ZS40 Image Evaluations

 


I compared the images from the ZS40 tests to evaluate the relative differences in resolution and IQ at different focal lengths, apertures and ISO settings.


Calculation of Resolution


The resolution is obtained from the coloured Es chart by dividing the image height in pixels by the line thickness (in pixels) of the lowest row of Es in which a distinct E can be seen.  That is, for the ZS40, the resolution in lines per picture height (LPH) = 3672/line thickness.


Resolution and Focal length


For the central coloured Es test image at 24mm (full WA), f/3.3 and 100 ISO, the lowest resolved E is on the 1.5 row, therefore the resolution is 3672/1.5, which is about 2400 LPH. 


For the central coloured Es test image at 720mm (max tele), f/6.4 and 100 ISO, the lowest resolved E is on the 2.0 row, therefore the resolution is 3672/2.0, which is about 1800 LPH.


Resolution and Aperture at the Image Centre


Below is a comparison of test image crops of the central coloured Es chart at different aperture settings at 24mm FL and 100 ISO (click on the image for a larger version):


24mm, 100 ISO Aperture Series


As can be seen, there is a gradual decline in resolution as the f number increases.  This is predicted to arise as a result of the effects of diffraction (as described in this section here).


Because the FZ40 is limited in its widest available apertures (as shown in the graph at the beginning of this section) and because diffraction depends on the f number and not the FL, the resolution decreases as the FL increases.  That explains (or at least explains in part, as there may be other factors associated with the lens itself) why the resolution at 24mm and f/3.3 (2400 LPH) is higher than the resolution at 720mm and f/6.4 (1800 LPH).


Resolution and ISO setting


Below is a comparison of test image crops of the central coloured Es chart at different ISO settings at 24mm FL and f/3.3 (click on the image for a larger version):



24mm, f/3.3 ISO Series Es Chart


While there is only a slight reduction in resolution and increase in noise from 100 to 400 ISO, above 400 ISO the noise increases at a faster rate, and it is accompanied by a gradual loss of resolution, an increasing amount of smearing of the Es, and an increasing loss of some colours, particularly cyan and yellow.  It is obvious that the best resolution and lowest noise is at 100 ISO.


Below is a comparison of test image crops of part of the circuit board and yellow fabric at different ISO settings at 24mm FL and f/3.3 (click on the image for a larger version):



24mm, f/3.3 ISO Series Crops


The crops above of objects at a larger scale give a visual appreciation of the gradual decrease in IQ as the ISO setting increases.



Resolution in the Image Corners


The crop below shows a 100% crop from the top left hand corner of the test image taken at 24mm, f3.3,and 100 ISO:




As can be seen, the corner of the image shows considerable blurring and distortion.  All of the corners of the image show this effect, although there are slight differences in the degree of distortion.


Examination of the corners of test images taken at longer focal lengths show much less blurring and distortion.  This indicates that the effect is caused by barrel distortion at WA which is not completely corrected by the in-camera JPG processing.  This is discussed further below, in the section about RAW with the ZS40.


It can be concluded from the above results that the maximum resolution is obtained with the ZS40 at or near the image centre, at low FL, widest aperture, and 100 ISO setting.



RAW with the ZS40


The ZS40 differs from the SX700 (and all other current travel zooms) in having RAW capability (RAW being the raw data from the camera sensor before any processing).  It is possible with the ZS40 to take a fine quality JPG concurrently with a RAW file, and all the tests in this section were conducted in this way.  RAW files from all of the tests carried out at 24mm and 720mm on the ZS40 can be downloaded from the ZS40 test images page above.


The free program Silkypix Developer Studio 4.1 SE can be downloaded (here) for developing images from the ZS40 RAW files.  The free open source program Rawtherapee (here) and several other commercial RAW developers are available which eventually should also support ZS40 RAW files.


Processing the RAW file offers the ability to recover blown highlights better than from the jPG, and also allows the WB to be adjusted as well as various other corrections and adjustments.


For the current tests I used Silkypix 4.1 SE at its default settings to process the 24mm, f3.3,100 ISO RAW test image and compared it with the accompanying JPG.  Below is a comparison of crops from the central Es chart (click on the image for a larger view):



Silkypix processed RAW JPG vs OOC JPG, 24mm, f/3.3, 100 ISO


At this magnification (300%) the Silkypix processed RAW image shows some faint noise in the white background but no sharpening haloes, whereas while the noise is not visible in the OOC JPG there are distinct sharpening haloes present.  The JPG processing has however made the Es more distinct for most colours, except for cyan and yellow, which appear somewhat more distinct in the processed RAW image. In the case of the black Es on the 1.7 and 1.8 rows, the centre bars are separated from the uprights in the JPG but not in the processed RAW image.  Obviously further rprocessing could be performed on the RAW JPG to improve it.


Below is a comparison of crops from the central Es chart for the JPG and Silkypix (default) processed RAW file for the 24mm, f/3.3 800 ISO test image (click on the image for a larger view):



Silkypix processed RAW JPG vs OOC JPG, 24mm, f/3.3, 800 ISO


In this case it is obvious that the OOC JPG processing has involved some noise reduction and the tradeoff has been some distortion and smearing of the Es, as is particularly noticable in the black Es in the lower rows.  The processed RAW image does not show the distortions or smearing but there is significantly more background noise.  It might well be possible with suitable RAW processing adjustments and some noise reduction to achieve a result that is better than the OOC JPG in this case.



Uncorrected RAW Processing


Current RAW processors usualy include automatic corrections for various lens aberrations, vignetting etc for a particular camera model.  I tried processing a RAW file from the ZS40 using the RAW developer called PhotoNinja (here) and found that although it did not yet have support for the ZS40 it could open the RAW file.  I processed the 24mm, f3.3,100 ISO test RAW file using only the demosaicing step, with no other corrections.  I was quite surprised at the result, as shown below (click on the image for the full sized view):



24mm, f3.3,100 ISO RAW file, demosaiced only, using PhotoNinja


Quite severe barrel distortion can be seen and the image is considerably wider than the 24mm OOC JPG, which is shown below (click on the image for the full sized view):



The corresponding 24mm, f3.3,100 ISO, OOC JPG


It is obvious that a large amount of distortion correction and cropping has taken place in-camera to produce the OOC JPG.  That would explain the significant amount of residual blurring and distortion that was noted above at the corners of the 24mm JPG images from the ZS40.  Actually it is quite surprising that the  result is as good as it is, when one is aware of the amount of correction that is involved.


Note:  I have since conducted further testing of the ZS40 resolution at focal lengths of 50mm and 250mm (equiv.) as shown on the ZS40 Resolution Update page as linked below.




Continue to ZS40 Resolution Update


Continue to SX700 Image Evaluations


Continue to Comparisons between ZS40 and SX700


Return to Previous Page


Return to Home Page