50mm Prime with Resolution for a D800? Come on Nikon/Sigma!

Started Jan 4, 2014 | Questions thread
MisterHairy Senior Member • Posts: 2,218
Re: 50mm Prime with Resolution for a D800? Come on Nikon/Sigma!

Scott McMorrow wrote:


How are you measuring resolution? Do you mean MTF50, MFT30, ,MTF10, or some other measurement criteria? I'm quite aware of Nyquist and aliasing. I deal with it all the time in my day job modeling, simulating, and designing ultra high bandwidth digital communication interconnect.

You've not yet defined what you mean. Low contrast image energy beyond Nyquist can cause aliasing, which due to it's repetitive nature sticks out like a sore thumb. Moire is very sensitive to even minute amounts of image contrast. Just because there is enough image contrast to cause moire does not mean that it's very useful contrast. Which is why measurements such as MTF50 are used for meaningful resolution comparisons.

If you tell me that a particular lens on the D800 out resolves the sensor using an MTF10 measurement, I'd absolutely agree with you. MTF30, then I'd start asking how the measurement was made, and how the image was sharpened. MTF50, except for a few lenses I would call BS. I"ve made measurements on a few of my lenses with Imatest, and have correlated the results to Roger Cicala's unsharpened RAW measurements. I've then looked at the MTF50 results with Imatest's optimal sharpening filter and with a sweep of different sharpening parameters using Lightroom. None of these results approaches the theoretical resolution of the D800 sensor when measured at MTF50, and definitely not when looked at from corner to corner, with center image focus.

Then we have some professional experience in common. While I worked in the real world, I spent a couple of decades designing and developing the hardware and software for scalable, simultaneous data sampling and transmission which is heavily in use around the world today. OK, not terribly fast - only up to a few MHz aggregate rate, but the basic tenets of digitisation are independent of sample rate. It is possible that you yourself use some of my IP in your work.

Your discussion of contrast is a bit of a red herring in this argument. One might consider a higher contrast source as one containing a greater proportion of harmonic content while a lower contrast one will have less - perhaps only the fundamental. A simple analogy, pertaining to linear, time based sampling, would be a square wave (highest possible contrast) versus a sine wave (lowest possible). The harmonic components of the square wave will induce aliasing on an unprotected sampler far sooner than the sine wave, even with a fundamental sitting well well below Nyquist. Any signal components (not just fundamentals) which sit above the Nyquist frequency for a given sampler will cause aliasing in the output of that sampler if adequate protection is not provided. Nothing new there. Digitisation 101.

All well and true and this explains why one might be prepared to see aliasing from a higher contrast source when testing a lens (or even just using it for, well you know, photography). However, to blindly state that higher contrast sources will promote aliasing without thinking about the implications for lens resolution is to miss the point. The high contrast source can only cause aliasing if those higher harmonics (the cause of the higher contrast) are transmitted through the lens to the sensor (sampler).

The only way in which we can experience aliasing in an image is if the lens is transmitting frequencies above the discrimination limit of the sensor. It does not matter if these frequencies are the fundamental of the sampled pattern or the Nth harmonic; they are still reaching the sensor because the lens is successfully transmitting them.

Therefore, the lens is able to provide frequencies above the discrimination limit of the sensor - it is effectively "out resolving" it (there's your definition).

Any time that we see aliasing (moire) in an image it is happening because the lens is doing this. If the sensor has alias protection in the form of a low pass filter then the intensity (amplitude) of the high frequencies must be commensurately higher to be still making their presence felt.

It has also been most amusing to see individuals quoting frequency limits for lenses measured on different hardware with different sampling capabilities. Using a lower frequency device with a strong low pass filter (a D3x) to predict lens characteristics on a higher frequency device (D800[E]) is a nonsense when the D3x and its AA filter are the limiting factors on the measurement capabilities.

 MisterHairy's gear list:MisterHairy's gear list
Fujifilm X100F Sony Alpha a7R III Sony Alpha a7 III Sony FE 35mm F2.8 Sony FE 55mm F1.8 +10 more
Post (hide subjects) Posted by
(unknown member)
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow