The myth of 14 bit vs 16 bit (Pentax 645D vs Hasseld H4d-40)

Jason

There have been many discussions on the image quality of 14 bit vs 16 bit captures, but no one seems to be able to give any definitive or authoritative answer.

In order to help to answer these questions, in the spirit of dymystification, we are going to test two medium format cameras. One is Pentax 645D, which uses 14-bit capturing. Another one is Hasselblad H4D-40, which uses 16-bit capturing. Both cameras share the same Kodak 40mp KAF-40000 CCD sensor. It makes them to be the perfect candidates for our test.

To rule out the influence of using different lenses, in this test, we will use the same Hasselblad 100mm CFi lens.
Stopped down to f/8.
Shutter speed 1/125 sec.
ISO 100.
White balance 5000k. Tint +14 lightroom.
Default unsharp mask in lightroom.

Jason

100% crop on highlight area:

Jason

100% crop on shadow area:

Jason

Comparison on micro contrast:

Jason

EV +1


EV +2


EV +3


EV +4

Jason

EV -1


EV -2


EV -3


EV -4

Jason

EV +3 (100% crop)


EV -3 (100% crop)

Jason

The left hand side is H4D-40, the right hand side is Pentax 645D

Some points to be made:

1. The color rendition between H4D and P645D is very different. It is without doubt the the Hasselblad produces more pleasing color out of the box. But it doesn't mean that Pentax 645D produces very bad color. On the contrary, 645D produces better color to most (if not all cmos sensors). But Hasselblad is really the industry leader in producing the best color in images.

2. While the two images were taken under sunlight, with about 10 minutes apart, we could see that the sun was changing its direction. (We can see different shades in the pink building in the first image). I carefully selected some areas for 100% crop with show little or no light difference between the two images for close inspection.

3. From the second image, we can see how the two cameras deal with extreme highlights. The difference between H4D and 645D actually is very small, though not indistinguishable. Pay close attention to the red arrow areas I made on the image. The H4D clearly shows better details on those extreme highlights. But please bear in mind that, some details are lost in these jpeg files on flickr. Pentax 645D actually IS ABLE to show extreme highlight details, though marginally not as capable as H4D.

4. Similar results can be observed in the shadows. But again, in these jpegs, the shadow details become even harder to see. But even without going into the details, H4D image clearly shows better clarity in shadows.

5. The Hasselblad also demonstrates better clarity and micro contrast throughout the whole image, even though both H4D and 645D share the same sensor. This is obvious in 100% view.

Jason

Some more thoughts:
If this test can tell any meaningful information, I would come to say that apart from color and micro contrast, the difference between 14-bit and 16-bit capturing actually is very small, particularly in terms of dynamic range. There is a visible difference in 100% view, even in 8 bit jpegs in consumer grade monitors, like iMac, but it won't be significant if the image is properly exposed.

If we look at the +3EV of the 100% crop above, we can see that H4D40 (left hand side) was still able to deliver the texture of ICC building even with 3 stops over exposed in extreme highlight. But in the Pentax 645D example, the texture becomes quite invisible.

For shadow areas, it becomes clear that 8-bit jpegs lose many details. In the 100% view of -3 EV above, H4D was able to show a faint shape of an air-conditioner (red arrow). It is clear in the RAW files. Unfortunately, in the jpeg i showed here, it becomes quite hard to see. Again, in Pentax 645D, even in the raw files, the shape of the air-con becomes quite hard to see.

Interestingly enough, actually one does not need to go to 100% view to see the (however very small) difference between the two files. In the +1/+2/+3 EV examples, it is clear that the 645D files are more "wash out". In the -1/-2/-3 EV examples, the 645D files are "duller". It does reflect there is a quality difference between bit-depth. 16 bit maintains higher fidelity.

Another point to be made is: the color difference between the two cameras is big. While I suspect it is the result of different rendering, it can also be a result of 14-bit versus 16-bit. But color, as I mentioned above, is something very hard to judge objectively. As far as I know, Pentax 645D produces better color with original lenses. This test, which uses a third party lens, may not be doing justice to the camera (or both cameras).

Conclusion:
With the huge price difference between H4D-40 and Pentax 645D, I would definitely say the Pentax is a great buy. It is a major upgrade from any DSLR (clearly better color). The difference in image quality between 14 bit / 16 bit is very, very small (though not indiscernible), at least for using Adobe softwares. The reason for buying Hasselblad should be better color rendition, handling, focusing ability and collection of lenses, etc.

If someone says Pentax 645D is bad "simply because it is 14-bit", ummm..... I think you may very well ignore that statement...

M3fans

Much appreciated. Thanks!!

YKCC

Thank you!

Jason

you are very welcome la, M3fans and YKCC hing

C9.99

Jason呢d就叫專業，我M9/M240個比較hea好多！

ghost03032008

C9.99 發表於 2013-8-20 12:56
Jason呢d就叫專業，我M9/M240個比較hea好多！

I like your testing model....

Jason

C9.99 發表於 2013-8-20 12:56
Jason呢d就叫專業，我M9/M240個比較hea好多！

i agree with ghost!!

grandetech

Although I am not into medium format yet, this testing is very well organised and easy to understand, and particularly the testing aspects are up to industrial standard, thus the ability of the cameras are fully revealed.  

Hope to see more testing in this stern level, there are other testing by other members which are just as welcome but not as for reference, thanks Jason.