D800 vs D4 at very high ISO

[arthurking83]

I'd downloaded some sample images from Imaging Resource from these two cameras, but being busy for the past few weeks I hadn't had much time to compare them.

So I did a quick'n'dirty test using CNX2.

CNX2 reveals some interesting insights into the workings of the two cameras.

1. at ISO 12800, CNX indicates that the D800 uses a lot of NR (an intensity value of 20) on the images when NR in camera is set to 0(no NR), and that by comparison the D4 only sets NR intensity to 6. This is out of a max value of 100. You can reset this on the NEF file down to 0 manually or by disabling the develop version of NR. (there are a few ways you can use NR in CNX2!!).

2. Once the NR is reset to zero, the differences become obvious. So I did the Tannin way of comparing the images properly and cropped to a certain point which was as close as possible without getting to tedious) and saved each of images at the same pixel resolution. Once the images are normalised, the D4 beats the D800 by a comfortable margin!!
The main area of betterment is in colour rendition of the D4's image(at least in these sample images) and if you look closely there's better resolution in the D4 images when viewed at the D4 max resolution(100% pixel view) even compared to the down scaled D800 images.

D4
D4_crop.jpg
D800
D800_crop.jpg
original raw images are courtesy of imaging-resource with the only intent being to show how the images react to Nikon software processing(or unprocessing!)

All I've done to these images was to
1: remove all processing from the camera. Reset the sharpening in the standard picture control setting from 3 to 0 for both images.
2: disabled the in camera NR completely. Even tho the in camera NR is set to zero, some NR is still applied to the raw files.
3: crop to a similar point of view and resize to the same pixel value, and then save to jpg(FSViewer was used to resave to jpgs less than 250Kb in size)


Other folks using other software may or may not see these results, but this is how Nikon's software works... so this is why I've posted these images. I was curious to see how much better the D800 is compared to the D4, and as I'd expect to see in 'reality' the D4 images are better at high ISO, even taking into consideration the more than 2x higher resolution from the D800.

Haven't even considered comparing the base ISO images, as the results should be a no brainer .. D800's resolution will always win out.

I'm still using my D300, and taking that into consideration the D800's ISO12800 quality is superb! .. it's just that the D4's ISO 12800 is superberer
There is still some colour noise that is easily removed from the D4's raw files using a much less aggressive NR routine in CNX2, eg 7-10 depending on level of tolerance and required image size. Whereas the D800's file needed a much higher value, and possibly greater loss if detail in doing so.

Theoretically speaking you could max out a D4 file at a print size of about 17" with a quality of print set to 300dpi, whereas a D800 image could go on to over 24" print at the same print quality setting.
So it's a given that the D800 will always offer the ability to print larger, but from seeing these images and the possibilities available in processing, I'm not so sure that the D800 offers such a greater opportunity when light levels get low and ISO has to increase too much.

The best way to see the differences is to download the images and compare them 'over each other' viewing them in your favourite editor/image viewer.
It's not immediately obvious, but there is better detail rendition in the wine bottle label in the lower RH corner(the statue with black background) in the D4 image as the cracked black background is rendered nicer and with less colour noise than the D800 image.
There's also slightly better detail rendition in the cloth swatches in the upper LH corner.
Once each image is viewed directly against the other, the much more vivid colour of the D4's images is blatantly obvious, where the D800 looks more washed out by comparison.


ps. D4's ISO 204K images are nothing special(as you'd expect), but the fact that there is any detail in the images is still quite amazing anyhow.

[I @ M]

Thanks for the time and efforts AK.
This looks to be the better way of comparing apples to grapes than simply having two different megapickle yield camera bodies presented with straight 100% crops of their respective outputs.

The only thing which grabs my attention though is that the D4 image appears to be slightly less exposed than the D800 image. I know that if using two differing models ( or even two different bodies of the same model ) it would be hard to have them expose perfectly identically but with the ability in CNX2 to be able to raise / lower exposure via a slider in .01 graduations it should be sort of easy enough to match the two images for intensity to a closer degree than in theses shots.

If one were to do that I feel that the D4 would look slightly better again against the D800 or are the slightly deeper blacks in the D4 masking some of the chroma noise in areas like the bottle caps?

[jim]

I'm still using my D300, and taking that into consideration the D800's ISO12800 quality is superb! ..

That was my first thought on looking at the pics.

[TEITZY]

Thanks AK. D800 is holding its own though a bit more chroma noise to deal with.

Here's a comparison at ISO6400 & ISO100 between the D4, D3, D800 & D3s. If you are lusting after a D800 and can't afford it, whatever you do DON'T look at these samples

http://www.fredmiranda.com/forum/topic/1098515

Cheers
Leigh