Pentax 645D issues and questions

[BillOConnor]

Has anyone heard if the Pentax 645D made its appearance in Japan? Or Poland? Or Spain?

One issue that worries me is Pentax's decision to only go with 14 bit color. Now this may be a non-issue, Kodak may be more conservative in their
ratings than Dalsa. Many of the Leaf and Phase One backs that advertised 16 bit color were really 15 bit according to Dxomark, so if the Kodak
chip is really closer to 15, then it could be moot. But make no mistake, there is a huge difference between 14 bit and 16 bit color. Do the math.
I guess we'll have to hurry up and wait and see.

[Peter Stacey]

Has anyone heard if the Pentax 645D made its appearance in Japan? Or Poland? Or Spain?

One issue that worries me is Pentax's decision to only go with 14 bit color. Now this may be a non-issue, Kodak may be more conservative in their
ratings than Dalsa. Many of the Leaf and Phase One backs that advertised 16 bit color were really 15 bit according to Dxomark, so if the Kodak
chip is really closer to 15, then it could be moot. But make no mistake, there is a huge difference between 14 bit and 16 bit color. Do the math.
I guess we'll have to hurry up and wait and see.

According to this on the Pentax Japan site:

http://www.pentax.jp/japan/news/announce/20100512.html

the release date was put back from 5/20/2010 until 6/11/2010 due to higher than expected demand and a desire by pentax to meet the expectations of all pre-ordering customers.

So it seems that they received more orders than they could fulfill by the 20 May date, so the release has been pushed back to 11 June.

Regards,

Peter

[BernardLanguillier]

Has anyone heard if the Pentax 645D made its appearance in Japan? Or Poland? Or Spain?

One issue that worries me is Pentax's decision to only go with 14 bit color. Now this may be a non-issue, Kodak may be more conservative in their
ratings than Dalsa. Many of the Leaf and Phase One backs that advertised 16 bit color were really 15 bit according to Dxomark, so if the Kodak
chip is really closer to 15, then it could be moot. But make no mistake, there is a huge difference between 14 bit and 16 bit color. Do the math.
I guess we'll have to hurry up and wait and see.

I believe that are you are confusing 2 different things

1. The DR of the sensor, measured in stops represents the range of illuminations that are neither fully blown (highlight side) nor impossible to distinghuish from noise (shadow side),
2. How many bits are used to sample this continuous range of illuminations.

It is possible to cover a wide range of illuminations with few bits, but, depending on the spatial resolution, that may result in visible banding in the image.

DxO tests measure #1 and have found the very best back to be around 13.7 stops if I am not mistaken.

Now, independant studies show that even on the backs using a 16 bits processing pipe, there is no useful information in the lower 2 bits, meaning that none of the backs today are actually more than 14 bits.

There is therefore no reason to think that the 14 bits pipe of the Pentax will induce any limitation whatsoever compared to products from Phase of Leaf.

Regards,
Bernard

[BillOConnor]

I believe that are you are confusing 2 different things

1. The DR of the sensor, measured in stops represents the range of illuminations that are neither fully blown (highlight side) nor impossible to distinghuish from noise (shadow side),
2. How many bits are used to sample this continuous range of illuminations.

It is possible to cover a wide range of illuminations with few bits, but, depending on the spatial resolution, that may result in visible banding in the image.

DxO tests measure #1 and have found the very best back to be around 13.7 stops if I am not mistaken.

Now, independant studies show that even on the backs using a 16 bits processing pipe, there is no useful information in the lower 2 bits, meaning that none of the backs today are actually more than 14 bits.

There is therefore no reason to think that the 14 bits pipe of the Pentax will induce any limitation whatsoever compared to products from Phase of Leaf.

Regards,
Bernard

thanks, Bernard, that comes as a huuuge relief. I photograph paintings and need every "bit" of color possible and would rather not go to a scanback, because I work only on location--making a continuous light source impractical. The Pentax seems like such a great fit.

[JonathanBenoit]

Now, independant studies show that even on the backs using a 16 bits processing pipe, there is no useful information in the lower 2 bits, meaning that none of the backs today are actually more than 14 bits.

There is therefore no reason to think that the 14 bits pipe of the Pentax will induce any limitation whatsoever compared to products from Phase of Leaf.

Regards,
Bernard

Except how do you know that the lower two bits on the pentax arent the same way?

[Doug Peterson]

DxO tests measure #1 and have found the very best back to be around 13.7 stops if I am not mistaken.

Now, independant studies show that even on the backs using a 16 bits processing pipe, there is no useful information in the lower 2 bits, meaning that none of the backs today are actually more than 14 bits.

The dark-frame technology Phase One uses (which is only applied when processing in Capture One and not included in any 3rd party measurements/evaluations as far as I've seen) can have a dramatic impact on the net usable dynamic range and the degree to which shadow noise and tonal transitions are organic and pleasing rather than blotchy and ugly. Having available a proprietary raw processor which is extremely fine-tuned to getting the most (and the best) out of that last stop of dynamic range is pretty huge - just take a Phase or Leaf file and process it with shadows boosted in C1 vs shadows boosted in LR or Aperture.

Try to forget about the numbers because they are very susceptible to marketing, and the parts that aren't require a PHD to 100% understand (the basic idea is very simple but like anything there are more nuances/subtleties/exceptions than is apparent at first glance). Probably the least useful measure of final usable image quality is to the number of bits in the A/D convertor - if you have a great audio amplifier but are playing music from a sub-par CD player or through sub-par speakers the end result is still sub-par. Even different A/D convertors labeled as the same number of bits are not guaranteed to perform their job equally well. There is exactly one way that we will find out how the systems compare and that is real world results - shooting scenes with challenging lighting and pushing around the highlights/shadows to see how much detail is retained, how accurate the color in those areas is, how organic the shadow noise is, and how pleasing the transition from clipped to non-clipped data is.

Moreover, since everyone has a different workflow, style, and needs, the only way to have a test like this really work for you is if YOU do it. If you ever need a Phase One or Leaf back to test you know where to call ;-).

Doug Peterson
__________________
Head of Technical Services, Capture Integration
Phase One, Leaf, Cambo, Canon, Apple, Profoto, Eizo & More
National: 877.217.9870  |  Cell: 740.707.2183
Newsletter: Read Latest or Sign Up
RSS Feed: Subscribe
Buy Capture One at 10% off
Personal Work

[BernardLanguillier]

Except how do you know that the lower two bits on the pentax arent the same way?

That is indeed a fair statement. We will need to check Pentax claims against facts.

It is well known that best in class DSLRs fully tap into their 14 pipe potential, so I would be very surprised if the 645D didn't though.

For now we have samples that appear to be excellent. Besides, since the 645D appears to be Kodak's best hope to make significant money out of MF sensors at this point of time, I am confident that they will have delivered top class support to Pentax to design an imaging pipe fully tapping in the porential of the sensor.

Generally speaking, high end is often assessed by their owners in terms of spec rather than being assessed in terms of results. The best Esoteric D/A converter (the D-05 for those who know it) features a nearly unique 32 bits processing pipe and has been selling like hot cakes in Japan (still 5000 US$ though...). It takes a triple golden ear to hear the slightest different with well designed 24 bits dacs (99.9% of the rest of the market), yet the owner I know swear that he is 2 generations ahead of the rest of the world (as if "he" had anything to do with it).

Cheers,
Bernard

[BJL]

Except how do you know that the lower two bits on the pentax arent the same way?

The uselessness of the last few bits from 16-bit ADC is due to the signal to noise ratio of the sensors themselves being at most 8000:1 about 13 bits worth. Pentax is using the same Kodak sensor technology as many Hasselblad MF backs (the H3D40 in particular) so the sensor noise should kick in at the same bit level (bit 12?, 13?) as with those other MF backs, unless Pentax has messed up badly on processing.

That is, there is no reason to expect an equal number of "bad bits", but instead a good reason to expect a roughly equal number of "good bits": about 12 or 13.

[MichaelChmilar]

It is misleading to speak of the low-order two bits as "good" or "bad". They are simply "noise bits", meaning that they do not hold any meaningful information.

[Doug Peterson]

Generally speaking, high end is often assessed by their owners in terms of spec rather than being assessed in terms of results. The best Esoteric D/A converter (the D-05 for those who know it) features a nearly unique 32 bits processing pipe and has been selling like hot cakes in Japan (still 5000 US$ though...). It takes a triple golden ear to hear the slightest different with well designed 24 bits dacs (99.9% of the rest of the market), yet the owner I know swear that he is 2 generations ahead of the rest of the world (as if "he" had anything to do with it).

One big difference between audio-listening and modern photography is that in photography we often heavily manipulate the source data. In fact the bit-depth of the final outputted image is really not demanding at all. In fact 8-bit is fine enough for 99% of output, but getting to the final outputted image often strains even high bit-depth images. Take for instance a night city scene. You might boost the shadows 2 stops, correct for an overall blue cast, correct for half a stop of vignette at the corners, and then dodge a particular detail a stop and then add some sharpening (increasing the local contrast - and therefore noise or banding - of detail in those data-stretched areas).

A better comparison would be scientific audio analysis / audio forensics or sound recorded specifically for audio FX; these applications share the need to manipulate and push around the data after-capture and before-output.

If you never touch the highlight/shadow sliders, exposure slider, contrast slider, curves, clarity/sharpening slider, lens correction, or use a WB outside of 5500-6500 than you only need a camera with 8-10 clean bits. But as soon as you do even the most rudimentary adjustment of your image you become dependent on the bit-depth and bit-fidelity of the image.

Doug Peterson
__________________
Head of Technical Services, Capture Integration
Phase One, Leaf, Cambo, Canon, Apple, Profoto, Eizo & More
National: 877.217.9870  |  Cell: 740.707.2183
Newsletter: Read Latest or Sign Up
RSS Feed: Subscribe
Buy Capture One at 10% off
Personal Work