Definitive A900/D3X noise/ISO samples

[mdcromer]

The images from IR look comparable once you give a bit of USM to the Sony (radius .3, amount .65). The only area of improvement in the D3x is the round dial crop.

Also you have to be certain that focus is identical with both cameras.

I don't see any amount of improved detail with the D3x at low ISO.

[David Kilpatrick]

The biggest difference is that the Nikon has a slightly steeper midtone section of its contrast curve. This emphasises some detail contrasts which are relatively low in the Sony. There is probably an AA filter difference too but my guess is that the Nikon filter will be prone to more variation in results between lenses. So far this is one of the main positive aspects of the A900 filter which is weaker in effect, but placed further from the silicon - it minimises the losses found with lenses using short rear nodal points (wide-angles and wide zooms without special telecentric design). I have never had a better performance from the 12-24mm Sigma than I get on the A900.

David

[j_p_p]

at low ISOs it's pretty similar...maybe the d3x is a bit sharper and somewhat "vivid" but it's not a Huge diffenrence ...the lens could also provide a part on this ... not a 5000 us$ difference in my opinion.
But of course at this level a small improvment will always cost more

Regarding High Isos...i think most people (i am) are expecting sony to better it via firmware ? pulling some sort of a700 v4 like "firmware magic"

Hope they release someting sooner rather than latter ...

[mdcromer]

David,

Let us know when you get those ISO 6400 D3x 14 bit to 12 bit 5o Alpha 900 comparison images done. I'm very curious if they look more like the 14bit D3x, or like the Alpha 900.

[mdcromer]

Don, why would they look like 14 bit D3x?

The supposition is that the 14 bit files are using a different signal processing pathway for the images (possibly involving 4x oversampling) that accounts for the drastic slowness in 14 bit mode, while the 12 bit files are using the standard hardware that is also used by the Alpha.

It seems very likely to me that the increased shadow performance of the Nikon at high ISO is related to using different hardware / digitization over the Sony.

[Vidgamer]

I've read multiple comments from people who think that 14 bits doesn't make much of a difference (vs. 12). If it did make a difference, then that might account for an improvement, but you may run into system noise that swamps the ability to actually extract 14 bits of true information. And if 14 bits doesn't make a difference, then you should be able to make up the difference through firmware/software. Maybe the truth is somewhat inbetween -- not all of the 14 bits are useful, but there is enough of a difference in the last two bits to be worthwhile, and then perhaps better processing adds to that.

[douglasf13]

The thing is, it's a trade off, because, while Nikon seems to do better in noise control, Sony does better in DR.

[mdcromer]

But the D3 can, in no way, compare for resolution.

[douglasf13]

Trust me, I've seen every test on the web known to mankind. I should have been more clear. I wasn't referring to total DR, but instead the handling of DR. ie. highlight roll-off, which is part of Sony's special sauce, just like Nikon's black clipping. This is why, regardless of test numbers, so many are being drawn to the camera. Granted, I rarely shoot over ISO 800, so noise issues don't apply to me.

[mdcromer]

Frankly Don, I trust David's tests far more than DPR's. 'Nuff said.

If in fact the 14 bit Nikon files are taking a significantly different signal reading and processing path than the standard Sony chip offers, while the 12 bit files use the stock Sony hardware, perhaps that accounts for the much improved shadow detail at high ISO for the Nikon 14 bit files over the Sony (12 bit) files.

[mdcromer]

But the D3 can, in no way, compare for resolution.

LOL MD..let's talk IMAGE QUALITY across the range. Resolution means squat to most folks...and then finding lenses to match the resolution across the frame is another story.

-Sonolta

Well resolution means a great deal to landscape shooters, and (low ISO) stock shooters.

The only reason I am buying a dSLR is to make much larger fine-art landscape prints. Going from my Sony R1 (10MP) to a Nikon D3 (12MP) is virtually a wash as far as print quality is concerned.

As for lenses, I'm quite satisfied with the results I have been seeing with many older Minolta lenses on the Alpha 900 when stopped down to landscape apertures.

If I were primarily an event photographer, I might well be interested in a D3 (although the lack of IS for most event photo lenses is a big drawback!)

[mdcromer]

Those are jpeg dynamic range measurements. A joke. R1 JPEGS are garbage. R1 RAW DR is extremely high. BTW the DPR resolution measurements are also from jpegs.

In real life you can't tell an R1 landscape image from a D3 image -- except the R1 will very likely be sharper due to a lot more DOF from the smaller sensor, or if the scene has such huge DR that the 1 stop advantage of the D3 shows up.

In real life, if you want significantly better images than well exposed, base ISO R1 images of static subjects, you can choose:

1) Canon 1DsIII
2) Nikon D3x
3) Alpha 900
4) Canon 5DII
5) Various MFDBs / cameras

That's it. There are no 10-12-14-15MP cameras that have any significant advantage for landscape shooting over the R1, unless you want wider than 24mm / narrower than 120mm. That's why I waited until now (actually come January) to buy something else.

Oh, and Don, kindly please shove your "fanboy" namecalling horse-shite back in your piehole (or whichever other hole it's gushing out of).

[alphaomega]

Some divided opinions about the Sony DSC-R1 above. I used one for a couple of years before the A700 came out. My son is using it now to photograph ships. He shoots Jpegs and they are all very sharp and with good tonal range.
I bought it in 2005 (I think) on the basis that it produced as good as or better (sharper) pictures than the 350D and it was proofed against dust. A major point for me then and now.
I have many pictures on Alamy taken with this camera and most are nearer 70 than 50 Mb. I found the dynamic range for landscape & architecture very acceptable. Do not recall suffering from blow-outs or clippings, but don't mention higher ISOs. It was simply terrible above 400 ISO or so. With the CZ 16-80 lens it certainly deliviered the goods at 200 ISO. It was a camera that acquitted itself well at the time but has been overtaken by newer models. The electronic viewfinder was poor and the buffer could take one RAW before it slowed down. The LV screen at 2" was too small to be of any use. I was happy to replace it with my A700, but it did the job at the time and quality wise could compete with the consumer DSLRs available at the time - and dust never settled on the sensor. The range was restricted by its 24-120 equivalent lens, but then I used my MD system and Mamiya 645 Pro when the R1 was found wanting.

I do not doubt the technical figures produced by Sonolta but then the R1 cannot be expected to compete with today's DSLRs. I mostly used mine at the wide end so did not really suffer from the tele end problems highlighted by Sonolta.

[David Kilpatrick]

Yes, my reply there was pre-V4 and also pre-AF adjustment which I made to the camera. My A700 results improved, but the A100 still remains almost unmatched (and I will include the A900 in that). Even so, the R1 files (I do not have an R1, but I keep an archive of all original raws shots during tests) are hard to beat for detail when revisited and examined using current raw processors. I do use my A2 but not regularly, it's a pity and it should get more use - the accuracy of focus is incredible considering what was actually involved.

The issues with the R1 had nothing do to with image quality, everything to do with functional response of the camera. The last day I used it was the day that a boatman fell off his boat and had to be rescued by his passengers in the very cold waters of Loch Leven; I was standing twenty feet away, and the R1 responded so slowly I did not get a shot of him in the water. Had I been using any DSLR, I would have done (especially a DSLR with 18-200 or 18-250). That was when I realised you can't rely on an electronic VF camera with off-sensor live focusing for quick shots. The A2 has, if anything, proved better. It has almost shot action now and then. What I like about the A2 is the movie function with fantastic sound quality, and I do take it to gigs.

David

[mdcromer]

There is no significant advantages to using any of the 12MP dSLRs over the R1 for landscape shooting. NB a 10% increase in linear resolution of test charts is not a SIGNIFICANT advantage, esp. given the huge advantage in DOF for the R1. For practical resolution, the R1 holds its own with the 10-15MP crowd very well. The idea that an Alpha 700 will "blow away" the R1 for landscapes is laughable at best, and utterly wrong if you are talking about "kit" lenses. . .

David is quite right, the R1 is slow and certainly not the best tool for moving / action photography. Although I have some great images of my daughter riding horses with it, go figure. . .

As for dynamic range, the point is NOT to get full benefit to the DR in a single RAW conversion. Instead, you can justmdouble-process the RAW when you want to recover blown highlights, and then combine the shots in photoshop. I've never seen ANY RAW convertor that can do an acceptable job with a 10+ stop scene in a single conversion -- they might be out there, but I haven't seen it.

Of course the R1 is not a BETTER camera for landscape than the D3 et al. Neither is it a WORSE camera for that purpose. See my website for some examples, most of them are from the R1, very few of which were shot "in the middle of a sunny day". And given that I've gotten 4 years of use with it without any significantly superior performers for landscapes in the affordable price range until now, makes me pleased as can be. Money well spent. Now I'm moving on to another groundbreaking price/performance camera that DOES offer significant IQ improvements. I could have chosen the Canon 5DII, and almost did because of the liveview and video, but in the end the antishake, higher resolution, better viewfinder, top-notch affordable FF legacy Minolta glass and reportedly better AF won me over.