Erik's web

  • Increase font size
  • Default font size
  • Decrease font size

Observations on Leica S2 raw images

E-mail Print PDF

Note: This article is based on two images kindly made available by Mr. Lloyd Chambers at http://www.diglloyd.com, these images are part of a long article on the Leica S2.

I very much recommend the original review which is: here (subscription to DAP required)

 

The Leica S2

Leica has reinvented a new kind of camera, a DSLR body with a larger size sensor. It's essentially a Medium Format DSLR. The new Leica of course stirs significant interest in the photographic community. The original invention was made by Mamiya, in the form of the original ZD. That said, the Leica has a new format, larger than full frame 135 but smaller than full frame MF and the whole system is optimized for this size

 

This camera is in principle interesting for users of advanced professional level DSLRs as a possible upgrade to achieve the next level of image quality. The camera is also interesting in the sense that it can be assumed that it's lenses are beyond reproach.

The camera also does use a mix of MF and DSLR technology. The sensor is a Kodak CCD sensor typically used in MF cameras and does not have an Optical Low Pass filter, but untypically for MF it has micro-lenses.

 

I have some special interest in the OLP filtering. According to imaging theory an OLP filter is essential if the lenses have significant MTF ant the pixel pitch, also called the Nyquist limit, aliasing artifacts will arise and they cannot really be removed in post-processing. It seems that most photographers only objects to color Moirés which are a special case of aliasing effect related to the Bayer pattern sensor used in nearly all digital cameras.

 

Lloyd Chambers has published an excellent review of the Leica S2 on his DAP (Digital Advanced Photogrpahy) site. The review also has a very carefully executed comparison shot between the S2 and the Nikon D3X which by many is regarded as the leading DSLR with regard to image quality. I warmly recommend the DAP site for anyone seriously interested in the Leica S2 or seriously interested in achieving optimal image quality at all.

 

Method of the review

Obviously the images are different. To begin with they are different size. I normally do my comparison at print resolution. In this case I choose to use a presumed print size of 50x70 cm at a resolution of 360 PPI.
There is some rationale of that choice, 50x70 cm is a reasonable print size. The resolution choosen is a bit high, but I could clearly see line patterns in Lambda prints, siized 70x100 cm at 200 PPI.  So  I felt that going to 360 PPI may be somewhat motivated. In addition I wanted to have a situtaion where both images were scaled in the same direction, so we wouldn't scale the S2 image down and the Nikon image up.
Sharpening is very important to the visual impact of a an image and so is tonality. Sharpening is also highly subjective.  In general an OLP-filtered image needs much more sharpening than non filtered images. In this review sharpening was chosen to achieve similar edge contrast. Tonality was adjusted to be similar.

These steps were used:
  1. The images were slightly adjusted in LR3
  2. Sharpening and noise reduction were set to zero
  3. Both images were opened in Photoshop CS5
  4. Image->Adjustment->Match color was done to the Nikon image based on selection around "Faith figure" in the Leica S2 image
  5. Image->Size was upscaled to 50x7x cm at 360 PPI using Bicubic
  6. Both images were sharpened by Filters->Smart Sharpen->Gaussian Blur (Setting Leica S2 Radius 1 amount 150, Nikon Radius 1.5, amount 3.11, advanced mode on both)
So what do I see?

First, let's examine the central figure "Faith".

On the left we Leica S2 and on the right Nikon D3X. The Leica S2 has obviously a significant advantage in detail but edge sharpness is quite similar (because of the Nikon image been more agressively sharpende).. Now, let's look at the crucifix "Faith" has in her hands. On the Nikon image it has a pretty uniform color while on the Leica image it has some color pattern, probably an interaction between the bayer pattern of the sensor and the pattern between the mosaic.

 

Klick on the image for actual pixels view, please!

 

Some more resolution and moiré

In the part shown below we have a stucco line pattern which on the Leica image is quite close to the Nyquist limit.
On the Leica S2 the lines are clearly resolved and on the Nikon image they cannot be seen. This is mainly an effect of the fact that the Leica has something like 24% better linear resolution, but probably also on the OLP filtering.
The Leica image has a typical colorful Moiré caused by interaction between the repetitive  pattern in the image and the color grid array in front of the sensor.
Klick on the image for actual pixels view, please!

 

A more obvious example of Moiré artifacts

The moiré effect we looked at until now has been more marginal. We have some quite heavy Moirés in this image as the image contains a lot of fine structures in form of stuccos on the wall.
These structures happen to be near the Nyquist limit on the Leica S2 image and handily illustrate OLP-less design at it worst.
Klick on the image for actual pixels view, please!

In this case the feature size in the stuccos is very close to wtice the pixel pitch on the S2 sensor, so we get aliasing. Because of the "bayer pattern" of the sensor the resulting moiré pattern is colorful. According to signal processing theory such patterns are virtually impossible to remove in postprocessing. In practice the perception of colorful artifacts is much disturbing while monocrhrome acrtifacts take some experience to detect. So many programs can reduce aliasing by reducing saturation on edges.

 

Update: A reader pointed out that the Nikon images has some ugly red edges. I don't know what this is, it could be lateral chromatic aberration or some effect of the sensor. Lateral chroma can be corrected in raw conversion, but I have not tried to do that in this case.

 

Looking at the corner of the image

The top left corner doesn't show visible color aliasing.
Klick on the image for actual pixels view, please!

 

The Dynamic Range issue

There are a lot of claims that MF equipment would have significantly more DR than smaller format equipment. There seems to be little support for this in imaging theory. If his was correct we would be able to extract much more shadow detail from the Leica image than from the Nikon image, assuming that both were reasonably exposed to the right. Also, would MFDBs offer a large DR they would perform very well at high ISO if ETTR (Exposure To The Right) was correctly executed.
Below is a screen dump from Lightroom 3 where both images had an exposure of +2.95 and fill light of 35 was added.

In my view it is quite obvious from this sample that we don't get significantly more shadow detail from the Leica S2 than from the Nikon, but also that the Leica image is more noisy.

 

Conclusions

The Leica image has significantly higher resolution than the Nikon D3X image.  The difference in sharpness due to the fact that the S2 has more pixels. The resolution is also affected by the Nikon having OLP-filtering while the Leica S2 has not. Finally the Leica lens in this test is said to be truly excellent.
The omission of OLP-filtering increases the risk of Moiré and other aliasing artifacts significantly and such artifacts are clearly present in this image.
Last Updated on Saturday, 04 February 2012 10:02  

Statistics

OS : Linux u
PHP : 5.2.9
MySQL : 5.0.67
Time : 11:32
Caching : Enabled
GZIP : Disabled
Members : 1510
Content : 72
Web Links : 1
Content View Hits : 331541

Login

Banner1

J!Analytics