A900 Update

[ErikKaffehr]

Well, mostly when I stitch I use it either instead of cropping or to increase the angle of view. If I had an A900 I would probably still stitch.

Erik

That is true, stitching will work, as long as the subject is static.  I would argue that if you are willing to stitch and the subject is static, even an extremely low-res APS-C camera can produce staggeringly high resolution images, as long as you stitch multiple takes.  

[Ray]

Dpreview should really have tested the A900 with JPEG/"Extra Fine" instead of the "Fine" that they tested with.  Maybe it is a matter of terminology but 'Fine' in the Sony lexicon seems to be clearly positioned between 'Extra Fine' and 'Coarse', while in case of a competitor, 'Fine' is in reality the 'Extra Fine' of Sony.  So for apples-to-apples comparison, they should have ignored the terminology and employed Sony's "Extra Fine" in their JPEG comparison....just a side observation I had.

We're all capable of making mistakes. Perhaps Dpreview are not used to seeing an 'Extra Fine' option. Perhaps they tried it and saw no improvement which they thought was significant for their demonstration purposes. Who knows! It's of no great concern to me because I don't shoot in jpeg mode.

============
However, the amount of detail that is rendered by the Sony's new 24MP sensor is quite simply astonishing. If resolution and detail at low sensitivities are high up on your priority list it'll be difficult to ignore Sony's new flagship.
============
Naturally the 24 MP Sony DSLR-A900 stands out in this test, offering considerably more resolution than its 12 megapixel competitors. The resolved detail is quite simply staggering.
============

"Staggering" and "Astonishing" are not terminologies that dpreview is fond of using, when it comes to cameras from Sony - based on past experience.

Bottomline, if you need ultra-high-resolution, then it would be impractical to expect stratospheric high-ISO performance and when you get extraordinary high-ISO performance, you will have to give up on ultra-high-resolution.  Just the "physics" of the situation.  

I think you made a statement earlier, that in the A900, detail is missing, while noise is present....dpreview seems to think otherwise, based on the above quotes.

"Staggering" and "astonishing" are not scientific terms. Anything, absolutely anything, can be described as astonishing in relation to something which is not astonishing, or in relation to something which is no longer astonishing but perhaps once was.

The second part of this quote should give you a hint as to what this degree of staggering astonishment is in relation to...."Naturally the 24 MP Sony DSLR-A900 stands out in this test, offering considerably more resolution than its 12 megapixel competitors. The resolved detail is quite simply staggering."

Dpreview have enough experience to know what to expect when you double the pixel count of a sensor, just as Imaging Resource does not have to make a song and dance about how BIG the A900 files are. We should all know that 24mp is precisely twice as big as 12mp. Reviewers sometimes feel the need to use a literary style to get the reader interested, which is fair enough provided they also show us the results of their careful testing so we can see for ourselves just how astonishing those results might be.

Anyone who has a 40D or other brand of cropped format DSLR of similar pixel count, can test for themselves just how much detail to expect from a 24mp full frame DSLR, at least in that central part of the 35mm frame which the cropped format represents. I've taken the trouble myself to compare my 5D with my new 15mp 50D, using the same lens at the same aperture from the same distance to subject. The 50D represents a 39mp full frame sensor which has been cropped in software to 15mp. Even at F22 I can find a very marginal increase in detail, but such extra detail would of course only be visible in a very large print on close scrutiny.

However, the increase in detail at F8 is just astonishing. (Not really. I expected it   ).

Ray, based on what I have seen around, the opinion seems to be that the D3/D700 are great all-purpose machines with a FF sensors and due to its relative low resolution (for a FF), it is optimized for high-ISO performance and sheer speed (FPS).  

The A900 on the other hand, is equipped with a high resolution sensor and thus expecting it to have comparable high-ISO performance as the D3/D700 is not very realistic.   But for a person who needs that resolution, the lower resolution sensor is not an option at all.

There's a fallacy here which needs addressing. In fact, it has already been addressed many times, even by photography enthusiasts with a PH.d. in Physics, such as Emil Martinec who occasionally makes an appearance on this forum.

Increasing pixel count does not necessarily have to result in greater image noise.  There are numerous examples in Canon's development of DSLR technology which demonstrate this. The D60 had twice the pixel count of the D30 which it succeeded, yet noise even at the pixel level was no greater than that of the 3mp D30. Likewise with the 8mp 20D which followed the 6mp 10D; the 8mp 1D2 wich followed the original 4mp 1D; the 16mp 1Ds2 which followed the 11mp 1Ds and the latest 21mp 1Ds3 which has slightly lower noise than the 1Ds2 despite a greater pixel count.

The other consideration is that all images without exception have to be viewed at a specific size. One has to be careful when comparing image noise in prints, or on monitor, when such images are different sizes, unless one is examining the noise characteristics of an individual pixel.

Whilst it's useful to know the comparative noise level of the individual pixel, which DPreview shows us, such noise at the pixel level will always be modified by the degree of enlargement of the final print. For example, a D700 or D3 pixel has less noise than a 1Ds3 pixel. However, when comparing equal size images, the D3 image will have to be enlarged to the same size as the 1Ds3 image, which inevitably results in an enlargement of D3 noise, or alternatively, the 1Ds3 image will be downsized to the D3 image size, a process which inevitably results in a discarding of some 1Ds3 noise (as well as resolution).

Whilst D3 owners like to assert that their camera has the lowest noise of any other DSLR on the market (because they are 'sort of' supporting their favourite football team  ), the facts of the matter appear to be that D3 noise is pretty much on a par with 1Ds3 noise, in practice.

I see no reason why the 24mp upgrade to the Nikon D3 should have more noise than the 12mp D3, when equal size images are compared. Unfortunately, Sony doesn't seem to have quite made it, in the high ISO stakes.

[Ray]

Just a small correction.  The new "G" SSM zooms are 70-300 and 70-400.  

In dpreview, there was a post relating to images from the A900 and the new 70-300 G SSM lens:




100% crop from the above image:

The 100% crop of the cock's head certainly looks very sharp and detailed. But I would expect similarly impressive results from a 40D with the same lens from the same shooting position. The cock's head would then almost fill the frame as shown in the following image with 40D format superimposed.

[attachment=8840:Cock_with_crop.jpg]

[aaykay]

The 100% crop of the cock's head certainly looks very sharp and detailed. But I would expect similarly impressive results from a 40D with the same lens from the same shooting position. The cock's head would then almost fill the frame as shown in the following image with 40D format superimposed.

[attachment=8840:Cock_with_crop.jpg]

Ray, this is getting really funny.  By your 40D crop logic, would we not get an even more detailed view if you were to take the picture with a 5 year old 7.2MP Sony DSC-V3 point-and-shoot (or the newer 13.5MP Sony DSC-W300 P&S), since such p&s cameras would magnify an even smaller cropped portion of the cock's head, since that 7.2MP (or 13.5MP) on such small sensors, are WAY denser than even the 50D or any other dSLR in existence ?  

[aaykay]

I see no reason why the 24mp upgrade to the Nikon D3 should have more noise than the 12mp D3, when equal size images are compared. Unfortunately, Sony doesn't seem to have quite made it, in the high ISO stakes.

Ray, you seem to have already made up your mind but is circling around this topic again and again.   You seem to be hunting around for statements (like in the IR review) that will "confirm your pre-set suspicions" about the A900 and ignoring others that are contrary to your already-made-up view.

I think several 1600, 3200 and 6400 ISO images from the A900 were shown on several foras, with terrific performance and remember these images were processed using very early versions of post processing tools (since none of them are "mature" enough when it comes to processing A900 RAW files - yet, and which in turn are being compared to "highly mature" processing tools for cameras like the 1DSMKIII, which have been on the market for a year now).  

The JPEG engines of Sony cameras are typically sub-par but that should not be a problem at all, for people willing to work in RAW.  I personally only work in RAW and never shoot JPEG.  But several of these reviews seem to focus a lot of JPEG output and berate the overall product for that.

The sensor in the A900, is the same design as the A700 and the Nikon D300, just that the pixels in the A900 are slightly LARGER and as long as you are willing to work with RAW,  I don't see any "problems" here.  I never heard of anybody complaining about D300 high-ISO images, which in turn are comparable to the A700 high-ISO images (after the V4 firmware upgrade)....the A900 can only be better at the pixel level, since it is the same design with further optimization for a whole year *AND* captures more light per pixel than either the D300 or the A700 along with 235% more light at the sensor level  *and* the A900 has slightly weaker AA filter (than the D300/D700/D3).  People shooting with both the Nikon D700 and the A900 have commented that the D700 images are slightly "fuzzy" because of the strong(er) AA filter when compared to the A900.

Don't know what will convince you but I think indications are that you will be a happy person with the new Canon 5DII and you are simply circling around here for someone to convince you about that.  I think based on the fact that you have a significant lens and body lineup from Canon, I would stick with Canon if I were you, since IF you REALLY want to get full advantage from the A900, you will need to invest in the top-of-the-line Zeiss AF lenses and the "G" series lenses, which are certainly not cheap by any stretch.  If you intend to use your decades old Minolta lenses on the A900, you will get reasonable results but those are simply not comparable to the images from the 2007/2008 designed Zeiss Full-frame lenses in the Alpha mount, that are specifically developed for Full-frame Digital sensors.....just the way things are.

[Ray]

Ray, this is getting really funny.  By your 40D crop logic, would we not get an even more detailed view if you were to take the picture with a 5 year old 7.2MP Sony DSC-V3 point-and-shoot (or the newer 13.5MP Sony DSC-W300 P&S), since such p&s cameras would magnify an even smaller cropped portion of the cock's head, since that 7.2MP (or 13.5MP) on such small sensors, are WAY denser than even the 50D or any other dSLR in existence ?

You've missed the point, old chap. I use the example of the 40D crop because the 40D pixel is about the same size as the A900 pixel. Both cameras are modern and therefore employ the latest technology. My point is, if one wants an idea of the astonishing quality the A900 is capable of, one can use a 40D to find out, provided one is able to use a lens of comparable quality. Of course, such a comparison tells one nothing about the edge performance of the full frame A900 with the particular lens used.

Another very practical issue flowing from this concept relates to the practice of using the cropped format camera as a lens extender when one's longest telephoto lens does not have sufficient reach. I do this myself whenever I travel. A higher pixel-density sensor is usually better than a 1.4x or 2x extender.

With an A900, there would be no advantage in carrying a 10mp cropped format camera as a lens extender. That would be a weight saving.

It would be interesting to see a comparison between the A900 and 14mp A350, using the same lens at the same distance. I think one would need a very high quality telephoto lens to provide any worthwhile additional detail in those circumstances. I haven't yet compared my 50D with the 40D, using my longest telephoto. I'm doubtful the 100-400 is good enough to reveal any worthwhile differences at F8 and F11, although I would expect to see some very minor improvement at 200% enlargement and above.

[Ray]

Ray, you seem to have already made up your mind but is circling around this topic again and again.   You seem to be hunting around for statements (like in the IR review) that will "confirm your pre-set suspicions" about the A900 and ignoring others that are contrary to your already-made-up view.

Not at all. Nothing could be further from the truth. I've always been very interested in the A900, since I first heard the announcement of its development. I have 3 Minolta-fit lenses which are all above the 4 out of 5 Photodo rating; the Minolta 50/1.4, the Sigma 24/2.8 and the Tamron SP 90/2.8 macro. I'm not sure that any modern equivalents to these 10 year old lenses would necessarily be much better.

I think several 1600, 3200 and 6400 ISO images from the A900 were shown on several foras, with terrific performance and remember these images were processed using very early versions of post processing tools (since none of them are "mature" enough when it comes to processing A900 RAW files - yet, and which in turn are being compared to "highly mature" processing tools for cameras like the 1DSMKIII, which have been on the market for a year now).  

The JPEG engines of Sony cameras are typically sub-par but that should not be a problem at all, for people willing to work in RAW.  I personally only work in RAW and never shoot JPEG.  But several of these reviews seem to focus a lot of JPEG output and berate the overall product for that.

That's a very good point, but you should understand that objective and fair-minded people like myself can only draw conclusions from the evidence available. If there is a general issue with sub-par JPEG performance from Sony cameras, one might wonder (1) Why has Sony not fixed this problem in their flagship model? (2) Why have very experienced reviewers like Imaging-Resource not mentioned a caveat in relation to this, along the lines, "In our experience, Sony JPEGS in the past have proved to be much noisier and of lower resolution than results from their RAW images at high ISO, so the JPEG images shown in our Comparator are not necessarily an indicator of RAW performance"?

Don't know what will convince you but I think indications are that you will be a happy person with the new Canon 5DII and you are simply circling around here for someone to convince you about that.  I think based on the fact that you have a significant lens and body lineup from Canon, I would stick with Canon if I were you, since IF you REALLY want to get full advantage from the A900, you will need to invest in the top-of-the-line Zeiss AF lenses and the "G" series lenses, which are certainly not cheap by any stretch.  If you intend to use your decades old Minolta lenses on the A900, you will get reasonable results but those are simply not comparable to the images from the 2007/2008 designed Zeiss Full-frame lenses in the Alpha mount, that are specifically developed for Full-frame Digital sensors.....just the way things are.

I don't need anyone to convince me. I can make up my own mind provided I have the information. At this stage, the information is not completely reliable for the reasons you've stated. I'm in no hurry. I can wait till the RAW comparisons are out, using production models of both the A900 and 5DMkII.

DSLRs are of no use without lenses. When initially choosing a system, or thinking about changing to another system, the quality of available lenses in the range one knows one is likely to use, is a very important consideration.

You must have gathered from my previous comments that I'm not entirely happy with the wide angle performance of my Sigma 15-30 and the telphoto performance of my Canon 100-400 zoom at 400mm and F5.6. Simply upgrading to a 5DMKII is not going to change this. I shall be looking closely at comparisons between the the Sony 16-35 and Nikkor 14-24 when (or if) they become available, as well as comparisons between the new Sony 70-400 and the old Canon 100-400. I haven't bought the Nikkor 14-24 yet, although I received a Mark Welsh adapter a few days ago. I probably will buy the Nikkor, if only to use it on my 5D. Should Nikon announce a real humdinger in the form of a 24mp upgrade to the D700 at an attractive price (before I get either the A900 or 5D MkII) then the fact that I already have an excellent Nikkor lens will of course influence my decision.  

[aaykay]

Not at all. Nothing could be further from the truth. I've always been very interested in the A900, since I first heard the announcement of its development. I have 3 Minolta-fit lenses which are all above the 4 out of 5 Photodo rating; the Minolta 50/1.4, the Sigma 24/2.8 and the Tamron SP 90/2.8 macro. I'm not sure that any modern equivalents to these 10 year old lenses would necessarily be much better.

I think the above lenses should work just fine - for the most part - but digital sensors behave significantly differently from film and thus newer digital-oriented lens designs have digital-specific lens coatings and specially "shaped" lens elements that are intended to optimize their use with digital sensors, along with improving its telecentricity.  In the film world, non-telecentric lenses would at best introduce some vignetting....in the digital spectrum, the same lenses would cause vignetting, light scatter (when light falls at an angle on the micro-lenses of a digital sensor) and color bleeding.  The older film designs are a gamble at best, on digital sensors (specifically on Full-frame digital sensors), even though they might perform adequately, for all you know.

That's a very good point, but you should understand that objective and fair-minded people like myself can only draw conclusions from the evidence available. If there is a general issue with sub-par JPEG performance from Sony cameras, one might wonder (1) Why has Sony not fixed this problem in their flagship model? (2) Why have very experienced reviewers like Imaging-Resource not mentioned a caveat in relation to this, along the lines, "In our experience, Sony JPEGS in the past have proved to be much noisier and of lower resolution than results from their RAW images at high ISO, so the JPEG images shown in our Comparator are not necessarily an indicator of RAW performance"?

Fair enough.  But having used several Sony cameras in the past, I can state that their RAW and JPEG engines are world's apart.  The RAW is mindblowingly good (except in case of the A700 initially, when they did NR on RAW - and reversed by Firmware Ver.4) while the JPEG is passable at best.  Olympus on the other hand, is extremely good with their JPEGs and people see very little improvement from shooting RAW.  Canon treats the JPEG outputs from their professional models, differently from their crowd-pleasing consumer models (output being audience-specific).  Just the nature of how the beasts are created.

Either way, it is your time and money.  I doubt you would go wrong by making a choice in any of these camps.  I have deliberately chosen Sony because of the in-body stabilization in a Full-frame body, which just is not available at all in the other options, along with ultra-modern lenses.  Having shot with the Canon 5D/1DSMKII and the 85mm f/1.2L (Ver.1), what I find is that even though the lens itself is built as a low-light tool, it is well nigh impossible to hold it steady in low light, since it does not have IS built-in.  When you want a slow shutter speed, even a 35mm or a 50mm will also benefit from I.S.  The 135mm f/2.0L is also another example where having I.S is a big plus, regardless of what the Canon PR machine wants you to think.  

Anybody stating that they don't WANT/NEED I.S in the 24-70 f/2.8L or the 16-35 f/2.8L or the 17-40 f/4L etc are not speaking truthfully.  Basically, it is just not possible to accomodate additional I.S lens elements into these primes or zooms, without significantly increasing their size and the PR machine of these camera companies, kicks in and tries to shove this problem under the carpet, by pretending it does not exist - while at the same time, denying I.S in the body.   Even a 24mm will benefit from I.S, believe it or not.  

Also, when you introduce compensating I.S elements into the lens, something has to give - you are introducing additional elements with more "matter" through which light has to pass through.  For instance, if they introduce a new 70-200 f/4L, with 2008 technology, it will almost certainly be superior to the 70-200 f/4L IS, since it will not have to have the additional compensating  I.S elements that move around in the lens (when I.S kicks in), along with the additional mechanical complexity within the lens through the additional gyro-sensors and such.

So Sony has some very real advantages here.  It has introduced a Full-frame camera with some really compelling features like a 100% VF etc, unlike smaller players like Pentax and Olympus, who seem to be happy operating in the small sensor space.  Most of the Sony FF lenses are ultra-modern and top-of-the-line and designed from the ground up, for digital Full-frame sensors.  None better in the market.  A similar A900-type sensor in the A700 and the Nikon D300, with SMALLER pixels, are doing just fine, with no complaints even at 3200 or 6400 ISO.  So where is all this ruckus about high-ISO noise in the A900 images, coming from ?  The Ver.4 firmware upgrade in the A700 (which has radically transformed the imaging in the A700 for the better, bringing it at par with the D300), was a DIRECT result of the research efforts around the A900 imaging engine.

Either way, good luck with whatever you decide on.  

[aaykay]

Hand-held (no tripod used) picture of a lens taken with the A900 and Carl Zeiss 24-70 f/2.8 Vario-Sonnar (ISO 200, 70mm, f/8, 1/13sec):

[BernardLanguillier]

That is true, stitching will work, as long as the subject is static.  I would argue that if you are willing to stitch and the subject is static, even an extremely low-res APS-C camera can produce staggeringly high resolution images, as long as you stitch multiple takes.  

Actually, stitching works pretty well with many non static subjects, they just need to be small enough to fit into one of the frames or need to be steady/stichastic enough (water in a river for instance) that the stitch will not be visible.

Other than that, I feel that DR and tone is a very important aspect of landscape photography where the D3/D700 is in a different league compared to most APS cameras I have had the chance to use. This is a tremendous asset when stitching.

Cheers,
Bernard

[Kenneth Sky]

What RAW converter are you using? My versions of LR and Aperture don't seem to handle A900 RAW files.

[Ray]

Anybody stating that they don't WANT/NEED I.S in the 24-70 f/2.8L or the 16-35 f/2.8L or the 17-40 f/4L etc are not speaking truthfully.  Basically, it is just not possible to accomodate additional I.S lens elements into these primes or zooms, without significantly increasing their size and the PR machine of these camera companies, kicks in and tries to shove this problem under the carpet, by pretending it does not exist - while at the same time, denying I.S in the body.   Even a 24mm will benefit from I.S, believe it or not.  

Also, when you introduce compensating I.S elements into the lens, something has to give - you are introducing additional elements with more "matter" through which light has to pass through.  For instance, if they introduce a new 70-200 f/4L, with 2008 technology, it will almost certainly be superior to the 70-200 f/4L IS, since it will not have to have the additional compensating  I.S elements that move around in the lens (when I.S kicks in), along with the additional mechanical complexity within the lens through the additional gyro-sensors and such.

aaykay,
I've never seen any statements from people claiming they don't want or need IS in ultra-wide zooms. However, if it's true, as you claim, that IS with a floating element built into the lens, will inevitably reduce the potential quality of the lens, then that might be a reason for not wanting it.

There's a certain logic to this claim, but how it works out in practice is not clear. The last wide angle zoom I bought, the EF-S 17-55/2.8 IS, is very sharp; as sharp as the Canon 50mm primes which don't have IS. However, because it's an EF-S lens designed for the crop format, it has a smaller image circle and therefore a slightly worse corner performance than the 50mm primes when they are used on the crop format.

The Canon 300/2.8 IS, 70-200/2.8 IS and 70-200/4 IS all have fine reputions for being sharp lenses. On the other hand, the very finest Canon lenses, such as the 200/1.8 and 85/1.2 don't have IS.

There's no doubt that a good anti-shake sensor is a big advantage. I wouldn't argue against that. But it's probably less useful for wide angle lenses than for telephotos. Having recently bought a very lightweight Manfrotto 190CXPRO4 carbon fibre tripod with 460MG pan & tilt head, ideal for travelling, I'm not too worried about a lack of IS in ultra-wide angle lenses, although I'd prefer to have it.

[aaykay]

The last wide angle zoom I bought, the EF-S 17-55/2.8 IS, is very sharp; as sharp as the Canon 50mm primes which don't have IS. However, because it's an EF-S lens designed for the crop format, it has a smaller image circle and therefore a slightly worse corner performance than the 50mm primes when they are used on the crop format.

Check out the size difference between the EF-S 17-55 f/2.8 IS and the Tamron 17-50 f/2.8.  Don't know if the massive size difference between these 2 EF-S lenses (Canon and Tamron) are due to the presence of IS on one (and the additional floating elements that will be added to the lens due to it) and lack of it on the other (which is equally highly rated for optical quality on EF-S bodies):






From the below link:
http://www.the-digital-picture.com/Reviews...ens-Review.aspx

[aaykay]

aaykay,
I've never seen any statements from people claiming they don't want or need IS in ultra-wide zooms. However, if it's true, as you claim, that IS with a floating element built into the lens, will inevitably reduce the potential quality of the lens, then that might be a reason for not wanting it.

If you go into several of the Canon/Nikon forums, that is exactly what several people claim.  That they don't need IS.  What a convenient viewpoint, especially since Canon or Nikon does not offer IS for those (35mm, 50mm, 85mm, 135mm etc).

Also, it is not *a* floating element built into the lens...there are *several* elements that will need to be added into an IS lens, to compensate for the image degradation that the floating elements would introduce into the design.   Check out the number of *additional* elements/groups in the 70-200 f/4L IS or 70-200 f/2.8L IS, when compared to their non-IS counterparts.  The saving grace for the IS versions in this case is that the non-IS versions are much older designs and thus the IS versions have got a leg up over them.

When we say the floating elements will inevitably reduce the potential quality of the lens, well, the "reduction in potential quality" is applicable to ALL lenses (specifically the ones currently having IS) -  not localized to the ones that do not come with IS, due to design problems in adding IS to those.

Obviously, when these companies started out by adding in-lens IS, the bodies most in use, were film bodies, where an in-body stabilization was not an option at all (can't shake the film, can we ?).   And when they moved full tilt into the digital landscape (and the demand for film bodies waned), they are unable to shake off the prior film legacy built into their lenses and begin all over again - probably due to the heavy marketing done around in-lens IS, and the price premium an IS lens commands over a non-IS lens.  As long as the users continue to defend the companies' mode-of-operation, why would they change course and add IS into the body (with a firmware option that would turn off in-body IS automatically, when it detects an IS lens being mounted and otherwise employ the in-body IS) ?  

Having said the above, I do realize that there are several longer lenses which do benefit from in-lens IS, regardless of the image degradation the additional IS elements would introduce.  More of a benefits outweighing the downsides, kind of deal, since it allows one to see the stabilized image on the VF, than merely allowing the *capture* of a stabilized image.  But quite frankly, for the really long lenses, where this phenomenon is most prevalent,  I prefer to break out my Series-5 or Series-3 Gitzos - works really well.

[Ray]

If you go into several of the Canon/Nikon forums, that is exactly what several people claim.  That they don't need IS.  What a convenient viewpoint, especially since Canon or Nikon does not offer IS for those (35mm, 50mm, 85mm, 135mm etc).

Exactly! This is a very convenient strategy that some people employ when their ego and self esteem is tied up with their choice of equipment. Whilst I haven't come across people claiming they don't need or want IS in the wider angle focal lengths (probably because I don't get around other forums much), I have come across similar comments when people are confronted with the fact that their camera is a bit noisy at high ISO. People sometimes then defend their choice of equipment along the lines, "Doesn't worry me. I never use ISO 1600." The obvious response to that is, "Of course you never use ISO 1600 if image quality is not acceptable."

Also, it is not *a* floating element built into the lens...there are *several* elements that will need to be added into an IS lens, to compensate for the image degradation that the floating elements would introduce into the design.   Check out the number of *additional* elements/groups in the 70-200 f/4L IS or 70-200 f/2.8L IS, when compared to their non-IS counterparts.  The saving grace for the IS versions in this case is that the non-IS versions are much older designs and thus the IS versions have got a leg up over them.

When we say the floating elements will inevitably reduce the potential quality of the lens, well, the "reduction in potential quality" is applicable to ALL lenses (specifically the ones currently having IS) -  not localized to the ones that do not come with IS, due to design problems in adding IS to those.

This is something which needs further investigation. Unfortunately, in the absence of proper testing of lenses, along the lines that Photodo employed several years ago, it's very difficult to compare lenses across different systems. I would never just assume that, because a lens does not contain the additional elements required for IS, it is likely to have better absolute image quality than a similarly priced lens that does have built-in IS.

Having said the above, I do realize that there are several longer lenses which do benefit from in-lens IS, regardless of the image degradation the additional IS elements would introduce.  More of a benefits outweighing the downsides, kind of deal, since it allows one to see the stabilized image on the VF, than merely allowing the *capture* of a stabilized image.  But quite frankly, for the really long lenses, where this phenomenon is most prevalent,  I prefer to break out my Series-5 or Series-3 Gitzos - works really well.

This characteristic of the anti-shake sensor worries me a bit. I find it very useful to see a stabilised image through the viewfinder and particularly useful when trying to manually focus using Live View with camera hand-held. At the same time, I would prefer that autofocussing be so accurate that manual focussing is never required.

Lack of accuurate focussing and/or lack of a sufficiently fast shutter speed for the conditions, are the two major culprits in most those shots of mine, over the years, which I consider are technically flawed, followed by incorrect exposure. I would find it very constraining if I were in a position again where I felt reluctant to go above ISO 200 because of image degradation. There are many situations outside of the studio where subject movement, in conjunctuion with less than ideal lighting, dictates that one use a fast shutter speed. The longer the lens, the faster the shutter speed required in such circumstances

[BernardLanguillier]

Exactly! This is a very convenient strategy that some people employ when their ego and self esteem is tied up with their choice of equipment. Whilst I haven't come across people claiming they don't need or want IS in the wider angle focal lengths (probably because I don't get around other forums much), I have come across similar comments when people are confronted with the fact that their camera is a bit noisy at high ISO. People sometimes then defend their choice of equipment along the lines, "Doesn't worry me. I never use ISO 1600." The obvious response to that is, "Of course you never use ISO 1600 if image quality is not acceptable."

As a Nikon user, I would love to have body IS and can only praise Sony for having gone that route in a FF body. This simply extends significantly the range of possible applications in photography.

It doesn't mean that I would use this in more than 10% of my images, but these 10% are important too. I am sure that Galen Rowell would have loved to have body IS on wide glass when shooting aerials with his 35 f1.4, or when wide taking shots hanging high on a an iced cliff with 100 kmh winds turing his rope into a pendulum.

Cheers,
Bernard

[Ray]

I am sure that Galen Rowell would have loved to have body IS on wide glass when shooting aerials with his 35 f1.4, or when wide taking shots hanging high on a an iced cliff with 100 kmh winds turing his rope into a pendulum.

Cheers,
Bernard

Bernard,
I'm not sure that IS, whether built in the lens or built in the sensor, would help much when swinging on a rope with 100km/h winds   .

Working from the principle that a sharp image of a static subject requires a shutter speed at least twice 1/FL and preferrably 3x, shutter speeds of 1/30th to 1/60th sec with a 15mm lens are often quite achievable without the necessity of jumping to a high ISO.

[BernardLanguillier]

Working from the principle that a sharp image of a static subject requires a shutter speed at least twice 1/FL and preferrably 3x, shutter speeds of 1/30th to 1/60th sec with a 15mm lens are often quite achievable without the necessity of jumping to a high ISO.

Well, often but not always. Achieving sharp wide angle images when strong wind is blowing, when your are exchausted after a quick dash up a hill, when you are shivering with cold, when the boat you are on is moving with the waves, when the old Russian helictoper is shaking, when night is falling on swampy ground too soft for a tripod... those are real world situations where IS helps on a wide lens.

If I am given the choise to either use IS at ISO 100, or no IS at ISO 640, I'll always opt for the former. Today the A900 is the only body giving photographers option #1 with pro grade bright lenses (zooms or primes) while delivering class leading image quality. It is simply making possible images that were basically impossible to take before with the same level of performance. From that standpoint, it is clearly as revolutionary a camera as the 1ds3 or D3 were when they were released, and perhaps even more if you consider its price point.

Cheers,
Bernard

[Tony Beach]

Well, often but not always.

More like never; MLU is required to achieve absolute acuity at anything below about 1/250.  It might be sharp enough for some, but I wonder how many of those 24 million pixels are being thrown away by shoddy technique considering that even at a mere 12 million pixels I see a noticeable loss of detail between using MLU and not using it at slower shutter speeds.

[Ray]

Well, often but not always. Achieving sharp wide angle images when strong wind is blowing, when your are exchausted after a quick dash up a hill, when you are shivering with cold, when the boat you are on is moving with the waves, when the old Russian helictoper is shaking, when night is falling on swampy ground too soft for a tripod... those are real world situations where IS helps on a wide lens.

If I am given the choise to either use IS at ISO 100, or no IS at ISO 640, I'll always opt for the former. Today the A900 is the only body giving photographers option #1 with pro grade bright lenses (zooms or primes) while delivering class leading image quality. It is simply making possible images that were basically impossible to take before with the same level of performance. From that standpoint, it is clearly as revolutionary a camera as the 1ds3 or D3 were when they were released, and perhaps even more if you consider its price point.

Cheers,
Bernard

Bernard,
I've tried taking photos from the back of a moving elephant and I can assure you that IS doesn't help at all. The best one can do is try and take the shots during periods of minimum movement between one lurch and the next, using the highest ISO one's camera is capable of, consistent with acceptable quality.

I've tried taking shots at 1/13th sec with my Canon 24-105 IS at 24mm. It's difficult to get a sharp image. Pretty much hit and miss at that shutter speed. Maybe the A900 would do a better job at such slow shutter speeds, but generally, any subject that's suitable for such a slow shutter speed must be very static. Not much use for street photography without flash. In fact, they are the sorts of subjects that would mostly lend themselves well to the slow and methodical procedure of setting up a tripod.