Sony A77 Image IQ at iso 6400 comp to A580

Started Oct 14, 2011 | Discussions thread
Flat view
Ma55l Contributing Member • Posts: 805
Sony A77 Image IQ at iso 6400 comp to A580

In the past I have shown direct side by side comparisons of the a77 sensor at moderate iso (1,000 to 1600) against several other cameras/sensor combinations from several different manufacturers as well as against earlier products from Sony. Those can be viewed on this forum in some of my other threads under Sony a77. My contention is that although the noise is low on the a77, the capturing of fine detail is severely lacking at elevated iso primarily because detail is traded for noise. The pictures below will try to prove that.

I thought I would place a side by side direct comparison image at iso 6400 for you to examination showing the a77 image in the lower section and the a580 image - one above the other.

This comparative image is a JPEG but is representative of what the cameras are capable of at iso 6400 when compared on the same platform and same subject matter.

If you will notice the objects in the images are the same size since I downsized the a77 image slightly to be the same size as the a580 representation. Please click on the image before making conclusions. You can go to the IR site to see the full images for both cameras to check them directly.

I respect your right to have the last word in your interpretation and whether the a77 is for you. Each person has different needs and concerns and cares. My point in this post is just to give you some representation of possible negatives that you might encounter that you might have not otherwise considered or recognized. Remember click on the image.

However, in my view, I feel that the IQ of the a580’s representation of the scene is a lot better.

1. The noise chart on the a77’s representation shows colors bleeding out of the box constraints that are representative of each color patch to a greater extent than on the a580. The a77 just doesn't look as good. This is related to loss of fine detail as the sensor is not able to determine if there is a color present or the gray border - and so blends the image – which destroys detail.

2. Saturation in the red, brown, gold and green is anemic in the a77 sensor. If the processing engine would turn that function up the bleeding of colors would worsen and so would the noise.

3. If you look at the image of the label on the bottle on the left and compare to the a580 you can see how much loss of saturation is affecting the quality of the image on the a77. And the a580 seems to be sharper - with 8 MP less.

4. Looking at the color chart you can also see how much more noise there is within most of the color patch on the a77 image.

5. Look at the label on the bottle, at the right upper border of the chart, and see the distortions in the background of the label which are not present in the 580’s image. The delicate rendition of the label’s background detail is obliterated and then redrawn by the processing engine but incorrectly. It even looks like the label was removed, crumpled and then reglued to the bottle. Therefore, I classify that as an image distortion as it is not what is actually there in reality. Look at the printing on the label on the lower two lines and see how the printing seems hand drawn and poorly done at that on the a77. Note how much better the 16 MP APS a580’s sensor does in comparison to the 24 MP sensor at this high iso and that with 8 million fewer pixels. The difference is that the pixels on the a580 sensor are simply larger and able to capture more light.

6. Then look at the brown bottle above the chart in each view for comparison. The brown on the a77 image is disorganized and irregular without a smooth texture.

I am aware that iso is an extreme example but that is where quality differences reside. Any camera can render a fine 3.5 x 5 print or good prints at low iso. If you wanted to buy a Corvette you would take it for a spin and then floor the accelerator while timing how long it takes to go 0 to 60. If there is a 1 second slower acceleration to 60MPH than expected, then the car is not right. Of course, there would be other checks that you would do as well.

If you wanted to know what is meant by the IQ of an image, visually compare the label on the left side of the image between the two cameras – one image is decidedly better to me.

IQ is a composite of several functions but the end result is the quality of the image. With good IQ you get a well exposed, properly saturated, and easy for your mind's eye to see and understand images. Some people say that with high IQ there is almost a 3D effect to the image and that the image looks real. That can be seen in the a580 representation above the a77’s. In this instance the A77’s IQ is low when anything is done to tax it’s ability. My feeling has been that the pixels are too small to be effective at low light as they just can’t capture enough signal.

To put things in perspective, the pixels for the 24 MP APS sensor are about 3 to 4 microns and that is the size of a germ. Ecoli is two microns and a red blood cell is 7-8 microns. A light microscope can see the bacteria but only render its outline as light microscopes can only magnify 1,000 times, showing that there are limits to the properties of light.

One last comment: I have seen discussions that taking 4 small 16 MP sensors from point and shoot camera and blending them into a 64 mp sensor would yield a better image than any APS sensor with less effective total real estate and smaller pixels. There are at least two objections to that. First is at how high or low an iso, but, the second is even more revealing. Point and shoot camera have short focal distances. Since the focal length of PS cameras are so much shorter than APS, there is less distance for the light to disperse in traveling to the sensor.

Flat view
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow