[opendtv] Re: Image quality
- From: Tom Barry <trbarry@xxxxxxxxxxx>
- To: opendtv@xxxxxxxxxxxxx
- Date: Thu, 10 Nov 2005 23:15:16 -0500
Mark -
Thank you very much. I've saved this one for future study.
I still don't pretend to understand it all but generally consider those
results fairly depressing. :-(
- Tom
Mark Schubin wrote:
> It IS a very small number. Let's look for red diffraction-limited
> resolution on a 2/3-inch camcorder. A 2/3-inch camcorder uses an imager
> with a 11-mm diagonal. In 16:9 mode, that's about 9.6 mm wide. 1920
> pixels per line is 960 line pairs. 960 line pairs in 9.6 mm is 100
> lp/mm. There's a range of wavelengths for red light, but let's pick 630
> nanometers, which is 0.000630 mm. Solving for f, we find that at any
> f-number from f/13 and lower, 1920 pixels on that camera has zero MTF,
> i.e., cannot be resolved. Okay, f/13 is not so bad. So let's move to
> 50% MTF -- very little energy. Now it's f/6.5. Let's move to 75% MTF
> -- certainly resolvable, but still pretty grey, with no black or white.
> That's at f/3.3. We can't quite achieve a diffraction-limited MTF of
> 87.5% on a 2/3-inch-format HD camera -- and this is ONLY diffraction,
> not lens MTF, which I'll get to in a moment.
>
> Now let's switch to a 1/3-inch imager. Its imager has a 6-mm diagonal,
> so the image width is about 5.2 mm. Now zero MTF at 1920 is at f/7.1,
> 50% is at f/3.5, and 75% is at f/1.8. So far, so good. The format
> factor says that f/1.8 on a 1/3-inch camera is roughly equivalent to
> f/3.3 on a 2/3-inch camera for sensitivity (except for microlens
> efficiency) and depth of field, too. But I can open up the 2/3-inch
> camera more; on the 1/3-inch camera, I'm stuck.
>
> Now let's consider lens resolution, using the same lp/mm factor. In the
> early days of HD, most cameras used 1-inch-format imagers. At 1920
> pixels per line, a 1-inch format imager (16-mm diagonal, 13.9-mm wide
> for 16:9) needed only 69 lp/mm. A 2/3-inch camera needs 100. A
> 1/2-inch camera (8-mm diagonal, 7 mm width) needs 138. A 1/3-inch
> imager needs 184. So the 1/3-inch-format camera needs roughly three
> times the lens resolution of the 1-inch-format camera.
>
> The glass in the 1-inch-format lenses was cooled over the course of
> months to reduce microfractures that could reduce MTF. That helped
> justify their extraordinary cost. Complete 1/3-inch HD camcorders are
> cheap. Yet they have lenses with three times the resolution of the
> 1-inch format??? I don't THINK so.
>
> Canon's Larry Thorpe gave a paper at SMPTE today in which he showed an
> MTF chart of one of their best 2/3-inch-format HDTV lenses. It fell off
> at high resolutions, but not too badly. Then he showed the MTF of the
> exact same lens reconfigured to work on a 1/3-inch camera; the MTF
> plummeted.
>
> I hope that helps your understanding.
>
> TTFN,
> Mark
>
>
> Tom Barry wrote:
>
>
>>Mark -
>>
>>I'm afraid I am not familiar with that formula and don't understand it.
>> Could you give some typical numbers?
>>
>>For instance I'd think lamda would be a very small number as the
>>wavelength of visible light.
>>
>>- Tom
>>
>>
>>Mark Schubin wrote:
>>
>>
>>
>>>>"high quality out of focus picture" does seem humorously
>>>>oxymoronic to me.
>>>>
>>>>
>>>
>>>Indeed! But it might be worthwhile for HDTV shooters to consider lens
>>>diffraction and other optical foibles that can degrade images.
>>>
>>>Here is the formula for monochromatic diffraction-limited modulation
>>>transfer function:
>>>
>>>MTF = 1-(1.22*f*lamda*lp/mm)
>>>
>>>where f is the numerical aperture,
>>>lamda is the wavelength of light, and
>>>lp/mm is the number of line pairs per millimeter (nominally the number of
>>>lighter and darker points per scanning line divided by two divided the width
>>>of the imager)
>>>
>>>Because it is a "one-minus" equation, the only way to get 100% MTF is to
>>>have either zero input frequency (the whole screen is one shade) or a zero
>>>numerical aperture (the lens is infinitely large).
>>>
>>>We do a lot of dancing around about 1920 x 1080 when it's becoming less and
>>>less likely (with smaller and smaller imagers) that there will be any
>>>significant energy at the high end of that.
>>>
>>>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.
>>>
>>>
>>>
>>>
>>
>>
>>----------------------------------------------------------------------
>>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.
>>
>>
>>
>>
>>
>
>
>
> ----------------------------------------------------------------------
> 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.
>
>
----------------------------------------------------------------------
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:
