Comparison of unsharpened resolution Sony IDC raw developer vs Iridium Digital
In the process of my ongoing obsession about comparing pixel quality between the Sony A7r and the Sigma Merrill sensors I found this interesting, at least to me, result.
The idea is to compare pixels without any additional processing to the raw data. Most vendor products appear to massage the data for their products so as to produce the best default image. Sigma for example sets 0 sharpening to what is actually +2 and the only way to get actual 0 sharpening is to set the sigma value in SPP to -2.
I was curious as to whether Sony did this as well. The answer it turns out is no, the 0 value in the sony slider actually represents 0 sharpening. I determined this by comparing it to an unsharpened and no noise suppression image for Iridium digital. Aftermarket raw developers are probably a more reliable way to get unmodified raw from a camera due if nothing else simply to their need to build a toolset for many cameras rather than just one or one brand.
What I found was that Iridium is significantly better at getting raw details from an A7r image than is Sony with its own raw devloper IDC.
The tests were done using Koren's charts and imageJ following the techniques described here: http://www.normankoren.com/Tutorials/MTF5.html
Note this page was intended for doing mtf tests. I am not doing mtf tests I'm simply profile plotting the results from the same raw file using different developers. The raw file contains an image of Koren's chart. Once the image is developed I used imageJ to plot a profile of the reproduction of the chart in the two dimensions of gray tone vertical and pixels borizontally. Each set of a max and min amplitude on the graph indicates a resolved line. The more lines that are resolved the better the raw developer is at finding the detail in the image. As it turns out Iridient is distinctly better than IDC.
So let's have a look:
Comparing Iridient results at the top to IDC both with 0 sharpening it can be observed that just before the 100 pixel mark on the horizontal the IDC developer is already finding less distinctive boundaries between shades of gray (amplitude of a single set of high and low values). Beginning at 100 Iridient has a few more sets of hi-lo pairs indicating it is finding more distinct lines indicating higher resolution. From 150 iridient continues to find more discreet lines up to about 180 while IDC has become much less distinct and pretty much fails after about 175.
From about 180-210 resolution has become largely undetectable by both converters. At about 210 for Iridient and 225 for IDC both start to appear to pick up distinct lines again. I don't believe this to be actual resolution from the chart but rather high frequency aliasing which again Iridient seems to be better at detecting.
The final profile plot shows what happens when sharpening is set to -100 on the sony IDC software. I suppose the - settings are there to provide blur for some reason. I was curious as to whether Sony was doing something similar to Sigma where -2 sharpening in SPP is actually 0 (also confirmed using this technique on a sigma image)
If anyone is interested I will upload the raw image and would like to see similar results from other converters particularly the high end converters. If you will upload full quality jpegs I can do the same ImageJ analysis or I can post my process for others to try.
I think this way of looking at raw developers is useful for determining the comparative quality of decoding the raw image only. Other processing functionality is not being tested and cannot be determined using this raw image.
However the tool can be used to check any techniques and/or impacts of various manipulations of raw converters on the original raw image.
The original image was taken at iso100 at f5.6 at 18mm on a sigma canon mount 18-35 f1,8 lens on a sony A7r in aps-c mode using a metabones auto adapter mk iii .
Here is the original image:
|Post (hide subjects)||Posted by||When|
|May 11, 2014||1|
|May 11, 2014|
|May 11, 2014|
|May 11, 2014|
|May 11, 2014|
|May 15, 2014|
|May 25, 2014|
- Canon EOS M58.8%
- Panasonic G85/G803.3%
- Panasonic FZ2500/FZ20001.9%
- Panasonic LX10/LX151.2%
- Panasonic GH5 development3.6%
- Sony a99 II15.9%
- Nikon KeyMission 170 and 801.0%
- Fujifilm GFX 50S development28.3%
- Olympus E-M1 II development18.7%
- Olympus E-PL80.1%
- Olympus 25mm F1.2 Pro1.5%
- Olympus 12-100mm F4 IS Pro1.9%
- Olympus 30mm F3.5 Macro0.1%
- Sigma 85mm F1.4 Art3.6%
- Sigma 12-24mm F4 Art2.6%
- Sigma 500mm F4 DG OS HSM Sport2.4%
- YI M12.2%
- GoPro Hero50.8%
- GoPro Karma drone2.2%
|Sadiqur_Rahman by Sadiqur Rahman|
from Ain't Going to Work on Maggie's Farm no More
|Airborne by John Beavin|
from - How to respect the Flag and Anthem - (Portrait in Full Colours + A Border)