A900 and D3X studio comparison

[David Kilpatrick]

This test actually involved four lenses (two Sigmas, Min 50mm f1.4, Nikon 50mm f1.4 AF-S G), two cameras, and several sequences of shots with adjusted settings using both RAW and JPEG, and was done at 1/3rd stop intervals from f1.4 to f4 then full stops to f/16 and f/22 (depending on the lens used). The amount of data in terms of image size is just too much to present properly via pBase, though I have learned a lot from it - including that I am very happy to use my A900 and 50mm f/1.4 Minolta AF in these conditions.

First, the Sony image at f/8, seen small:



http://www.pbase.com/davidkilpatrick/image/109053091

Visit the pBase page to view the level 12 JPEG at full size, it is over 15MB and takes a seriously long time to view or download. Read my notes there - the Sony is at least one stop less sensitive than the Nikon at ISO 100 given identical ACR processing and carefully matched exposures.

Next, the Nikon image, which will look superficially more attractive since the higher saturation and warmer colour balance, at the tungsten fixed setting, have that effect:



http://www.pbase.com/davidkilpatrick/image/109053545

I remain unconvinced by the DxOMark etc view that the D3X has the greater dynamic range. Even when more accurately density-matched exposed raws are worked on, the Nikon image clips at one end of the histogram or the other, while the Sony does not. The wool ball labels on the Nikon image are prone to a colour shift and slight burn-out where the Sony perfectly handles this subtle bright detail. The Nikon has a much more consumer-camera like contrast and saturation, and the JPEGs also reflect this (I set up the raw processing parameters to get as close to the JPEG as possible).

I'd love to post the entire series of shots including the lens comparisons and all the apertures but it's just an exhaustive amount of stuff to process and upload (and my pBase account is expensive enough already).

David

[InTheSky]

A900 + some Minolta lenses =

Thank you for the test David.

This is a very good thing that Sigma design this beautiful 50mm too, now we have a tool to really compare both sensor in same condition. (flare, coating, resolution ).

Do not post too much pictures, on my side, i have learn to trust your words.

Regards,

Frank

[InTheSky]

Ok ... I'm guilty, I have done some pixel peepering .

You remember when you told me that my 28-75D was not good ... , but when I'm looking at this lens a F8.0, i was expected better result corner to corner. The middle is sharp, but not the end (for a 50mm ... at F/8.0). http://www.pbase.com/nadeauf/image/1077 ... iginal.jpg(28-75D, my copy has problem on the left side, but look not so bad compare to the sigma).

The Nikon looks like resolving better result detail on the money piece, but not on the wool (the Sony looks better). But again, i think Sony is the winner for using the best of the lens. Clearly the Nikon have difficulty with angle light than the Sony. In fact i have found that the picture taken with the Nikon look a little bit more zoom on the subject. Making effect that there is better resolution i think.

Funny things ... i don't know if it is my cheap Dell Latitude from my job , but the dark area looks more pleasant on the Sony than Nikon.

OK ! I stop .

[Greg Beetham]

DXO's system seem to be based on theoretical simulations, Don seems to think they are gospel, but I still have my doubts.
David what might be useful is if you had a control image alongside those two shots of what the eye can actually see, no easy task though as the human eye even with it's limitations has a much larger dynamic range than film or a sensor. Probably though that might not work either as the eye wouldn't capture those shadows depthwise that are probably only there during the flash, and the whole thing would most likely look a lot flater in tone etc. But still might be useful from a colour/tone perspective. Mind, I'm only speaking from an image consumer standpoint, not so much a technical one.
As far as the images themselves go it probably depends on individual taste which is more appealing, too me for example the A900 one looks very slightly underexposed and muddy, while the D3 looks too my eye is very slightly overexposed and tonaly artificially brightened, but the D3 has much better detail in the shadows, but then the A900's shadows would probably be a snap to lift a tad anyway in PP, so in the end it's probably comes down to picking nits...basically.
Myself, I would be perfectly happy with the A900's output...being sort-of biased towards reality as much as possible.
Greg

[Dr. Harout]

If I had to pick one, it would definitely be the first, though the second is much brighter. But looking at familiar colors, in the first they are more natural (such as the wood structure and color); the second shot is a bit overexposed I think.
David, is there a way to compare both with identical exposure, i.e. both pictures to have the same amount of light/exposure and then compare. Would reducing the brightness in the second shot lead to a better color rendition?

[aster]

Thanks for the comparison David.

I find the first photo by a900 more 'real' with truely out-of-life colours while the second is looking artificially bright and over-saturated. The shadow fall-off on the guitar-clock's contours is very nicely rendered by a900 while it's almost lost in the second photo.

Yildiz

[David Kilpatrick]

Some comments - first of all, there is a slight difference in focus point despite great efforts to check on the screen that is identical (both cameras were set to manual focus on the lens after focusing, and locked, the Sony definitely held it MF position - maybe the Nikon did not). Secondly, the three-D subject with various degrees of out of focus area does not agree much with the curvature of the Sigma's plane of focus at this distance. On distant or flat subjects it is much cleaner. Finally, no matter what exposure is given to the Nikon or what processing adjustments are made (apart from artificial ones like fill light and highlight recovery) the Nikon file is not capable of matching the Sony rendering (it is impossible to pull the highlights back in without getting colour casts) but the Sony file is very easy to make look like the Nikon, just change to a Strong Contrast tone curve, add +0.3 stop exposure shift in raw, pull the black point up, change the WB and increase saturation.

The left-hand ball of wool is completely wrong in the Nikon shot. That is not the colour. The Sony colour is dead accurate - extremely accurate.

But I suspect most people would pick the Nikon shot and say it looks better. They would not be aware of the colour shifts, or the loss of detail on the wool labels and other bright areas.

The tripod had to be adjusted between the shots, because the Nikon body is so much thicker between the lens and tripod mount. This made it almost impossible to match the view perfectly, despite measuring the gap and rising the centre column the same amount.

David

[Greg Beetham]

Have a look at the peacock feathers too, the long tail feathers are slightly better from the D3, that's something like I remember them in bright sunlight, (how much trouble would it be in PP to fix the A900 one's...probably not much) but the light tan (D3) feathers are a bit sus, David will be able to confirm how accurate that colour is, I suspect the A900 grey brown will carry the day there.
Greg

Sorry David I was in the midst of posting, my post may no longer apply.

[David Kilpatrick]

Because the results are not really fair to the Nikon using identical processing parameters, I have replaced the conversions and the captions on pBase.

Both raw files are now set to Brightness 45 (this is actually the default I use on Sony files anyway) and the Nikon was set to -1 EV exposure correction, the Sony to -0.3 EV. The only remaining correction not done would be to cool down the Nikon colour balance. I found that the Alpha 900 consistently produces a very accurate tungsten (or other artificial light) WB compared to the Nikon D3X, which tends to undercorrect in all conditions. This may be deliberate - undercorrecting keeps the 'feel' of the lighting better (so tungsten looks warm) and introduces less noise.

Because the original post was linked to the pBase files, the results will now have changed visually, and I'll edit the first post to ensure there is no mismatch.

David

[Dusty]

Fanboys will be fanboys....lol.

-Sonolta

And guys who believe one lab and one lab only will be .... annoying.

This is a test to show what our eyes see, not what some test lab's faulty numbers show. This is photography, not image science. If the human eye doesn't agree with the lab results, then that just shows that the lab's model isn't right. After all, no one has created an eye-brain link yet except God, and He didn't bother to write the specs down where we can find them.

No matter what the lab may say, if the human eye finds one more pleasing than the one the lab says is technically superior, the lab loses.

Dusty

[Dusty]

The a900 Raw plots look worse at full rez...these are normalized at 8MP.

-Sonolta

I'm gonna race my VW against your Ferrari. By the way, we've 'normalized' them both down to 50 horsepower!

di Vinci's version:


'normalized' version:


Any other silliness you want to engage in?

Dusty

[Dusty]

You are a clueless fanboy idiot who clearly can not read! I have posted info from FOUR LABS to this thread! Popular Photography, Imaging Resourse, DXO, and Diwa. Get a clue you moron.

Actually, you post 2. The other 2 just copy DXO Labs. An authoritative resource is someone with original research, not someone who quotes someone else.

Stop ignoring the picture's and testing done by REPUTABLE labs, test sites, and reviewers. NOT SONY FANBOYS THAT OWN SONY MAGAZINES AND WEBSITES!

Well, I don't own any Sony magazines or websites. I'm also not saying that Sony is the best at everything. As I've stated before, I believe Sony is the clear price/performance leader, and their DSLR products meet me needs.

You are as dumb as they come.
I will repeat, you are as clueless as anyone I have ever met.
You are a moron plain and simple.
Get a clue, and a life, and please stop telling lies and peddling your nonsense around the forum.

-Sonolta

Honestly, Don, I'd like to debate you on your level, but I just don't know how to stoop that low!

Dusty

[Dusty]

Dusty clearly overlooked the Diwa Labs results as well...here they are again:

-Sonolta

I saw those, Don, but even being blind in one eye, I noticed the DXO watermarks on the results!

Dusty

[caporip]

good to see you learnt some restraint in your absence Sonolta..............

[Dusty]

You must be blind in both eyes....DIWA does the tests with the camera and a lens using the DXO Anylizer you freaking goofball!

http://www.diwa-labs.com/wip4/detail.epl?cat=8009" onclick="window.open(this.href);return false;

FTR almost every manufacturer, review site, and lab on earth uses the DXO Anylyzer for gosh sake. BTW...you keep skipping the image resource studio images as well as the Popular Photography conclusions as well.

Get a clue!

-Sonolta

Don, I believe I can see more with one eye than you can with 2. The title of this thread is "A900 and D3X studio comparison", yet your above link shows nothing about a D3X, it shows Canon's 5D Mark II.

The same for your Image Resource link, it's comparing the Sony to a Canon.

At one time almost everyone thought the world was flat, what does that prove? If I use a test suite that doesn't give real world results, and everyone else does, too, does that make us all right or all equally wrong?

Dusty