which lens to buy?

[agorabasta]

I have a specific problem, called Alamy. I submit images which are all examined at 100% pixel size.

Well, as the digital res goes up, it must at some moment decouple from the real effective res. So the problem you face now must become real later, when APS sensors reach 50+ Mp.
But the problem is that 24Mp is not even close to that threshold at f/8, while being still very far from the edge even at f/11.

So for now, you just have to know which particular lenses work with a77 for your purposes within certain particular limits as the problem is with particular design flaws of the a77.
Then maybe someone puts up a simple proggie that saves the image automatically at a minimum pixelcount that's still not limiting its optical res, simply because not all real images need all those pixels present. Then if the 'alamies' of this world stay as stubborn as they are now, you simply choose only the images of the required pixelcount to send over, as those images are automatically pixel-perfect by design.

But it's only a matter of time when all the cams on offer are going to have pixelcounts greater than the real optical resolution. And that's quite normal. So we have to adapt anyway and we have to automate our workflow somehow.

Still it would seem that as of right now, the better 24Mp APS solution is the yet unreleased Nex7 plus the EA2. From whatever I see in the samples available, its sensor toppings/filters act in very much the same way as the original Nex5's did. So you get the topmost res for the pixelcount and at the same time never get the aliasing problems of the Nex5 because of the much higher digital res.

[alphaomega]

Agorabasta wrote

Still it would seem that as of right now, the better 24Mp APS solution is the yet unreleased Nex7 plus the EA2. From whatever I see in the samples available, its sensor toppings/filters act in very much the same way as the original Nex5's did. So you get the topmost res for the pixelcount and at the same time never get the aliasing problems of the Nex5 because of the much higher digital res.

Just what I wanted to hear as my next (and only) Sony purchases will be that EA2 and NEX-7.

[Greg Beetham]

Agorabasta, the diffraction limit calculator at Cambridge in Colour when set to 24MP on a 1.5 crop factor sensor shows that the A77 is exceeding the diffraction limit by a good margin at f11 with lenses that are built to standards that we are used to. That calculator has the pixel size of the sensor at 4um but that is only approximate because of a few unknown factors, like if the sensels are square or rectangular and how much space there is between sensels etc. Convexly if you set the calculator at 16MP (4.8um) the result shows that at f11 the sensor is only just exceeding the diffraction limit.
http://www.cambridgeincolour.com/tutori ... graphy.htm
That’s the reason why I say optically the 24MP sensor on APS-C format is now exceeding the limit of most of the lenses we have for A-mount. You can build lenses with a smaller diffraction limit of course…if you want to pay the money, Bill Gates would probably not hesitate to pay for them, I however with more modest means am restricted to what is available and in production commercially.
Greg

[bfitzgerald]

I watched a video from some photo tech professor/expert and he pretty much said about 3-4um is all you'll get and any more won't yield any actual improvements to images even with the extra pixels. I'll dig about on that to see if I can find it stuck into a few pc jobs right now.

Of course for a FF sensor you can go higher but we're really not going to get a lot more out of this one. I know this is crushing news to resolution lovers (Hello DK!) but really I think it's time the tech moved beyond this one track mission and focussed on some other important areas. The diffraction part is of course right I've a pal who tried some of that super high res Adox 35mm film and he said you're diffraction limited even at modest apertures you've to shoot it fairly fast speeds to actually get the res. You can stop down but obviously you've hit the diffraction wall/ I think above f4 and you're in trouble on that film it's that bad for diffraction and you'd want to use some pretty damn good primes too.

Unless this sounds like a great idea (not being able to stop your lenses down without mashing your images) I'd say call it quits and let's get off the pixel bandwagon

[Greg Beetham]

Yep I reckon Canon have seen the light (so to speak) with this memory filling folly.
Greg

[Greg Beetham]

David I can see what you are saying, that if you use the A77 within a blinkered range of settings and use high class lenses you can take advantage of the resolution on offer for commercial benefit, but for me I think the A77 is not really a forgiving enough camera, my photography is rather fly by the seat of the pants when I’m in the mood, I don’t really have a lot of interest in speciality cameras you have to tinker with and only use within a restricted range to get the best out of them.
Honestly I think the A900 is a better choice for your Alamy files any day of the week.
Greg

[agorabasta]

Greg,

You really must be believing those Cambridge tales more than you believe your own eyes. Because on a 12Mp body that you own you should never get a pixel-sharp image at f/22. But guess what - you get them every single time at f/22 with good lens. Maybe it's about time to start thinking about why is it so?

And then, building a zoom lens capable of exhibiting real diffraction effects in the image is possible but it would be absolutely prohibitively expensive. Nobody builds lenses like that, they simply let other factors to limit the lens performance well before that - it's much cheaper and often also looks better.

But there is still a chance to see diffraction in APS cams. It will be happening on a regular basis when the theoretical diffraction limit gets hit at about f/2 - then all DSLR lenses would be transparent enough to show real diffraction effects while wide open - the sensors would be 200Mp at APS. Until then, the best place to look for diffraction is in the microscope.

[Greg Beetham]

Agorabasta I will admit the A77 does have great looking JPEG’s just the same with amazing detail, it’s just the range of situations and settings where you can actually get those great JPEG’s is restricted more than in previous models it seems. People who shoot landscapes for instance (never mind for a moment, the people who do close-ups and macro) aren’t going to be impressed to learn that using f11 on the A77 could actually degrade their images quite a bit more than f11 on the A55.
You can’t keep blaming sensor toppings for every foible that comes along without any concrete evidence that that is the definitive causal agent; besides there is no absolute evidence one way or another for sure that the inclined mirror is not contributing to an increase in diffraction, when the NEX 7 is generally available and comparisons can be done we will know more.
There is evidence from Cambridge in Colour that diffraction increases as the sensel size gets smaller with current lens optical design limits. The design envelope for a given lens has to have quite a few practical limits set in stone before it’s put into production, like what is the expected resolution at a given viewing distance versus print size versus cost of the lens to deliver that resolution etc.
Yes I agree you can have theoretical optics that will be ok with very small sensels but we have to use what we have, not theoretical lenses, to make those theoretical lenses in reality will cost….lots and lots of coconuts, there would likely be very few customers world-wide for lenses of such high cost, it would be a very game lens maker that would embark on such a venture I feel.
Greg

[agorabasta]

Greg,

Could you please answer a simple question - do you ever get pixel-sharp images at f/22 from a 12Mp sensor?

If yes, then the 'diffraction limit' is getting somehow broken.

Then the question becomes - how exactly can some not-so-expensive lenses break that limit?

I actually have some answers to that question, but that would be another matter to discuss afterwards.

P.S. Those Cambridge boys need to create a computer-simulated image of a photo showing those diffraction effects they profess there. Because the real captured samples they show there have absolutely nothing to do with diffraction blurring.

[agorabasta]

You can’t keep blaming sensor toppings for every foible that comes along without any concrete evidence that that is the definitive causal agent; besides there is no absolute evidence one way or another for sure that the inclined mirror is not contributing to an increase in diffraction, when the NEX 7 is generally available and comparisons can be done we will know more.

I can keep blaming sensor toppings simply because there's nothing more to blame.

The evidence is - a55 and NexC3 are equally good to the best lenses, the mirror in a55 does not make it worse than C3, the mirror in LA-EA2 has no effect on image quality neither with Nex5 nor with Nex5N/C3, the Nex5N is much worse with sharpest lenses I have than the NexC3 is. Then mind also, that a55/5N/C3 all have the essentially same 16Mp sensor.

What follows is that neither the SLT mirror, nor the pixelcount may produce the differences observed. It's the sensor toppings that cause all the hoopla, period.

[Greg Beetham]

Actually agorabasta I don’t use f22 (that I can remember) because of the softness factor, I suppose I should investigate f22 more thoroughly to see just how soft it is, and I never bother looking at any of my images at pixel level either, I give the image a pass or fail as it looks on the screen.
There has been numerous statements by others that f22 is soft, softer than f16 which is in turn softer than f11 so I guess I go along with that at face value, I’ve never taken the time to investigate just what degree of softness we are talking about, it may be that it’s a storm in a teacup on some cameras, or it might become a more pronounced issue on others.
But whatever, the results speak for themselves in my book, and the appearance of that A77 ISO640 shot of David’s at f11 is not up to scratch.
Like I say, wait for the NEX 7 and then we might be able to prove things one way or another.
Greg

ps. If the NEX 5n is worse then the others why do people think it's so wonderful? That looks like yet another mystery.

[stevecim]

Have not seen anyone at DPR talking about f11 issues, but then, I seen heaps of post there, like "man look at all then sensor noise in iso100 shots" and lot of poster agreeing , yet what they a looking at is just not so nice bokeh , it's seems that every time an images does not have silky smooth out of focus areas it's a 24MP fault.

David , please keep us posted on your A77 findings, I stopped reading most post at DPR , they create a lot of NOISE but not much value.

P.S, had my first 20min play out doors with the sigma 18-250, just before sunset, so lighting was not the best for my sort of shooting
but the A550 feels like a whole new camera, the sigma 18-250 HSM , beats the the combo of sigma 18-50 DC + Tam 70-300 in every way.

[redsim74]

Have not seen anyone at DPR talking about f11 issues, but then, I seen heaps of post there, like "man look at all then sensor noise in iso100 shots" and lot of poster agreeing , yet what they a looking at is just not so nice bokeh , it's seems that every time an images does not have silky smooth out of focus areas it's a 24MP fault.

I didn't think it was possible, but the Sony SLR/NEX forums at DPR are actually getting worse. Very few posters left with anything useful to contribute.

[agorabasta]

ps. If the NEX 5n is worse then the others why do people think it's so wonderful? That looks like yet another mystery.

Actually, the 5N is a very fine camera in most aspects under most shooting conditions. But its image quality simply falls apart in some very specific ranges of use.
Below is quite an example shot at f/11 with the 5N. First the full resized image, then a centre crop and another crop showing the problem. It looks like an object well within DOF is very blurred and also shows tons of CA; mind also that the CA is well corrected for the sharper parts of the image.
The lens used was the 16mm pancake with the UWA adapter, so it's 12mm lens with tons of DOF. And it's actually near pixel-perfect if used on the original Nex5, except for the corners and extreme edges.
That's why I also never use the 14mm Samyang with the 5N, I use it with C3 and a55 instead, and also with the original Nex5.

[jcoffin]

Here's a series I took a while back, all on an A700, a few seconds between shots, so the conditions are essentially identical throughout (on tripod, mirror prefired, same light, etc.):

f/11:


f/16:


f/22:


At least to me, it seems quite clear that the shot at f/11 is the best of the three. At f/16 the degradation isn't too terrible, but there pretty clearly is some. At f/22, it seems pretty obvious (at least to me) that the quality has deteriorated quite badly compared to f/11.

Anybody who cares can look at the resized but uncropped version: http://s1212.photobucket.com/albums/cc4 ... SC4210.jpg.

Admittedly, we're pixel peeping here (these are 100% crops), but Agorobasta's comment was specifically about being pixel-sharp at f/22, so at least a little pixel peeping seems justified.

Of course, it's possible I'm just incompetent, and it's purely an accident that the shots get progressively worse as the aperture gets smaller -- but if you're going to claim that, at least have the courtesy to point out what I've done wrong.