Home
SearchSearch MemberlistMemberlist RegisterRegister ProfileProfile Log in to check your private messagesLog in to check your private messages Log inLog in

low_contrast VS high_contrast lenses
View previous topic :: View next topic  


PostPosted: Tue May 27, 2008 6:10 pm    Post subject: low_contrast VS high_contrast lenses Reply with quote

I checked back my files and it seems the contax file I used for the comparison was a modified one so all this theory was based on sand.
Sorry Embarassed


Last edited by poilu on Thu May 29, 2008 1:49 pm; edited 1 time in total


PostPosted: Tue May 27, 2008 6:29 pm    Post subject: Reply with quote

Thank you for this illustration - the contrast differences are clear in the histograms and photos.

So putting your statements together: a low contrast lens, when auto contrast in pp is applied, can give the appearance of a high contrast one, but at the price of lost information.


PostPosted: Tue May 27, 2008 6:33 pm    Post subject: Reply with quote

Thanks for your test


PostPosted: Tue May 27, 2008 6:51 pm    Post subject: Reply with quote

Thanks for that very good point + explanaition!

This is what Zeiss made so famous: high sharpness + high contrast = high acutance. Today it might even way too much for some DSLRs...


PostPosted: Tue May 27, 2008 8:19 pm    Post subject: Re: low_contrast VS high_contrast lenses Reply with quote

poilu wrote:

if we lower the contrast, we compress the histogram and we don't lose info
if we increase contrast we expand the histogram and we lose info as they have to be interpolated


I am reflecting on what you write, and I am not sure.
If increasing contrast means to interpolate to create the missing information, I think that decreasing contrast by compressing the histogram should mean to drop some existing information. Otherwise it would not be possible to push a full image into a smaller portion of the luminance spectrum.
In both cases I think there is a loss in the quality of the image, for different reasons. In the first case, you would have a loss of definition, because you would reach a higher resolution by creation of artificial values that don't add any real data to the existing values. In the second case, you would simply have a loss of resolution, in the form of loss of microcontrast (and therefore of detail, as neighbouring values would become identical by the compression).
I am not an expert, only a user, so I can surely be wrong... so, I ask the experts, am I wrong in this reasoning?


PostPosted: Tue May 27, 2008 8:25 pm    Post subject: Reply with quote

Now that I'm thinking of it - perhaps flare is one of the differences in contrast behavior. I have an old hazy lens Franka which simply refuses to give anything but thin negatives... because the internal flare is so high it destroys the contrast towards the left side of the histogram. I see that sort of behavior in the three examples.

Towards the right, I'm not sure what would cause a drop in the peak with the Jup?


PostPosted: Tue May 27, 2008 8:28 pm    Post subject: Reply with quote

Nesster wrote:
Now that I'm thinking of it - perhaps flare is one of the differences in contrast behavior.


This is certain.


PostPosted: Tue May 27, 2008 8:29 pm    Post subject: Re: low_contrast VS high_contrast lenses Reply with quote

Orio wrote:
poilu wrote:

if we lower the contrast, we compress the histogram and we don't lose info
if we increase contrast we expand the histogram and we lose info as they have to be interpolated


I am reflecting on what you write, and I am not sure.
If increasing contrast means to interpolate to create the missing information, I think that decreasing contrast by compressing the histogram should mean to drop some existing information. Otherwise it would not be possible to push a full image into a smaller portion of the luminance spectrum.
In both cases I think there is a loss in the quality of the image, for different reasons. In the first case, you would have a loss of definition, because you would reach a higher resolution by creation of artificial values that don't add any real data to the existing values. In the second case, you would simply have a loss of resolution, in the form of loss of microcontrast (and therefore of detail, as neighbouring values would become identical by the compression).
I am not an expert, only a user, so I can surely be wrong... so, I ask the experts, am I wrong in this reasoning?


As far as I am concerned, Orio, your thought make perfect sense.
And they concur with my experiences: any change of the histogramm can cause loss (although the image seems to be improved). Thus the best way to achieve a very good photo still is a very good lens and an almost perfect exposure. Wink


PostPosted: Tue May 27, 2008 8:40 pm    Post subject: Re: low_contrast VS high_contrast lenses Reply with quote

poilu wrote:

if we lower the contrast, we compress the histogram and we don't lose info


Approximately correct, provided the number of bits per component is greater than that in the desired output image. (If this operation is done at 8 bits, for example, then you do lose information as dissimilar values map to the same value in the output).

poilu wrote:
if we increase contrast we expand the histogram and we lose info as they have to be interpolated


Both operations are interpolations.

Working from a lower contrast image means there is more headroom and footroom to preserve shadow or hilight detail.

Of course, if the source image is a JPEG then all bets are off (you are starting with 8 bts per component, and shadow detail has been discarded).


