which lens to buy?

[agorabasta]

OK, the 16-80 is totally fine if it's fine, which is rare.

The OP range has a very good lens that somehow didn't get discussed. It's the Sigma EX 17-50 DC OS HSM.

I have yet to see a bad copy of that lens. And the OS makes it much more usable on a Nex with adapter, even more so if you do video.
In many aspects, it renders just as pleasant as some Tamron 28-75, but it is much better at rendering the mild blur and bokeh, and that mild blur is the most important thing to the general impression of quality in many cases, especially at indoor shooting. The mild blur with that lens is very smooth, natural and symmetric and it shows no excessive CA on the mildly blurred edges, so being natural to the eye, the blurred areas don't attract the attention that may otherwise completely ruin the overall impression from the image.

[David Kilpatrick]

I can only say that something has gone wrong with my 16-80mm, or my A77 has an issue. Checking manual focus, I was not able to focus on infinity last week, and far too many of my shots are not sharp. They don't even begin to compare with Shirley's shots in the same places using the Sigma 18-250mm OS on the A580. This was not the case using the 16-80mm on the A55 in September.

We plan to check the 16-80mm on the A580 and on the A55 again, and check other lenses on the A77, as the focus travel should go through infinity and beyond, not be cut off about 300 feet.

David

[alphaomega]

Agorabasta, I am sure the Sigma 17-50 is a fine lens. Maybe I have not made the point forcefully enough, I will only contemplate a standard zoom that starts at 16mm (24mm) and goes as far as IQ maintenance will allow. For my Sony A mounts the CZ 16-80 is the only one that fulfills this requirement. My lens performs well and provides the image quality I desire. I have never had any focus problems either. I have just traded in my LX3 for LX5 to get the longer reach to 24-90mm in 35mm terms. I would not even look at the quality alternatives for a P&S type that can do raw because they start at 28mm mostly such as the S95.

[artington]

I can only say that something has gone wrong with my 16-80mm, or my A77 has an issue. Checking manual focus, I was not able to focus on infinity last week, and far too many of my shots are not sharp. They don't even begin to compare with Shirley's shots in the same places using the Sigma 18-250mm OS on the A580. This was not the case using the 16-80mm on the A55 in September.

We plan to check the 16-80mm on the A580 and on the A55 again, and check other lenses on the A77, as the focus travel should go through infinity and beyond, not be cut off about 300 feet.

David

The a77 has the ability to adjust the AF function on a lens by lens basis. Have you tried this with your CZ?

[David Kilpatrick]

I don't know the cause, and I aware of the adjustment, but this does not help with infinity collimation - it does not move the physical location of the sensor relative to the mount. I've just ordered a 'white box' 16-50mm f/2.8 SSM removed from a kit, which popped up at a price I was able to haggle down a bit - I was actually buying a Mamiya 120mm f/4 Macro MF for 645 and LCE's spot ad for this Sony lens came up when doing the transaction.

First of all, the inconsistent poor focus error with the 16-80mm makes me have some doubts about my experiences with the 16-50mms I tried at Sony's launch, therefore I'm willing to risk the 16-50mm. Secondly, I'm considering pulling the trigger on selling the Alpha 900+Grip+newSAM28-75 combo, and also on the Alpha 700 and a bunch of lenses - 12-24mm Sigma, 70-200mm HSM non-OS Sigma and converter, 17-35mm KM, 28-75mm KM, 500mm Minolta mirror AF, Minolta RS 28mm f2, Minolta 100mm f/2.8 SF, Sony 70-300mm SSM G, Sigma 70mm f/2.8 macro, and a whole cabinet full of strange items - tilt and shift adaptors, NEX adaptors, many manual lenses, Pentacon six 50mm/Kiev 80mm and so on.

Reason - I've sold my Hasselblad 16mpixel digital kit, but bought a Mamiya 645 AFII with ZD back and 80mm, and just ordered a secondhand MF 35mm f/3.5, a rollfilm back, and the 120mm macro. Reason: for tripod mounted or studio flash work at ISO 50, Adobe Camera Raw will directly interpolate a 28.5 megapixel image from this sensor (which has no AA filter in use) and what I'm seeing from this is well above any DSLR though the A900 comes extremely close if I stick with prime lenses and ISO 100. This outfit returns me to my mid-1990s operating mode with medium format for commercial work, and it's upgradeable if I start to take on such work.

The Alpha 77 will clearly deliver exceptional results but its main benefit will be long reach, high resolution or faster action work. The plan is to obtain a 70-400mm G SSM, keep the 8-16mm Sigma, maybe get a Tamron 60mm f/2 macro (which I very much liked). Then to add a NEX-7 in 2012.

From the early to late 1990s I always worked with medium format for studio and commercial, SLR for tele and macro or specific types of general work, and rangefinder (two CLE bodies and a kit of five lenses) for travel. I had three separate kits each neatly bagged up ready to roll. Now I have a nightmare of bags and store cupboards with far too many choices, I can spend a day just wondering what to take, which bodies, which lenses... intention is to tidy that up by simply removing overlap and redundancy.

I should have the 16-50mm tomorrow or the next day and will be doing some comparisons directly with the 16-80mm and focus distances. I know the 16-80mm is 100% perfect on the A55.

David

[Greg Beetham]

David why not pop a few of those other A-mount lenses you have on your A77 and see if they all go to infinity focus ok, if there are no problems with those then it might be the 16-80 that’s sus, although if it still works fine on the other A-mount cameras there must be another reason, maybe a lens ID foul up of some sort specific to your A77, or specific to that 16-80 on that A77.
Greg

[David Kilpatrick]

If there was an error in registration with the body/sensor distance, it would have shown up on the 8-16mm, and all the shots on that I've checked so far have been good. The reason I have bought the 16-50mm 'split from kit' lens is that I suspect the 16-80mm has developed a problem after four and half years of daily use.

I'm processing images today and have spotted some interesting things - diffraction has a very marked effect at f/11, and most of the unsharp shots have been set at f/11. This softening is absent at f/9, a setting I often use in more variable light. Also, there is a quantum leap in 'something' between ISO 500 and ISO 640 on the A77. I now need to test the 1/3rd steps in ISO setting because I've picked up on some truly strange effects, including completely different colour and tone renderings even when identical development parameters are used.

But the few comparisons I have are not scientific or consistent. Here's the problem: the best A77 shots I have, including those with the CZ 16-80mm, are amongst the best quality images I can get. The worst shots are so poor you wouldn't even rate a point and shoot which produced them (at 100% view). I need to find out how to guarantee, without fail, the optimum result and never get the poor result.

David

[Greg Beetham]

Well I hate to say it but it sounds like some sort of processing/firmware bug, if the hardware (sensor, processor, exposure, AF and SSS) works ok at ISO500 it should still work ok at ISO640…one would think. It might be interesting to find out if it can keep producing the same strange behaviour shot after shot, if it can be discovered exactly what settings obtain the less than ideal results ‘on demand.’ I wonder if it’s a power demand thing, higher ISO should require more power be supplied to the sensor too gain it up....
Greg

[David Kilpatrick]

As I said, the comparison is odd and demands that I test the intermediate ISO steps. Here is the difference between two shots, both taken on auto with auto ISO, one at f/11 (giving a nominal ISO 640) and other at f/9 (apparently needing the same 1/60th but at ISO 500). Technically, the f/11 one is 1/3rd stop less exposed than the other but this was not deliberate and seems to be a product of the variable parameters - EXIF reports different brightness values for example.

What struck me was that 640 at f/11 is a whole big step worse than 500 at f/9 - grainier, softer, and also with colour shifts. The change in the red of the label and the outline round it are inexplicable. To be absolutely sure no camera calibration or other parameters could have affected this, the same raw settings were copied to both files including Agorabasta's 'Deep' camera profile. This is absolutely not a contrast, brightness, sharpening, calibration or other issue. It is an odd shift between very close ISO settings, and very close apertures. I have since determined that f/11 really does introduce softness - even more visible on lower contrast subjects and with higher ISOs. But the change in noise level is marked too.

60thf9iso500.jpg (174.53 KiB) Viewed 4174 times

Above - the ISO 500 example

60thf11iso640.jpg (167.73 KiB) Viewed 4174 times

Then the ISO 640 example. Both are 100% clips. The 500 shot conforms to the sharpness and smoothness, and also the colour and contrast, I need. The 640 shot falls short.

I'm sure that a shift in the position of the red tag may have produced the change in coloure (wind?) but not all the effects.

David

[David Kilpatrick]

I must add something to this - for me, almost none of the landscapes taken at 80mm have worked. At f/7.1, I get pixel-sharp focus. But... I have never been exposed to so much lack of depth of field!

Here is a shot taken at f/7.1 - I might have used f/8 equally well, and I repeated the shot at f/10 after checking images:

f7p1-fullimage.jpg (342.59 KiB) Viewed 4167 times

Here is a 100% clip from the shot showing just what this 80mm shot looks like that wide open:

f7p1-dofexample.jpg (220.93 KiB) Viewed 4167 times

The f/10 is still sharp, which I think f/11 or smaller apertures would not have allowed; but it also has not quite enough depth of field for a full 24 megapixel image, so it has been reduced to a mere 9 megapixel image for my library files - 2400 x 3600 pixels. I am finding that over half my A77 images are ending up reduced to this size, something which never happened with the A900.

David

[agorabasta]

David,

The first example looks like there was some passing beam of stray reflected light that momentarily changed the lighting and also lit the dust definitely hanging in the air of a shop so full of cheap garments. I definitely would reserve any further judgement until we see any controlled lighting samples to compare.

The second example looks badly weird indeed... The only idea that comes to my mind is that the LPF is playing some very bad tricks there. If it's the LPF trick, then there must be more DOF in front of the focus point than there is behind it. That's because some LPF designs may be acting differently upon the converging vs the diverging beams. That would create exactly what we see there. (And then maybe a bit too much of luma NR, too.)

I looked up a sample shot with 28-75 f/2.8 at 75/7.1 with an a300. Below is full downsampled image, and two 100% crops from the image enlarged to 6000x4000px, one showing the area in focus and the other showing the transition area. I think it looks too different from your sample, despite being overcompressed to fit this site limitations.

[Greg Beetham]

David if I were too have a wild stab I’d say the 16-80 is having erratic diaphragm problems and the camera not getting the expected T value is push processing too compensate. It might be interesting to use the 28-75 alongside the 16-80 at a target for a series of shots.
Greg

[David Kilpatrick]

Because of the way the A77 works, exposure is accurate (the main sensor is also the meter). However, I can see a possibility that the EXIF data may be recorded on the basis of the full aperture reading prior to pressing the shutter, and the actual auto ISO gain might be different. This is not exactly what you've said but probably what you mean, Greg. Any Minolta/Sony lens is only repeatable within ISO limits, so 1/3rd of a stop error is within limits. Also, I'm using the electronic first curtain and that may create different noise and gain conditions from using a physical first curtain which briefly blacks out the sensor.

So there are variables present. A recipe for optimum quality with the A77 might include NOT using auto ISO, and NOT using electronic first curtain. Both factors may reduce the risk of unexpected noise levels.

Anyway, I'm still processing my way through a thousand raw files and must stop to do other work then resume at the weekend. I have found so many unexpected things, including a few shots where the foreground and distance are both sharper than the centre - something I've seen using Canon, but never using Sony. It's very possible with zooms like the 16-80mm (it happens with the Canon 17-85mm and 24-105mm lenses). I know that Sony claim a specially flat field for the 16-50mm, and that they claimed a higher degree of plane flatness for the A900 sensor than any full frame sensor made. So they understand these issues.

Could be that the A77 sensor is not perfectly plane, or could be that the higher resolution is revealing field flatness variations in the 16-80mm combined with alignment and focus tolerances.

David

[Greg Beetham]

What I was thinking was, the camera expected the T value of f9 and actually got f16, then pushed the result and called it f11….something like that.
Greg

[David Kilpatrick]

What I was thinking was, the camera expected the T value of f9 and actually got f16, then pushed the result and called it f11….something like that.
Greg

Which it can't do, and the lens is perfectly accurate on other cameras, so I don't think that hypothesis works.

David