[opendtv] Re: Image quality
- From: Mark Schubin <tvmark@xxxxxxxxxxxxx>
- To: opendtv@xxxxxxxxxxxxx
- Date: Fri, 11 Nov 2005 12:00:38 -0500
As my name has been invoked here, I suppose I need to make some comments:
Craig Birkmaier wrote:
>It is not the number of samples per line or number of lines that are
>important per se'.
>
>
Agreed.
>What is important is oversampling relative to the format that you
>plan to use for emission, and care to make certain that the integrity
>of the samples that will be encoded for emission are of the highest
>quality.
>
Oversampling is, in general, good. But it is only good if the system
can deal with it. If the lens or the prism are limiting factors,
oversampling isn't necessarily a good idea.
>The film source is oversampled
>
Except in the temporal domain, film is not sampled. Grain is randomly
distributed.
>The cheap 1/3" cameras are undersampled - the sensors typically have
>no more than 1440 samples per line and possibly even less.
>
The cheapest of the 1/3-inch HD camcorders actually has a 1920 x 1440
sensor, of which 1920 x 1080 is used for HD.
> The
>acquired samples are then resampled to about 960 samples per line for
>compression and recording.
>
HDV, in 1080 mode, records 1440 x 1080 luma -- same as HDCAM and more
than DVCPROHD. And, in the horizontal direction at least, it actually
captures MORE color detail than does HDCAM (or DVCPROHD).
>To
>do 1920 x 1080 right you need to sample at AT LEAST 2880 x 1620 with
>progressive frames, then resample to either 1080i or P. 2x
>oversampling is even better.
>
>
If you do that oversampling on a 2/3-inch imager, you will have greater
diffraction-caused reduction of MTF (at least in the vertical direction)
than going to a 1/3-inch imager. You can't just increase the number of
sensors without considering the optics. My paper at SMPTE yesterday was
called "The Optics of Small-Format HD Acquisition."
>Since most 1080 line cameras do not oversample, and they use
>interlace for high temporal rate acquisition, the result is that the
>potential spectra of the format is not filled. Another way of saying
>this is that the frequency response for 1920 x 1080 should extend out
>to at least 27 MHz. Unfortunately, when you sample at the same
>density as the format you are trying to create, most of the
>information above 20 MHz is just noise, or has virtually zero MTF -
>i.e. NO DETAIL.
>
>
Not so. There is significant energy at even 30 MHz. Larry Thorpe
showed MTF at 30 MHz for various imager formats. 1/3 inch was lowest,
but even it was well above zero.
>And Mark nailed it with respect to the limitations of 1/3 inch sensors.
>
>HDV is a low cost stop gap format.
>
The issue is not HDV (although it has its own issues related to
compression). It's the imager size and video format. JVC, I seem to
recall, has a 2/3-inch HDV camcorder, which should perform nicely. And,
although Panasonic has not yet released details on the imager in its
HVX200, it is 1/3-inch even though the camcorder is not HDV.
>As Mark will tell you, the human visual system is more sensitive to
>contrast than to fine detail.
>
>
Not exactly. Contrast without resolution is as meaningless as
resolution without contrast. But the psychovisual sensation of
"sharpness" is proportional to the square of the area under an MTF
curve, and the bulk of that area, thanks to such issues as
diffraction-limited MTF, is at lower resolutions rather than higher.
Still, there are contributions to image quality from higher
resolutions. In Larry Thorpe's paper yesterday, he spoke of huge
differences that are apparent when shooting fabric textures based on
energy at higher resolutions. I did something similar with some
pictures of corduroy. You can find something along those lines here:
http://www.cambridgeincolour.com/tutorials/diffraction-photography.htm
Click on the various apertures below the corduroy.
There is talk of a limit of human resolution around 30 cycles per
degree. If that were true, we couldn't see the stars at night. We can
see those stars because of the huge contrast between them and the
background. The 30 cpd figure is for TV-viewing conditions. And
Craig's 22 cpd has a strong basis in some researchers' findings of our
high-end resolution sensitivity at about 50% MTF (we're also limited on
the low end, as a glance at a contrast-sensitivity-function grating will
quickly show:
http://ohzawa-lab.bpe.es.osaka-u.ac.jp/ohzawa-lab/izumi/CSF/CSFchart640x480.gif
Do you see the downward-shaped curve across the top? It's not there.
Your visual system is putting it in. The right side of the curve should
change significantly as you get closer to and farther from the screen.
TTFN,
Mark
----------------------------------------------------------------------
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:
