[opendtv] Re: Public M/H information

  • From: Craig Birkmaier <craig@xxxxxxxxx>
  • To: opendtv@xxxxxxxxxxxxx
  • Date: Mon, 22 Dec 2008 09:37:06 -0500

At 7:33 PM -0500 12/21/08, Manfredi, Albert E wrote:
Craig Birkmaier wrote:

 That being said, Apple is selling 360P via the iTunes store...

That's close to right, if you use the 1.6 arcmin criterion for visual
acuity. The closest viewing distance should be 6 picture heights, at

I believe that the 1.6 arcmin criterion is unrealistic.20/20 vision is 1 arcmin.

People expect pretty pictures, it's really that simple. It might not be
all they expect, but nevertheless...

I'm intrigued by Mark Schubin's quote of the NHK research findings. I'll
bet the fact that we can discern differences of much less than even 1
arcmin has to do with our eyes rescanning the image from slightly
different viewing angles. Whereas the 1 or 1.6 arcmin acuity assumes
looking straight at the image without moving the eyes at all. Some
effect like that.

You are confusing the issues here.

It is one thing to talk about the smallest detail that the human visual system can discern under ideal conditions using a static test image such as a line pair. But this is completely IRRELEVANT for motion imaging systems.

There are two significant factors at work here:

1. Nyquist filtering to produce fluid motion - directly related to the temporal sampling rate

2. The ability of the human visual system to resolve fine details in moving images

Nyquist filtering is a huge hit, especially at lower frame rates, which is highly relevant to a mobile video service. 24P require significant motion blur - lower frames rates as might be used in a mobile service require even more severe filtering. And remember, digital compression acts as a low pass filter, removing image detail, so aggressive compression will also reduce the available detail.

The ability of the human visual system to resolve fine details in moving images is also a significant consideration. With motion tracking it is possible to follow fine details, but even here it is not possible to resolve as much detail as is possible with static test images. And motion tracking is far less relevant for a small hand held display, which will typically only cover s small portion of the visual field.

So I would warn against putting too much faith in theoretical limits of human vision when we talk about motion imaging systems.

As you say, people like "pretty pictures," not necessarily sharp pictures.


You can UNSUBSCRIBE from the OpenDTV list in two ways:

- Using the UNSUBSCRIBE command in your user configuration settings at FreeLists.org
- By sending a message to: opendtv-request@xxxxxxxxxxxxx with the word 
unsubscribe in the subject line.

Other related posts: