Just thought you'd like to know that now the ACR4.6 beta is public I can process A900 files alongside 1Ds3 files.

Based on a very quick look today I'd say that the Sony files still falls slightly behind the Canon on sharpness and smoothness of tone. You have to look closely but the Canon is simply 'crisper' whilst the Sony has this odd granularity in smooth areas, not really noise, just a subtle lumpiness in areas where the Canon shows flat tone.

I'll look at some other files in a while, maybe when I get some time over the w/e.

The attached file is a screenshot with both images processed the same in ACR4.6. It may well be that the Sony file (on the right) can be 'massaged' to match the Canon.

Nick,
The differences are subtle but noticeable. But what ISO were these shots?

My main gripe about the A900 is its performance at ISO 1600 and 3200. If the 1Ds3 can better it, then the 5D MkII might even better better it.

No doubt, the Canon file is indeed cleaner and seems to resolve a little bit more by looking at the rendering of the boat's mast cables on the left of the image.

Cheers,
Bernard

Could you fix the CA on the Sony one?

The Sony has direct light so the contrast is higher, but the detail on the Canon is SO much better. This just proves that just because the MP count is larger doesn't necessarily mean you're going to get better IQ.

A friend that sent me this thread also sent me this interesting article:

Canon engineers held back by marketing department's "megapixel race"

I have looked at some other files and the Sony can look just as sharp and finely detailed as the Canon. In fact in one shot using the 135F2 and 135/1.8 the Sony was better. I also noticed that I shot the Sony at 200ISO and the Canon at 100ISO The Sony has a base ISO of 200 (I think) so this seemed reasonable.

Noise is definitely higher at base ISOs but merely adding 5 to the NR in ACR4.6 pretty much cancels this out.

There is still a slightly less smooth tone to the Sony, but the detail is certainly there. Only bench tests can really distinguish these camera as far as resolution is concerned so to all intents and purposes they are as good as each other IMHO.

Detail - draw
Noise - Canon by a head but at 1 stop less ISO.

Nick,
It sounds as though we're into pixel-peeping territory here. One would expect two modern sensors of the same size and type (CMOS) and similar pixel count to produce very similar image quality at base ISO.

Did you not take any shots at ISO 800, 1600 and above? It's at these ISOs one might expect to see the greatest differences.

Quote: A friend that sent me this thread also sent me this interesting article:...

NEWSFLASH: Engineer found who agrees wholeheartedly with his organisation's marketing department !

We are delighted to bring you a startling world exclusive interview with an Engineer who claims to be impressed by his company's marketing department.

"Not only do they work ever so hard," he was quoted as saying, "with long hours, often tiring themselves with late nights of corporate entertainment, they manage despite that to keep a coolly quantitative and analytical view of exactly how customers are segmented into different groups and are able to map feature desirability to these groups. I'm extremely respectful of those insights. Of course, I'm also delighted that without that vital step in the sales and development cycle, my colleagues and I wouldn't have jobs to do at all."

We were able to confirm our contact's Engineering credentials by the soup stains on his tie, his disarmingly friendly manner and his frequent inspections of his shoes.

- - + - -

Oh, beg pardon. I seem to have ambled off into the realms of fantasy for a moment....

Y

Why?

Market competition plus the fact that this always seems to be the case, judging by the comparisons at dpreview. Differences at base ISO seem to be overshadowed by factors such as processing technique, raw converter type and compatibility and, in relation to jpeg output, in-camera default settings.

At least, this seems to be the case to me. High ISO performance seems to me to be the frontier where differences are most marked.

Iliah Borg, in this dpreview thread, confirms that the *real* base ISO of the A900 is around ISO 130.

http://forums.dpreview.com/forums/read.asp...essage=29320995
http://forums.dpreview.com/forums/read.asp...essage=29321180

Thanks for all the hard work Nick. It appears that the ultimate quality of output from these cameras will come down to the user, RAW conversion, and the lenses.

He's talking about RAW headroom. To a good approximation, all DSLR's leave about a stop more highlight headroom than the ISO standard for metered middle gray relative to saturation, very likely to compensate for the abrupt response cutoff at saturation of digital sensors instead of the smooth rolloff that film has. In such a case it makes sense to leave some more headroom for properly rendered highlights, at the expense of a bit more noise in shadows. The A900 is not special in this regard, so if you like, according to this standard, ALL DSLR's overstate their ISO by around one stop.

Hi Nick-
Thanks for your A900 preview and these test shots... When you looked at other shots, was there still the odd clumpiness in the A900 files? I'm looking at moving over from the 1Ds mkII, mainly for the ergonomics (so much lighter, much better control placement), Zeiss glass (the Canon 24-105 is an OK to good lens, the Zeiss 24-70 a great one) and in-body IS (IS on lenses where Canon's version dosen't have it) ... The extra 8 million pixels are nice, but they're a third or fourth reason to consider the Alpha, not the decision driver. However, that clumpiness could be a deal breaker compared to my very smooth Canon...

-Dan

Dan, the following contains a few samples. Please be warned that these are all a bit premature in judging what the camera is capable of, since none of the support for the images are anywhere close to being "mature" and fully supported products yet:

http://www.dpreview.com/gallery/sony_a900preview_samples/
http://www.alphamountworld.com/image-sampl...e-image-gallery (contains some fairly high-res images)

There are some other places where they show higher resolution mindblowing images, but those have some members-only restrictions.

Also, if you can get a copy of the French magazine "Chasseur d'Images" (October Issue), it has a detailed review of the A900 (they are not typically Sony fans), along with the 5DII and so on.

And they gave the A900, 6 stars out of 5, for landscape use ! All other categories were scored as 5 stars.

Their additional verdict: At ISO 1600, the A900 has less noise than the Nikon D3. At ISO 6400, the D3 wins. Of course they did imatest/DXO tests etc., and printed the results at A2 size, while considering their scoring.

Bottomline, anyone intending to use the A900, better be prepared to invest in the latest crop of ultra-high-end lenses, to get the benefit of what it is capable of. Anything less, and prepare to be disappointed, since the sensor resolution is pretty merciless on even average lenses, especially when one moves to the FF borders.

It makes me wonder just how much better the latest crop of ultra-high-end lenses are compared with lenses of yesteryear that scored 4 and above out of 5 on the Photodo MTF tests; lenses such as the standard Minolta 50/1.4, the Sigma 24/2.8 (one of the better lenses from Sigma) and the Tamron SP 90/2.8 which is also one of the better lenses in the Tamron range.

My Minolta 35-105 zoom would be the walk-around lens and is probably no better than the Canon 28-135 within the range common to both lenses. My Sigma 400/5.6 prime, although probably at least as sharp as the Canon 100-400 (and slightly sharper at F5.6 according to Photodo tests) does suffer from obvious vignetting at full aperture and even at F8.

I had my eye on the A900 because I already have a few Minolta-fit lenses. However, should I feel the need to buy additional lenses, the switch becomes an expensive exercise which is difficult to justify.

On the other hand, there are two areas in my Canon lens range which I feel need upgrading, the extreme wide end and the extreme telephoto end. My Sigma 15-30 does not really have adequate edge and corner resolution, and an upgrade to the Canon 100-400 would be much appreciated.

If I'm going to end up buying replacements for these two lenses anyway, they might as well be Sony/Zeiss or some other Minolta-fit alternative, if such lenses exist.

Of course, I won't be pre-ordering any camera till I've seen full reviews and comparisons between all options. There are indications that the 5D2 has less noise than the D3 at ISO 6400 when same physical size images are compared.

I find myself very exited about the next few years because I think we will continue to see several new and improved lenses due to the image quality the current DSLR's are capable of.

Dave Chew

Me too! But I hate buying gear that becomes under-used and sits on my shelf. The 15-30 and 100-400 range are much used by me. The 24-105 completes the range and is also much used..

My immediate concern in considering the purchase of an A900 would be the availability of a better quality 15-30 and 100-400, or similar equivalents. Is there anything of the quality of a Nikkor 14-28/2.8, for example, with a Minolta fit?

I should add, the sample images from the new Zeiss 16-35/2.8 I've seen so far seem a bit disappointing at the edges and corners. Whilst this lens would probably be an upgrade to my Sigma 15-30, it's not clear at this stage if it would be better than the Canon 16-35 MkII on a 5D MkII.

So far as I know the only lens in the ballpark of the 14-28 is the oly 7-14. Unfortunately it doesn't fit on a 20+mp FF body.

It looks as though that might be the case. If Nikon were to announce soon its own 24mp DSLR at a price competitive with the the A900 and 5D MkII, I might opt for the Nikon for the sake of that Nikkor 14-24/2.8.... if the Nikon were also to include video .

Good job I'm in no hurry.

From what I hear, the Minolta/Sony 50mm f/1.4, the 85mm f/1.4 Zeiss, the 135mm f/1.8 Zeiss, the 24-70 f/2.8 Zeiss and the 16-35 f/2.8 Zeiss, performed outstandingly (excellent++) on the A900.

The same beta-testers (Nikon D3/D700 shooter included) had both the Zeiss 16-35 f/2.8, the Nikon 14-24 f/2.8, and also the Canon 16-35 f/2.8II (and 17-40 f/4L).....corner-to-corner, the Zeiss was SHARP (as I would have expected), with the 16-24 range going with a razor thin margin to the Zeiss, vis-a-vis the Nikon 14-24.

Again, remember the 16-35 f/2.8 Zeiss, is NOT a production version and such pre-production versions can be expected to be tweaked further.

Let's hope this is the case. Did you read this thread? http://luminous-landscape.com/forum/index....showtopic=27853

I think any of the Full-frame premium lenses, designed/built over the past couple of years, have been benchtested to perform well on ultra-high pixel density (Digital) Full-frame sensors - specifically the ones coming out of the armoury of sensor makers like Sony or Canon.

The "Digital" aspect needs to be stressed, due to the different requirements of the lenses optimized for a digital sensor, as opposed to a film-era design.

During the A900 press conference, Yoichi San, the guy in charge of the Sony/Alpha Camera division's engineering group, stressed the point (with graphs etc) that showed that the latest crop of Zeiss and "G" lenses can out-resolve Full-frame sensors having over THRICE the resolution of the 24.6MP A900 sensor. So unless they cross that threshold, these lenses should be more than sufficient, even though the prior film-era lenses may have to be shelved.

I think this is one of Sony's key competitive advantages (along with the Zeiss AF lenses in the range), over other competitors having a lot more wider range of film-era lenses, since most of the Sony lenses are optimized for (and built/designed) around high-pixel density digital FF sensors. The Nikon 24-70 f/2.8, the 14-24 f/2.8 etc are obviously new designs from the Nikon range and almost certainly Canon would be updating their range too.

In my opinion it is the exact opposite. Pre-production lenses are typically better since they are designed by higher precision equipment in small series. The key challenge for high quality lenses is to maintain that accuracy accross a massive production line.

Cheers,
Bernard

The Minolta User Club has now a new website focussed on the Sony system with infos around the Sony DSLR's:

http://www.photoclubalpha.com

Yes, I read that thread. That poster (David K), seems to have shot a picture with a prototype (one of the SEVERAL prototypes) and even more controversially, seems to have even published it. He actually was allowed into the Sony introduction event in Edinburgh, after snagging a press-pass.

This lens would not even be produced (after finalizing and fine-tuning the performance of the prototypes), for several months (retail release date being January, 2009).

I would say he did a grave faux paus in publishing samples from the specific prototype, he managed to lay his paws on ! If I am Sony, I would not let this fellow into any future event of mine !

Actually, this was not even a pre-production sample. This was one of SEVERAL prototypes, shot through a drab hotel glass window. In other words, the "properties" of the lens are yet to even be finalized, since the production versions would only be released in January, 2009.

Surprised that Sony has not controlled their pre-public event better, with tighter agreements with the participants.

Yes, the above mentioned site has some good information and is managed by David K.

A similar US-based site is http://www.alphamountworld.com, managed by Carl Garrard.

But the very BEST site for anything related to the Alpha/Sony mount, bar none, is http://www.dyxum.com.

dyxum has specifications and detailed reviews of every lens, every camera body and every flash ever sold in the Sony/Alpha mount, with thousands of user reviews on all of them. dyxum also has links to all other Alpha related sites.

It also has a VERY well moderated forum, with moderators from the US, from the UK, from New Zealand, from Australia, from Sweden, from Poland etc., and is a VERY well managed and respectful community of Alpha-mount users. Registration is also free.

Thats a software problem not a hardware problem. The A900 is supposed to have a NR off setting. Sensor pixels dont clump.

I would expect ACR to make a better interpretation of the Canon over the unreleased Sony at this point. The question is why would ACR create lumps from pixel data? Noise I can understand. Not lumpy pictures. ACR did the same to the A700 which gave it a bad rap when it was new. They cleaned that up.

The latest Sony firmware on the A700 is supposed to make a big improvement as well. That technology should be in the A900 firmware.

The situation is looking very promising regarding both new lenses and new camera offerings.

I notice at dpreview that Carl Zeiss has launched a new version of its famous 21mm Distagon with a Canon mount, available also at the end of the year. Seems like this is going to be an expensive Christmas for camera enthusiasts.

With HD movie capability on the 5D2, that new Epson HD projector (TW-5000 or 7500) with a claimed dynamic contrast ratio of 75,000:1 might also prove to be irresistable.

I get the impression that the Distagon 21/2.8 is unbeatable for corner to corner sharpness. It's certainly noticeably better than the Nikkor 14-24 at 21mm, using a 1Ds3. Check out this review at 16:9 http://www.16-9.net/lens_tests/nikon1424_2...1424_21mm1.html

If the Sony 16-35/2.8 proves to be no better than the Nikon 14-24/2.8, then it's likely I will be leaning very heavily towards the 5D2 with Distagon 21/2.8 prime, especially considering the HD video capability of the 5D2, which I would definitely use.

On the other hand, these wide-angle lenses do not have IS or autofocussing (the new Distagon by design and the Nikkor 14-24 no doubt disabled when fitted to a Canon). The anti-shake sensor of the A900, especially if there really can be as much as a 4 stop latitude, is very appealing.

That's really great news. We can all only benefit from the A900 being a top level entry!

Cheers,
Bernard

Yes, the introduction of the legendary 21mm Distagon will be a great thing for the Canon users, who have not had a truly great wide-angle available, that was natively made for the mount. As long as you don't mind working with MF, I think it would be an excellent choice and obviously if the objective is shooting landscapes, having AF may not be a big priority.

Whether the A900 or the 5DII, the resolution from either of these is just mindblowing and truly require great lenses to really take advantage of.

Based on JPEG samples from both products, I would strongly lean towards shooting RAW-only with either of these, if the objective is to get the best results from them. I personally only shoot RAW.

Incidentally, here is a hand-held ISO 1600 shot from the A900, using the Carl Zeiss 135mm f/1.8, shot at f/1.8, 1/60s:

Very nice, is she a friend of yours?

Cheers,
Bernard

Ha, ha not really. This image was posted on dpreview and I just linked to it.

I haven't tried all these lenses on the A900 but did try the 85/1.4 and 24-70/2.8 on Saturday. Having pixel-peeped and printed the results I can confirm that no one is exaggerating their quality. Incidentally, the Sony branded 70-200 I tried produces stunning results on the 900, both at the wide and long ends. I only tried it wide open, so was quite surprised.

Don.

Y'know what, I used the 14-24 Nikon in a D3 review last month and was rather underwhelmed by it. It had much better corners than other equivalent lenses but the sharpness at f5.6 - 11 was no different to my 17-40, at least not noticeably so in real world shooting.

I've seen the tests and read the threads but my own experience was slightly different. I suppose 'my mileage varied'.

Your experience could be due to QC variation, Nick. However, the new Zeiss 21mm Distagon for the Canon might be a better option.

Hi,

The shot on DPReview is taken with 24-70/2.8 at f/8.0, I would still argue that it is significantly less sharp at the edge then at the center. Just follow the rails from the center to the right and left edges.


http://a.img-dpreview.com/gallery/sony_a90...dsc01990_dw.jpg

Erik

This shows how benchtests sometimes can differ from real world shots. I see the same edge softness on many of my own test shots on the same lens. Not too bad, but certainly visible.

OTOH...

Photozone

shows this lens as being very consistent across the frame at F8.

Confusing...

I can't say I did specific "chart type" testing but I did shoot a couple of shots at 24mm and 70mm to look for distortion on the 1st 24-70 into New Zealand and those shots were evenly sharp right across the frame. I guess we're not really going to know how consistent this lens is until there are many production lenses in use.

This disagreement reminds me of how people talk about the Canon 24-70L. Mine was excellent, one of my all time favourite lenses in fact, but there seem to be quite a few that aren't. From my small test, the Zeiss had sharpness to match the Canon and better distortion control but I can't say about the contrast because the camera was set to produce very flat jpegs when I tried it. What I did like was that the Zeiss does not extend to go wide like the Canon and, from memory of my Canon, it seems a bit more compact.

Photodo uses a crop camera to test (A700), so the corners are never tested. Same issue with Nikon lenses. No sure about Canon.

bob

Hi,

Were you testing thae 24-70/2.8 on an Alpha 900? Any other Alpha DSLR is APSC format and does not utilize the full frame.

It's really a qustion what is the state of the art in "normal zoom" lens design. Some of the new Nikon lenses show that the obtainable quality is is pretty high.
The 14-24/2.8 seems to be impressive and so does the 16-85 APS zoom.

Best regards
Erik

Yes, I was testing on the Alpha 900. Sorry, I should have made that clear. I tried a number of lenses that day: 85/1.4; 24-70/2.8; 70-200/2.8; 35mm/1.4; 16-80DT and 11-18DT (the last two were decidely weird to use as you see a vignetted picture in the viewfinder, with faint corner guides, but the photos work like a crop sensor, just fine). The four big lenses were all top quality but that sensor is quite brutal - if you miss focus it stands out hugely at 100% on a computer screen, if not so much in print!

Nick,
As Bob McCarthy mentioned, Photozone test all lenses with APS-C size cameras. Nevertheless, I suppose one can extrapolate the results and not be far wrong, but that's not ideal. Some lenses can have a very flat response to about 15mm from centre of frame, then take a steep dive towards the corners.

Yes, re-read the specs on the Photozone site, I missed it before.

Big difference between 22mm across and 36mm. The DPR Lens Reviews show the difference clearly.

There are some posts in a different Sony/Alpha website, where the 24-70 f/2.8 Zeiss clearly shows corner-to-corner sharpness on the A900. Consistently.

Here are some additional A900 + 24-70CZ pictures:

http://www.dyxum.com/dforum/forum_posts.asp?TID=36184

Another A900 + 24-70 f/2.8 image (slow download since large image but don't see any softening that was mentioned above and is sharp across the 36mm frame):

http://74.86.43.122/a900/2008-09-10-123927-09790.jpg

I think there is nothing more to say, and this is a slap for canon. No way any Canon L zoom will produce such corner to corner sharpness. Kinda sad.

I must be missing something. I don't see any sharp corners in any of these shots.

Most of the images have too shallow a DoF to be sharp in the corners and the image of the building at F6.3 is reduced resolution and doesn't have any fine detail in any of the corners. In fact the resolution of the F6.3 shot, when opened on my computer, is only 17MB, less than the file size of a 6mp camera, never mind 24mp.