Canon EF 14mm f/2.8L II USM

wimg wrote in post #3798685
First prices I see over here are 2200 euros....


Please note that the notion that with dslrs we have lens limited resolution is a typical internet myth. Resolution of lenses is way higher than that of sensors. The 21 MP sensor will bring us to about 78 lp/mm, while the best lenses currently manage up to 400 lp/mm at F/4 (which is about the theoretical maximum).

Note that the sensor of the 1Ds Mk II has a limit of about 65 lp/mm, which is what can be measured, more or less (62 lp/mm) with a good lens (85 F/1.8 ), at F/8!

It is just that sensors are less forgiving than film; it is easier to see any aberrations, because it is so easy to pixelpeep.

Kind regards, Wim

Hi Wim

Whilst what you say is true it should need some qualification for example ~69 lp/mm is the Nyquist limit for the 1Ds II and is the mathematical maximum that can be sampled without aliasing.

The actual capability will be less than that depending on the MTF characteristics of the anti-alias filter and MTF characteristics of the sensor (which is not unity from 0 lp/mm to the Nyquist limit as many think). The combination of the two MTFs is known as the sampling aperture for reasons dating back a long way.

Another factor comming into play is how much sharpening is applied. My estimate for the 1Ds II with Canon recommended sharpening is a reasonable MTF up to about 52 lp/mm.

Now the 400 lp/mm for a lens is a bit good, in any event these sort of values are old style 'cut-off' resolutions with probably a 5-10% MTF, so not really of much practical interest.

Lens 50% MTFs, a much more relevant factor, can be much lower even around 30 lp/mm or less for wide angle and ultra wide angle specially in the corners. But even an improvement in lens MTF from say 90% to 95% at 10 lp/mm will have a significant impact on image quality.

My point is that lp/mm is just one dimension and is meaningless without the MTF dimension.

On full frame the most significant resolution range for image quality is 10-30 lp/mm, really fine resolution up to about 40-50 lp/mm, what maters in this range is the overall MTF values. This is what will notice in a print, even a bit one as it depends on the frequency response of the human eye.

Going back to film. They have MTF curves too, film manufactures publish them. They tend to be dropping off sharply at 100 lp/mm and quite a way down by 50 lp/mm, so the film digital divide is not so great.

for example Kodak E100VS

DavidEB wrote in post #3768614
think about a 24mm T/S. good luck with the cash flow.

I sure with that the 24 was a skoosh wider. I miss my LF camera but not the hassles of using it or it';s expense. But haveing a WA with drop front or tilt would be pertty cool. Something in the 16mm range would be ideal. Horseman makes an adapter for Canon cameras to be used on their l series bodies, but it costs an arm and a leg...

The 14 is rectilinier, but the true parallax correction would have to take place in PS or other software. I have never really been comfortable with that.

Sigh,,, oh well. I got to borrow the mk1 at Udvar Hazy from Canon reps. It was the way to go for many static A/C. I would love to see the MkII on a FF camera. So after I get my 500 I may bite.

Les

Lester Wareham wrote in post #3798883
My estimate for the 1Ds II with Canon recommended sharpening is a reasonable MTF up to about 52 lp/mm.

I can only find this:
As a rough guideline, we usually recommend Unsharp Mask settings of 300%, 0.3 pixel radius and 0 pixel threshold for letter-size prints from EOS Digital SLR images, but please feel free to vary these settings according to your needs and/or personal taste.
from Chuck Westfall on Canon recommended sharpening

Any thoughts on larger or smaller print size 'ideal' starting points for sharpening, and why 0.3 of a pixel how can you sharpen less than one pixel?

Sorry- should this be a new thread?

I Simonius wrote in post #3798991
I can only find this:
As a rough guideline, we usually recommend Unsharp Mask settings of 300%, 0.3 pixel radius and 0 pixel threshold for letter-size prints from EOS Digital SLR images, but please feel free to vary these settings according to your needs and/or personal taste.
from Chuck Westfall on Canon recommended sharpening

Any thoughts on larger or smaller print size 'ideal' starting points for sharpening, and why 0.3 of a pixel how can you sharpen less than one pixel?

I think this is what Canon think is needed to overcome the effect of the AA filter, what we call capture sharpening.

The sharpening for a print size should be done at the output resolution (up or down sized) the amount of sharpening depending on the printing process.

I Simonius wrote in post #3798991
Sorry- should this be a new thread?

Yes probably, and in the right section too.

Hi Lester,

Lester Wareham wrote in post #3798883
Hi Wim

Whilst what you say is true it should need some qualification for example ~69 lp/mm is the Nyquist limit for the 1Ds II and is the mathematical maximum that can be sampled without aliasing.

The actual capability will be less than that depending on the MTF characteristics of the anti-alias filter and MTF characteristics of the sensor (which is not unity from 0 lp/mm to the Nyquist limit as many think). The combination of the two MTFs is known as the sampling aperture for reasons dating back a long way.

Another factor comming into play is how much sharpening is applied. My estimate for the 1Ds II with Canon recommended sharpening is a reasonable MTF up to about 52 lp/mm.

Now the 400 lp/mm for a lens is a bit good, in any event these sort of values are old style 'cut-off' resolutions with probably a 5-10% MTF, so not really of much practical interest.

Lens 50% MTFs, a much more relevant factor, can be much lower even around 30 lp/mm or less for wide angle and ultra wide angle specially in the corners. But even an improvement in lens MTF from say 90% to 95% at 10 lp/mm will have a significant impact on image quality.

My point is that lp/mm is just one dimension and is meaningless without the MTF dimension.

On full frame the most significant resolution range for image quality is 10-30 lp/mm, really fine resolution up to about 40-50 lp/mm, what maters in this range is the overall MTF values. This is what will notice in a print, even a bit one as it depends on the frequency response of the human eye.

Going back to film. They have MTF curves too, film manufactures publish them. They tend to be dropping off sharply at 100 lp/mm and quite a way down by 50 lp/mm, so the film digital divide is not so great.

for example Kodak E100VS

[GIFS ARE NOT RENDERED IN QUOTES]

Thank you for the additional explanation; since the thread wasn't specifically covering this subject, I didn't expand more than I did already.
400 lp/mm is actually more at 2-4% MTF, BTW. And in order to show this in print, you'll need a very high resolution, and slow, B&W orthofilm. In a way it is a bit of a moot point, but the way I was taught, total resolution of a lens-imaging system can be calculated as:
1/system-resolution = 1/lens-resolution + 1/medium-resolution (and medium-resolution obviously being a sensor's or film's resolution)

Now I know this is not entirely applicable to sensors, due to the cut-off caused by the AA-filter, but even so, system-resolution of a dslr is extremely close to the Nyquist frequency, actually higher than you indicate. William Castleman reports 62 lp/mm for the 1Ds II, for example.

Furthermore, what a lot of people do not realize, is that with film for "normal" use, for prints from negatives, the average amateur slr photographer gets about 20-30 lp/mm from his negative, an advanced amateur maybe 30-40 lp/mm, a pro 50-70 lp/mm, and an advanced pro maybe 60-80 lp/mm. this is in 35 mm film format.

So, even at this stage, with regard to dslrs, we are, for amateurs, beyond negative film, and pros are at the same level.

Also, there still is some way to go before we exhaust the possibilities of current lenses, especially primes, and that is a good thing, because the really good ones are close to theoretical levels already. Even so, there still is a 1:5 factor involved, in favour of lenses, and that is for FF. That was what this was all about, really.

Kind regards, Wim

wimg wrote in post #3800541
400 lp/mm is actually more at 2-4% MTF, BTW.

OK, I am just pointing out the lens is not a simple order system, roll-off slope may vary, in fact the characteristic may not be monotonic at those sorts of frequencies.

So a lens might get to MTF 20% by 50 lp/mm and then slowly roll-off to 5-10% at 100lp/mm, so the 10% MTF resolution is not a meanigfull guide to effective sharpness.

wimg wrote in post #3800541
Now I know this is not entirely applicable to sensors, due to the cut-off caused by the AA-filter, but even so, system-resolution of a dslr is extremely close to the Nyquist frequency, actually higher than you indicate. William Castleman reports 62 lp/mm for the 1Ds II, for example.

Again, it depends where you set your cut-off MTF, set it at 10% and you will get right up to the Nyquist limit (the AA filters get down to about 10% MTF it seems). Set it to 30% MTF and you get the sort of number I estimated, 90% MTF is probably down at 20lp/mm.

We are probably boring the pants off everyone, so PM me to chat more on this.

Kind regards, Lester

Nope - highly edificatious!:p

Lester Wareham wrote in post #3804415
We are probably boring the pants off everyone, so PM me to chat more on this.

Kind regards, Lester

I Simonius wrote in post #3804566
Nope - highly edificatious!:p

Aww, your only saying that because youve just come back from the pub. :p

Lester Wareham wrote in post #3804572
Aww, your only saying that because youve just come back from the pub. :p

at this time of day???

(I wish!)

you just having a liquid lunch then?:p

I Simonius wrote in post #3804589
at this time of day???

(I wish!)

you just having a liquid lunch then?:p

No I wish, just squeezing in a bit of PC time before the wife does the shopping on line (and I have to do the washing up). Wonderful life!.

Lester...don't disappear into a whispering gallery with Wim, we all want to hear...eh lads?

Bob

Canon Bob wrote in post #3806045
Lester...don't disappear into a whispering gallery with Wim, we all want to hear...eh lads?

Bob

yearrr! That we do Jim lad ( I mean Bob) that we do!

Now it is time to start the archive.