, to assess display resolution differences because it controls for artifacts that could be inherent to consumer 3D video content. Further, it is a non-biased methodology that removes any potential bias or variability, unlike the Soneira’s testing which relied on the test administrator to select visible text within a consumer video on an ad hoc basis.

Dr. Soneira further criticized Dr. Banks’ research for requiring participants to guess what letter they think they saw after seeing it for 600 msec, Samsung countered that this criticism is unfounded and is based on a lack of understanding of consumer testing methodology. The use of a forced-choice methodology in this study and even shorter durations than 600 msec. is standard in vision science (and other consumer testing). Letter acuity is unaffected by stimulus duration for durations greater than 400-500 msec. According to Dr. Banks, “[b]y forcing subjects to make a response even when they are uncertain, we can eliminate performance differences due to personality variables (i.e., the willingness to make a possible mistake) The issue at hand is the effective resolution of the TVs, not the subject’s willingness to make a response.”

LG responded to the Banks report by noting:

The Banks report, for all of the technological information provided, employed the use of visual acuity charts. While such testing might well be sufficient for visual acuity or other technical scientific discussion, insofar as it is offered for advertising claim substantiation (indeed, for superior picture resolution and 3D picture quality claims), it bears no resemblance to actual 3D content typically viewed by consumers.

It added:

The Banks report relies on a test methodology that has no relation at all to actual 3D content viewed by consumers and did not involve actual 3D TV content but, rather, employed the use of test charts — hardly adequate support for Samsung’s broad claims that its active 3D TVs provide double the resolution and superior picture quality over passive 3D TVs. Moreover, Dr. Banks purportedly found that Samsung had higher 3D TV picture resolution at distances of 1.5 and 3 times picture height — far less than the viewing conditions found in most households, as well as Samsung’s own recommended viewing distances (i.e. three times or more the height of the screen.) Indeed, Dr. Banks’ study failed to test picture resolution at 4.5 times picture height (a closer approximation of normal viewing conditions) and inexplicably jumped from 3 times picture height to the much farther 6 times picture height.

(My bold.) The bolded phrase is irrelevant to the current discussion (I for one tend to the view that LG’s 3D actually provides a slightly ‘superior picture quality’ in 3D mode, all things considered.)

NAD was persuaded by LG’s claims. I am not.

As I have noted elsewhere, one important reason for high definition is to allow you to sit closer to the screen so that it can occupy a larger angle of your vision, when compared to SD. If you are going to invalidate tests which suggest that you get more detail at those closer distances, then why even bother with HD? Of course if you sit far enough back from a 3D TV, any loss of resolution will become imperceptible.

Why Samsung recommends a viewing distance of 3 times the screen height I do not know. LG, with its current range, recommends that a viewing distance of twice the screen diagonal in 3D mode. Consider a 55 inch TV. Three times the screen height (Samsung’s apparent recommendation) is 2.05 metres. Two times the diagonal (LG’s recommendation) is 2.79 metres. To put in another way, LG’s recommended viewing distance is 36% greater than Samsung’s. Perhaps there is a reason for this.

I would note that I’ve just finished reviewing Sony’s Personal 3D Viewer headset. This purports to offer an image equivalent to viewing a 750 inch screen at a range of 20 metres. That’s the same angle of view as watching a 55 inch screen at 1.47 metres.

And that only offers a 720p display!

I have watched a lot of stuff on both active and passive 3D TVs. Fact is, at a range of 2.7 metres the same real world 3D content looks more detailed and somehow whole on my 83.5 inch front projection screen using an active DLP 3D system than it does on a 55 inch passive 3D system. That’s not to say that a Samsung direct view TV of a given size is ‘better’ in an overall 3D sense than an LG set. It is to say that that the vertical resolution as perceived by a real human being on a current model passive TV set is less than that available from a best practice active display.

In reality, I find ludicrous the suggestion that a passive 3D TV somehow offers full vertical resolution. It simply doesn’t.

This entry was posted in 3D, Testing. Bookmark the permalink.

3 Responses to Details, 3D details

  1. Ryan Sandford says:

    Hi Stephen,
    Excellent post and article. It seemed from reading that what Dr Banks study was alluding to comes back to my previous mention of the temporal vs spatial element in Sten de Witts blog which I linked to earlier. Since last posting I have had an opputunity to have a look at a new sony active set, and whilst I honestly couldnt perceive the flickering and was impressed with the image generally, I did notice a ghosting I have never seen on my passive, similar to the old ghosting on anolod tv broadcasts.I wonder if this is common to active displays?
    I have also found another interesting take on the debate, which is well worth a read though covers alot of the ground previously, it does seem to enscapulate it well. The comments also are very informative – http://www.audioholics.com/news/editorials/active-vs-passive-3d .

    I hope this helps with your ongoing work. It seems to me a battle between the views of Dr Soniera vs Joe Kane & Dr Banks, with yourself as an impartial judge/referee weighing the available evidence,testing it and making a educated &informed ruling(so to speak).
    Anyway,thanks again and I hope you continue this great research.

    ps. I recently imported the 3d blu ray animated Sammy’s Adventures. If you have not seen this, I highly reccomend it to you and your readers, as the picture quality & 3D are incredible and I would like to hear your review if possible. The quality seems to surpass even Toy Story 3,Open Season & Despicable Me, which I find pretty amazing!

  2. My new post discusses some stuff from the audioholics post. Regarding ghosting, this is indeed a major problem with 3D TVs and has been from the start.

    But last year’s TVs tended to be better than the previous year’s, and I expect that this year’s will be better than last year’s, when I get to review them.

  3. Incidentally, one exception to the improvement last year was Sony. I hope it’s better this year.

Leave a Reply

Your email address will not be published. Required fields are marked *