[clackers]

So Sony tricks its customers into believing they are getting a 16-50 mm lens, when in fact the distortion and vignetting are so horrendous that corrections have to be applied that push the corners out of the image, turning it into something like an 18-50 mm lens. This is pure deceit.

I have that lens, MFF, and verily, it sucketh big time. :-(

[Kunzite]

Some of their lenses are made by Zeiss. Some, by Dyson?

[clackers]

They've got the Zeiss name on them, true! :-D

[Quicksand]

I have that lens, MFF, and verily, it sucketh big time. :-(

The highest compliment I ever heard about that lens is "really, it's not so bad!" And it's $300.

[rawr]

And it's $300.

I paid AUD $320. But for that price it came with a camera [A5000] attached .
Lenses like the 16-50 PZ still sell, most average users won't notice that they have a lens that, technically, sucks.

[MrNPhoto]

Because it slows down processing and thus overall operational speed. The K3 already offers automatic correction for vignetting, aberrations and distortion, but if you switch all these corrections on the camera takes significantly longer to save the images you take.

CPU power in digital cameras will improve, algorithms will get better.

The simple fact is that digital image capture allows camera manufacturers to alter lens characteristics within the image processing pipeline. This technique can make good lenses great, or as in your example, try to extend the operational range of a lens beyond what is useful. In-camera HDR, dynamic range compensation, all manner of exposure and white balance adjustments, noise reduction, sharpening, digital image stabilization - all processing tricks applied post-capture to improve our images. Lens corrections in post-processing are no different, and I expect they will become more commonplace as CPU power improves.

[derekkite]

CPU power in digital cameras will improve, algorithms will get better.

The simple fact is that digital image capture allows camera manufacturers to alter lens characteristics within the image processing pipeline. This technique can make good lenses great, or as in your example, try to extend the operational range of a lens beyond what is useful. In-camera HDR, dynamic range compensation, all manner of exposure and white balance adjustments, noise reduction, sharpening, digital image stabilization - all processing tricks applied post-capture to improve our images. Lens corrections in post-processing are no different, and I expect they will become more commonplace as CPU power improves.

The problem comes when the correction required subtracts from the image quality. If I'm taking a shot at the limit of dynamic range or low light, any correction would narrow the band that I can work with. And frankly, I might as well take the shot with my phone.

I very seldom have 5ev to give to vignette correction.

[rawr]

The problem comes when the correction required subtracts from the image quality.

Additionally, I think another problem with in-camera lens corrections is that they rely on correction algorithms derived from a perfect factory lens sample.

But as we all know, not all lenses come out of the factory the same. Some will be de-centred, some will vignette more than others, etc. So the in-camera corrrections often won't be able to work as designed, since they assume a perfect lens sample, and so may make things even more flawed.

[Nicolas06]

CPU power in digital cameras will improve, algorithms will get better.

The simple fact is that digital image capture allows camera manufacturers to alter lens characteristics within the image processing pipeline. This technique can make good lenses great, or as in your example, try to extend the operational range of a lens beyond what is useful. In-camera HDR, dynamic range compensation, all manner of exposure and white balance adjustments, noise reduction, sharpening, digital image stabilization - all processing tricks applied post-capture to improve our images. Lens corrections in post-processing are no different, and I expect they will become more commonplace as CPU power improves.

Algorithms are already fine and great and can't invent what is not there.

Progress has slowed down a lot on CPU power, if you want more computing power you also often require more juice and that mean less autonomy or bigger batteries. But if you accept to do it at home, you can use the much more powerfull computer do it and let him work for a long time is quite acceptable.

[Cynog Ap Brychan]

The problem with lens correction, either in-camera or at post processing, is that there are always trade-offs. Distortion correction effectively crops the picture, while vignetting correction creates noise. The better the lens, the less these factors come into play.

[Rondec]

CPU power in digital cameras will improve, algorithms will get better.

The simple fact is that digital image capture allows camera manufacturers to alter lens characteristics within the image processing pipeline. This technique can make good lenses great, or as in your example, try to extend the operational range of a lens beyond what is useful. In-camera HDR, dynamic range compensation, all manner of exposure and white balance adjustments, noise reduction, sharpening, digital image stabilization - all processing tricks applied post-capture to improve our images. Lens corrections in post-processing are no different, and I expect they will become more commonplace as CPU power improves.

I would like lenses engineered the best they can be before I decide whether or not to apply a correction feature. Corrections are generally applied to jpegs, not RAW photos when used in camera, they do slow processing speed meaning that your buffer fills faster, and there is no free lunch. Bad vignetting can be fixed if the image is shot at low iso and has dynamic range to spare, but if you are shooting at iso 800 or 1600, it probably can't be fixed without introducing a bunch of noise in the corners. Fixing distortion on a wide angle lens takes away some of the width of the image.

I think Sony, in particular, under engineers their lenses. Maybe it is to save cost for the lenses, more likely it is to try to have smaller lenses to go with their smaller cameras. Either way, I don't think using a computer to fix stuff that just should have been made right in the beginning is a really good way to go.

[MrNPhoto]

The problem with lens correction, either in-camera or at post processing, is that there are always trade-offs. Distortion correction effectively crops the picture, while vignetting correction creates noise. The better the lens, the less these factors come into play.

There are always tradeoffs at every stage of design, nothing new there. Of course a better lens will have better performance characteristics, and require fewer corrections. It's just that it's now possible for manufacturers to design a lens to meet their needs (for both performance and cost), and save money by using post-processing algorithms to apply corrections, instead of physically redesigning the lens. And it is possible (with firmware updates) to 'upgrade' the performance of a camera-lens combo over time.

[derekkite]

There are always tradeoffs at every stage of design, nothing new there. Of course a better lens will have better performance characteristics, and require fewer corrections. It's just that it's now possible for manufacturers to design a lens to meet their needs (for both performance and cost), and save money by using post-processing algorithms to apply corrections, instead of physically redesigning the lens. And it is possible (with firmware updates) to 'upgrade' the performance of a camera-lens combo over time.

Sure. I already have one of those in my phone.

Nothing can create what the lens and sensor didn't record. Period. Shots I take with my phone don't matter too much and have lots of leeway, so it can do whatever it wants with it. But when I shoot in the conditions that I face, I spend lots of money to get a stop of light, to get a sharp lens to record as much detail with as little noise and as much dynamic range as possible.

This same conversation happened with anti-aliasing filters which purposely introduced a slight softness to prevent moire. A sharpening algorithm can make it look sharp. Except the detail isn't there. With higher resolution aa free sensors I am capturing detail that I couldn't with the K5.

This is why I don't like the mirrorless short registration systems. To get a very good lens without these shenanigans you end up with something larger and heavier than a dslr with longer registration distances.

[Winder]

I think Sony, in particular, under engineers their lenses. Maybe it is to save cost for the lenses, more likely it is to try to have smaller lenses to go with their smaller cameras. Either way, I don't think using a computer to fix stuff that just should have been made right in the beginning is a really good way to go.

Sony does both. The 28mm F/2 FE needs a good bit of correction, but its small and compact with fast AF. The 90mm FE macro on the other had is large and needs no correction. The Sony 35mm F/1.4 is also a large lens that needs no software based correction. I think when Sony is done, you will see the Sony G series lenses will be large, expensive and relatively distortion free while the regular Sony glass will be compact and depend on software based correction. The Zeiss Batis lenses use a good bit of software based correction, but still produce really good IQ. I have the 90mm FE, and I'm waiting on the Sony 85mm F/1.4 G before I decide between the Batis or the G version.

LensRentals.com - Sony FE 90mm f/2.8 G OSS Resolution Test
"Please don't take this out of context. These are simply Imatest MTF50 numbers. By that standard the new Sony 90mm f/2.8 G OSS Macro seems to be a superb addition to the E mount lineup, and the lens on an A7r should provide superb resolution. We'll have to look at more in-depth, hands on reviews to see how it actually performs in the field, what the bokeh looks like, how well the OSS performs and a host of other factors to decide how great (or not) the lens performs in the real world. But these preliminary results look very good."

[pathdoc]

I think Sony, in particular, under engineers their lenses. Maybe it is to save cost for the lenses, more likely it is to try to have smaller lenses to go with their smaller cameras. Either way, I don't think using a computer to fix stuff that just should have been made right in the beginning is a really good way to go.

I think it reflects a problem with their mindset. They are excellent at creating technology, not so good at building cameras. Their emphasis is on image capture, not system building, and they haven't spent the last sixty or seventy years doing it the way Nikon, Canon, Pentax, etc. did. I handled one of their mirrorless cameras yesterday, and it seemed that the design aim was to determine how small an image-capture device they could graft to the back of a lens. As far as I could tell, the "Spotmatic D", K1000-D or whatever steampunk-retro digital version of an old film camera you want is eminently possible; Sony could package the guts of it for you now and stuff it in the back of the old film body to boot. But the one thing they could never do is create the feel of that old body which made you want the digital version in the first place.

IMO Sony should get out of making cameras and concentrate on providing core components to all comers.