PostPosted: Tue May 27, 2008 8:48 pm    Post subject: Reply with quote

The comparison of these three lenses is quite striking. I think Carsten nails it.
Quote:
Thus the best way to achieve a very good photo still is a very good lens and an almost perfect exposure
Thanks for the presentation. Poilu which version of the CCZ 1.4/85 do you have AEG or MMJ?


PostPosted: Tue May 27, 2008 8:56 pm    Post subject: Reply with quote

Very interesting comparison !

And this has changed my thinking, it does not seem that low contrast out of the lens can be completely compensated by post-process.


PostPosted: Tue May 27, 2008 9:10 pm    Post subject: Reply with quote

Orio wrote:
I think that decreasing contrast by compressing the histogram should mean to drop some existing information

if you decrease the contrast of the contax to match a lower contrast you will not loose info and probably gain in gradations (compared to the lower contrast image, of course you loose info compared to the original)
if you reduce a 10mb image to 50% to match a 5mb image you will get better result than enlarge a 5mb image to 200% to match a 10mb image

Nesster wrote:
because the internal flare is so high it destroys the contrast towards the left side of the histogram

the T* of Zeiss certainly help to reduce internal flare

sunshine wrote:
Poilu which version of the CCZ 1.4/85 do you have AEG or MMJ?

I have the MM. I think it is superior to the AE at least for bokeh.

Chris wrote:
Working from a lower contrast image means there is more headroom and footroom to preserve shadow or hilight detail

true in theory but the demonstration is midday in strong light and the sensor record all the light so theory doesn't stand


PostPosted: Tue May 27, 2008 9:49 pm    Post subject: Reply with quote

Nesster wrote:
Now that I'm thinking of it - perhaps flare is one of the differences in contrast behavior. I have an old hazy lens Franka which simply refuses to give anything but thin negatives... because the internal flare is so high it destroys the contrast towards the left side of the histogram. I see that sort of behavior in the three examples.

Towards the right, I'm not sure what would cause a drop in the peak with the Jup?


A very good post and I agree with most of what has been written.
Internal flare is one thing RAW conversion can't sort out and it does destroy lens performance.
A good RAW conversion doesn't destroy information. A bad one with excess increase of exposure, contrast etc can degrade it but one has to be very silly to go that far.
I have just posted a test shot from the Tamron SP 35-80 and have made a comment that may interest some.

I should add there is of course a vast difference between adjusting a scan from a negative (Even if scanned at 16bit) to adjusting a good exposure from a Digital camera RAW file One can adjust the levels and curves on a file from a negative scan but if the information isnt there little can be done without degrading the image. While a digital RAW file can have a wealth of hidden information just waiting for the user to reveal. The way round this with a film scan is to make a few scans and blend them but still one is limited to the information on the negative.
How many true levels does a BW film negative contain? Or are they in fact true levels?


PostPosted: Tue May 27, 2008 11:27 pm    Post subject: Reply with quote

Nesster wrote:
Towards the right, I'm not sure what would cause a drop in the peak with the Jup?

it seems the jup9 is slightly underexposed vs the 2 others (mechanical precision ?)
that could explain the drop in the right


PostPosted: Wed May 28, 2008 7:38 am    Post subject: Reply with quote

Chris wrote:
Working from a lower contrast image means there is more headroom and footroom to preserve shadow or hilight detail


to evaluate the dynamic range one method is equalize who reveal shadows and highlight details
you can see that the Zeiss don't lose highlight or shadows and the micro-contrast on the stone wall is better

jup9-canon-contax


PostPosted: Wed May 28, 2008 2:42 pm    Post subject: Reply with quote

poilu wrote:

you can see that the Zeiss don't lose highlight or shadows and the micro-contrast on the stone wall is better


Yes, and it's exactly what gives that sensation of "air between objects" that I love so much about the Zeiss lenses. Your samples prove it to all evidence. The Canon's and especially Jupiter's images look flat in comparison. Which is bad especially for the Canon because it's a recent AF lens that is supposed to be a top performer. But we're on this forum for a reason... Wink


PostPosted: Wed May 28, 2008 3:12 pm    Post subject: Reply with quote

[quote="poilu"]
Chris wrote:
Working from a lower contrast image means there is more headroom and footroom to preserve shadow or hilight detail


to evaluate the dynamic range one method is equalize who reveal shadows and highlight details
you can see that the Zeiss don't lose highlight or shadows and the micro-contrast on the stone wall is better

Have you tried taking these 3 images and doing a simple 'Match colour' on them?
I just made 3 dub copies and tried it.


PostPosted: Wed May 28, 2008 3:29 pm    Post subject: Reply with quote

Tha high contrast of the zeiss, may can confusion about the sharpness of the three lenses.
Zeiss always expresed that the sharpness is different to the contrast and, with pictures, emphasized that the contrastier image seemed to be the sharper, but didn't it.
I not say that the zeiss isn't contrastier, I put over the table that the image form the jupiter have more detailes in some areas. Look at the red area in the center, all of the things (plants and the wall) present more detail in jupiter's image. I can see more in the jupiter's picture than in the zeiss one in the red area. And then, Because the jupiter is less contrast than zeiss is less sharp too? I don't think so. May be, as in the zeiss illustrations, that jupiter is sharpness but less contratier than the zeiss.
The high contrast of the zeiss impede me to know the real sharpness of that lenses.


PostPosted: Wed May 28, 2008 3:43 pm    Post subject: Reply with quote

Rob wrote:
Have you tried taking these 3 images and doing a simple 'Match colour' on them?

If I use autocolor & sharpening on those samples I can hardly see any difference between the lens.
This post try to find some difference by using histogram or equalize and it is more a try to justify to myself the cost of some of my lenses.
Of course the jupiter is a top lens from F2.8 and the canon is one of the best EF lens
Finding difference is harder on those top performer but as sensor get better I think we will need top lens very soon.


PostPosted: Wed May 28, 2008 3:48 pm    Post subject: Reply with quote

I think such tests of lenses are useful but to prove any theory or judge a lens one has to do ones best to produce an image that Matches whatever other one you are testing it against.
To do the same RAW conversion or a Default one will of course always produce a different result. The difference may be the lens, it may also be a very slight change in the light you did not notice, or could be a silly as the camera shutter speed being different by a few percent.
Digital exposure is very sensitive so shooting in any auto setting could produce a different result. If (More likely) the shots were done in manual, then who says f8 on one lens is exactly the same as the other lens? And again how do you know that light hasnt just changed by 1/6 of a stop?

The present example shows that the 3 example images all have a very slightly different FOV, so are in fact recording 3 different scenes. Such a slight difference is enough to say that the exposure for all three cannot be exactly the same. A look at the three histograms confirms this difference in exposure. IMO it isnt showing anything else.
You dont think that dark land mass top of frame in the first two shots is going to alter any recorded exposure?
I think all the ideas and theories put forward are very good and well worth considering, however it does seem the simplest and most obvious one has been ignored?


PostPosted: Wed May 28, 2008 3:51 pm    Post subject: Reply with quote

Here's a good short article on flare and contrast:
http://www.luminous-landscape.com/columns/sm-03-01-12.shtml

I note that in older lenses, Zuikos had the reputation of using higher contrast to compensate for a slightly lower resolution. How much of that is slander, I'm not sure, but I do get a different feel from Zuikos than e.g. Nikon or Takumar. In modern AF lenses, I feel the Pentax 16-45 does this sort of thing - it hides its lack of resolution with contrast.


PostPosted: Wed May 28, 2008 4:07 pm    Post subject: Reply with quote

Yes, Nesster.
Not only zuiko. LEICA did it too. My personal tests said me that elmars and olds summicrons were better in constrast than in sharpness (never matched the 90 pairs of linesp/mm) while eg the Konicas hexanon 50/1,7 at F/8 matched the 110 pairs of lines in the center - down to 80's in the borders, and had low contrast
If you look to the rendition of the summiluxes at the borders, they are bad in sharpness but so-so in contrast, and the same with the summicrons 50/2 collapsibles (not the rigid that is best).


PostPosted: Wed May 28, 2008 4:12 pm    Post subject: Reply with quote

Rob Leslie
I think that all of us know that the talent is better than any lens. But in the forum we trait this themes to know more of the characteristics of each lens. Play the game. Laughing Laughing


PostPosted: Wed May 28, 2008 4:26 pm    Post subject: Reply with quote

This has been an incredibly informative thread and I think I even understand some of it. I think at least part of the problem is that the sensors involved have relatively limited dynamic range and that is where future developments have to go if they are to keep us moving forward. We need the histograms to stretch further and have fill in both extremes (detail in the shadow and highlights). At that point we will be able to make some fairly definitive comments on "micro" and "macro" contrast, resolution and acutance. The main effect for me of these samples is to demonstrate that mid-day sun offers a dynamic range that is much too wide for any sensor to handle properly - all of these shots have blown highlights and impenetrable shadows. There is a strong color cast to the Jup that conceals some of its strengths. The Zeiss still manages to show its class. The canon makes me wonder where modern development of lens design has really moved forward.


patrickh


PostPosted: Wed May 28, 2008 4:37 pm    Post subject: Reply with quote

estudleon wrote:
Look at the red area in the center, all of the things (plants and the wall) present more detail in jupiter's image. I can see more in the jupiter's picture than in the zeiss one in the red area. And then, Because the jupiter is less contrast than zeiss is less sharp too? I don't think so. May be, as in the zeiss illustrations, that jupiter is sharpness but less contratier than the zeiss.
The high contrast of the zeiss impede me to know the real sharpness of that lenses.


If I understand correctly, you are evaluating from the whole image resizes posted top of thread. It is impossible to make any judgement on resized images. One has to examine the full image size without alterations.