Sony a77 sensor disappointing - revisited - do we need new testing parameters Locked

Started Jul 28, 2012 | Discussions thread
Flat view
This thread is locked.
Ma55l Contributing Member • Posts: 805
Sony a77 sensor disappointing - revisited - do we need new testing parameters

Sony a77 Sensor Disappointing - revisited - do we need ssitional testing parameters

In August of 2011 I started a thread about the disappointment I had about the A77 sensor. There were a lot of negative comments about my feeling but I showed with comparative photos why the images were poor and disappointing at high iso. I believe you can go back and check those older posts and the many comparative photos in the original thread. At higher isos the excessive processing to handle the noise was distorting the image in a terrible way.

So now comes the August 2012 “ConsumerReports” magazine. They rated 11 SLR like cameras (the a77 is part of that list) as well as 14 of the current DSLR cameras. There were a total of 25 cameras ranging from a low price of $380 for the Lumix DMC-GF2K to a highest price of $1,850 for the Canon EOS 7D. The Sony a77 is priced at $1400 for comparison. Of the 25 cameras evaluated by “ConsumerReports” only two cost more than the a77.

One of the items evaluated in their work was image quality for each camera. If you look at the magazine you will see that the two Sony’s with the same sensor (a77 and A65) were the lowest rated cameras in the entire group for image quality.

My feeling expressed in the original thread was that there were too many small pixels on that small sensor to easily capture light. Therefore excessive obligatory amplification was needed and the extensive processing required to control amplifier noise was causing the problems. It is interesting to note that every other camera in the article (23 of them), ALL had lower megapixel ratings than the Sonys and still were rated with better image quality. Also notable is that the Sony a55 (15 MP) was the at the top of the list for SLR like cameras. It seems as if my assertion that the pixels were too small on the a77 sensor for the best quality image with today’s current amplification technology. I also mentioned that I thought that 15 to 18 MP was the ideal for a 24 x 14 mm. sensor at high iso.

It is also interesting that the D800, which I have purchased, has a 36 MP sensor and is wonderful. But that is a full frame sensor and is more than twice the surface area at 36 by 24 mm. than a 24 x 15mm. plot. Cropping the surface area of the D800’s sensor to the same size as an APS-C sensor puts the pixel count at 15 MP. That means that the surface area of each photosite in the D800 is significantly larger than those on the a77 sensor and so can capture more light thereby needing less amplification at low light levels. Again I feel the best size for today’s noise reduction algorhythms as well as the physical properties of light for an APS C senor is 15-18MP and full frame at 36 to 40MP.

Be aware that “ConsumerReports” magazine receives no advertising support from Sony or any of the other cameras manufacturers as do many other magazines and internet sites. Therefore CR magazine is an excellent source for evaluation of these cameras as they do not step on anybodies toes when they speak their mind. In regard to this site it is duly noted by DPreview that there were problems with the a77 sensor at low light and I believe that the testing done here was perfectly accurate as far as lines of resolution and noise etc. But the image degradation by the distortions caused by processing very weak light levels was only hinted at in a snippet of that review .

And for those reason I believe that some time and effort should be assigned during the extensive lab reviews done at DP review, to look not just at resolution and color accuracy etc., but curvilinear distortions and smearing caused by the processing engines at high isos. Otherwise we will be misled. Some cameras at high iso give stunning images – in my opinion the a77 does not, in spite of similar resolution testing results compared to others. In the case of the a77, the images at high iso looked like the mirage effect that you see on a hot highway surface or desert. And there was smearing of one boundary into another. At high iso in the a77 straight lines become wavy, and detail as well as IQ are lost. And then there are the tell tale signs of over sharpening with its white rimming at dark-light borders trying to compensate for those losses - that just doesn’t happen in my Nikon D800.

One last point here, is that I believe that the manufacturers know how to make resolution testing figures look better by their processors. Those figures sell cameras. So when the camera senses line pairs, the image processor draws the final image enhancing the line pair numbers artificially. Then cameras with the same resolution numbers come up with very different fine detail and IQ ratings. Fine detail and IQ are more related to two dimensional properties i.e. sharpness, rather than one dimensional properties i.e. resolution. So the a77 Sony can test well but still have poor IQ for its images at high ios – and that is likely what “ConsumerReports” had noted.

And that is why DPreview should at least investigate whether curvilear distortion along with smearing and oversharpening evaluations should be added to their reviews even if it is just a subjective rating of good, fair, or poor.

Flat view
Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow