rumor: Pentax 645D Price: $6500 USD

[EricWHiss]

Sorry if you find this post tech useless, but
What I still do not understand is: why can't we finally have now in 2010 a CMOS MF sensor??
What's wrong with that?

CCD's are still capable of better IQ at base ISO.

[rogan]

CCD's are still capable of better IQ at base ISO.

But image quality is over rated and I am not interested in base iso. I want to shoot mf at asa 1000-1600.

[rogan]

sorry, repost.
Why can't you delete a post here?

[rainer_v]

If I'm working in studio, with flash, I find the digital backs (at least the phase backs I own) to have a lot more detail and a much deeper file to work in post.

But for me it starts and stops with flash in studio, maybe location if there is no continuous ambient light.

Now if I work with any continuous light, especially mixed lighting, I find the exact opposite and having shot the 5d2, next to my phase backs, in these conditions the Canon has at least as much dr or whatever anyone wants to call it and overall is a much easier file to work.  The detail I find even between the Canon and the Phase with continuous light.

At 400 iso I find the shadow noise with my backs to be much more pronounced (and not near as pretty) as the Canon, but the biggest killer for me working with continuous and ambient light and the phase backs is the color proliferation.  They seem to be tuned to pick up every single color in the room and we are always painting out blotches in skin, or contamination from some ambient light, even backlight.

Shooting outside, in ambient light, let's say shade under a tree, the green of the grass just illuminates the subject in yellow/green where with the Canons the skintones actually have skin color.

This isn't a knock on phase or a rave on about Canon, I just believe that different makers tune their systems for different applications and if I was only shooting flash in blacked  out studios  or non people based images like food I'd be fine with the Phase backs for about everything, though I don't work in just one genre or set of circumstances and rarely shoot anything that's not breathing.

Regardless of my experience with multiple cameras, a lot of this comes down to personal taste, or lighting styles, though given all of that, I find it amazing that Pentax and every maker always compares their files to the 5d2 which even at the lowest price medium format system is still 1/3 of the costs.

BC

i am just travelling through the states shooting many buildings in many cities, being asked and budgeted for the best i can do ...
and i took the luxury to shoot several shots side by side with my artek and my canon, comparing after editing which was looking better.
came out what? 50% for each system. of course the other 50% of each system would be usable shots also, these just felt a bit apart for several reasons and i choosed the ones which were looking simply better from each system. most important here are not the back or camera decision, 14 or 16bit, a bit more or less noise or resolution, most important are the lenses. and here the least important factor is the pure sharpness ( although i still like that the rodenstocks are tack sharp even wide open, compared with the schneiders - but the canons are too and shifted it doesnt matter to stop them down because the canon can make exposures as long i ever need ) . everything goes about distortion correction, and here not only about edge correction also about center correction ( if u have long tall buildings in the middle of the frame the difference in this might look as heaven and hell ). and about flare for interior shots or night shots. i talk about big interiors where you even have not to start to put flashes out but have 1000 lamps shining with 5 different light temperatures directly in the lense.
before canon brought out the new 17 and 24mm shifts i wouldnt have dreamed from such 1:1 comparison or at least not from similar hi-end usable results without making handstands in stitching images with the canon system andwhatever exotic shift lenses or shift lense adapters. but now the cards have been mixed again. my results are funny cause they tell me that either i`ll go on to shoot in parallel and carry around the double stuff or i can decide for one of both systems and will end up with the same final quality, doesnt matter with which i`ll shoot.  
but there are so much more factors in this game than just resolution and/or if the back finally can make some nice iso400 results ( which i dont care in any case).
if i would believe that a p/h/a10/50/60/65/100 would bring up my image quality in any visible/usable way either for me ( and my own work ) or for my clients and their needs, in two weeks i would have one.  but i am so less convinced right now of the mf advantages .... but on the contrary i will immediately buy the new 1ds4 just to sell it 9 months later if the 5dmk3 will be out ... ( i hate the big clumsy 1d bodies ).

[BernardLanguillier]

The best test yet of the 645D with lots of samples.

See http://translate.google.com/translate?js=y...sl=ja&tl=en

I'm amazed at the lack of moire in some of the shots.
I don't know how Pentax pulls this off, or maybe I'm missing something.

Good processing I guess.

The jpg image of the first distant landscape shows a lot of moire in the tiles of the facade bottom/right third, but the DNG conversion is much cleaner.

Cheers,
Bernard

[ErikKaffehr]

Hi,

Could probably be done, but DALSA and Kodak makes CCD sensors and Canon/Sony/Panasonic makes CMOS sensors for DSLRs. I don't think there is any real evidence for the superiority of either technology. Read noise may be lower on CMOS (using correlated double sampling and possibly on chip pre amplifiers) but noise is normally limited by shot noise (variation if the number of photons reaching the sensor).

I don't think that either technology is a magic bullet. A bigger sensor collects more photons and therefore has less noise. This does not explain any "magic" differences between DSLRs and MFDBs, the difference should be around one stop.

Best regards
Erik

Sorry if you find this post tech useless, but
What I still do not understand is: why can't we finally have now in 2010 a CMOS MF sensor??
What's wrong with that?

[BJL]

Could probably be done, but DALSA and Kodak makes CCD sensors and Canon/Sony/Panasonic makes CMOS sensors for DSLRs.

And then comes the question of what neither Dalsa nor Kodak makes large CMOS sensors for cameras, even though Dalsa also makes very large custom CMOS sensors (for X-ray equipment and such), and Kodak did make CMOS sensors, but dropped them a while ago.

My guess is that this is because:
a. the main IQ advantage of CMOS sensors is in low light sensitivity (high ISO speeds), and even then only with noise control technology that first Canon and then Sony and Panasonic have developed, but the far smaller Dalsa/Kodak CMOS efforts have not yet,
b. the Dalsa and Kodak large sensor divisions are driven by industrial, military, medical and scientific customers who do not need extremes of high sensitivity except in situations where high sensitivity and great dynamic range are both desired, like some X-ray equipment, and this combination is best achieved with very large photosites, where CCD works reasonably well. Kodak makes sensors of about 50x50mm with very good low signal sensitivity, but they are only 4MP, with huge 24 micron pixel spacing.

Some references to indicate what the Dalsa and Kodak sensor divisions are mainly interested in:
A. For CMOS sensors up to wafer size, and even beyond to 300x240mm (but with slightly visible seams) from Dalsa:
http://www.dalsa.com/sensors/Products/custom.aspx
B. For big, square CCD sensors with really big photosites from Kodak:
http://www.kodak.com:80/global/plugins/acr...ductSummary.pdf

[MichaelChmilar]

The biggest advantage that CMOS appears to have over CCD is possibility of live view.

Having live view for a medium-format digital back could have a significant impact for the users of "technical cameras" such as the Alpa, Arca rm3d, or Sinar ArTec. It would eliminate the need for a sliding back to change between a ground-glass and the back (or the risk of removing the back). It would also eliminate the need for a rangefinder (ie. laser or sonic) or other focusing aid.

Live view might not eliminate the MF SLR style of camera, like the Pentax, but it would provide a second, more precise method of operation with SLR cameras.

[EricWHiss]

B. For big, square CCD sensors with really big photosites from Kodak:
http://www.kodak.com:80/global/plugins/acr...ductSummary.pdf

Wow!  A 50mm x 50mm square sensor with 15 stops DR - sounds perfect!   I'll take one of these for my Rollei 6008  please!

[Doug Peterson]

If I'm working in studio, with flash, I find the digital backs (at least the phase backs I own) to have a lot more detail and a much deeper file to work in post.

But for me it starts and stops with flash in studio, maybe location if there is no continuous ambient light.

Now if I work with any continuous light, especially mixed lighting, I find the exact opposite and having shot the 5d2, next to my phase backs, in these conditions the Canon has at least as much dr or whatever anyone wants to call it and overall is a much easier file to work.  The detail I find even between the Canon and the Phase with continuous light.

At 400 iso I find the shadow noise with my backs to be much more pronounced (and not near as pretty) as the Canon, but the biggest killer for me working with continuous and ambient light and the phase backs is the color proliferation.  They seem to be tuned to pick up every single color in the room and we are always painting out blotches in skin, or contamination from some ambient light, even backlight

Yep that's about right.

At base or low ISO the P30 has much more fidelity / file-flexibility. At higher ISOs the Canon has more fidelity / file flexibility.

That's one reason why Phase developed the Sensor+ system which allows you to gain two stops of ISO without any loss in quality (though at a lower pixel resolution) so that you can retain the enormous file-flexibility and fidelity of the image as you increase the ISO.

Color accuracy always comes at the price of color accuracy :-). Being able to distinguish very subtle variations in color and areas of low saturation is great when doing product photography / art reproduction / landscape but not so great for fashion in uncontrolled lighting environments. Personally I'd rather have the accuracy and retouch out any undesirable accuracy than to have a camera which cannot pick up on subtle colors. I've had more than a few calls from photographers who had dSLRs that could not pick up the difference between two thread colors in an outfit (usually red), two hair shades, two paint colors etc and the client was not happy.

The Skin Tone tab in the Color Editor can create a gorgeous skin-tone profile (either a generic one that works pretty well for most models, or a very specific one that works perfectly for a specific model)

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

[bcooter]

but not so great for fashion in uncontrolled lighting environments. Personally I'd rather have the accuracy and retouch out any undesirable accuracy than to have a camera which cannot pick up on subtle colors. I've had more than a few calls from photographers who had dSLRs that could not pick up the difference between two thread colors in an outfit (usually red), two hair shades, two paint colors etc and the client was not happy.

The Skin Tone tab in the Color Editor can create a gorgeous skin-tone profile (either a generic one that works pretty well for most models, or a very specific one that works perfectly for a specific model)

I'm not talking about uncontrolled lighting, I'm saying if you go out side on green grass with a 12x rag over the talent, expect yellow/green people and yes that's a big deal to fix.  

I've tried that skintone thing, it's ok, nothing that keeps you out of the paint brush and photoshop, but the medium format backs I own are just too damn color sensitive.  They see every blotch, every slight variation in skin tone and you can't globally correct for that.  I guess the skin tone editor thing would work better if you shot blue clothes, blue backgrounds and blue makeup, then it could single out the skintones, but it's not a fix and don't believe, then set it up next to the Canons and have a go at it with anything other than one softbox pointed at a face.

Try it and show it.

BC

[ErikKaffehr]

Hi,

That's just 4 MP. If you downsample a 40 MP sensor to 4 MP you would get about the same DR. The Kodak sensor has a lot of read noise, BTW, around 20 electrons that would reduce DR to around 25000:1.

So, you would be better of with a smaller pitch sensor in almost all cases. If you need a 50mm x 50mm square sensor, it's a different case, but how much are you willing to pay for that?

Just to explain how it works:

The Kodak sensor has few but large pixels, so each pixel can hold about 500000 electrons. A 40 MP digital back may hold about 60000 electrons in each pixel. If we downsample the 40 MP sensor to 4 MP we get about 600000 electrons per (downsampled) pixel. In practice DR is limited by shot noise (random variation of photons falling on sensor pixels) and that variation would be the same weather 4 MP (downsampled) or 4 MP (real).

Best regards
Erik

Wow!  A 50mm x 50mm square sensor with 15 stops DR - sounds perfect!   I'll take one of these for my Rollei 6008  please!

[pschefz]

I'm not talking about uncontrolled lighting, I'm saying if you go out side on green grass with a 12x rag over the talent, expect yellow/green people and yes that's a big deal to fix.  

I've tried that skintone thing, it's ok, nothing that keeps you out of the paint brush and photoshop, but the medium format backs I own are just too damn color sensitive.  They see every blotch, every slight variation in skin tone and you can't globally correct for that.  I guess the skin tone editor thing would work better if you shot blue clothes, blue backgrounds and blue makeup, then it could single out the skintones, but it's not a fix and don't believe, then set it up next to the Canons and have a go at it with anything other than one softbox pointed at a face.

Try it and show it.

BC

I totally agree.....some by the name of James wrote here a little while ago that the dmf backs show it in your models face if a red car drives by outside the studio.....and I completely agree.....

i always laugh when people talk about correct color.....most skin tones I see in print don't actually exist....but they look pleasant....
yes, there are situations where color is very important (fabrics, product,....) but in my experience the dmf backs don't really  have a much better track record then the dslrs.....and when you take moire into consideration the choice is clear again...

I am surprised to read about someone like rainer using dslr now.....but then again the 2 architectural photographers I know switched to dslr exclusively in the last year.....but I am with him all the way....I will get whatever canon comes out with next....

I have never liked shooting slrs.....I prefer mf and even rf.....but I really have no choice anymore....

[rainer_v]

I totally agree.....some by the name of James wrote here a little while ago that the dmf backs show it in your models face if a red car drives by outside the studio.....and I completely agree.....

i always laugh when people talk about correct color.....most skin tones I see in print don't actually exist....but they look pleasant....
yes, there are situations where color is very important (fabrics, product,....) but in my experience the dmf backs don't really  have a much better track record then the dslrs.....and when you take moire into consideration the choice is clear again...

I am surprised to read about someone like rainer using dslr now.....but then again the 2 architectural photographers I know switched to dslr exclusively in the last year.....but I am with him all the way....I will get whatever canon comes out with next....

I have never liked shooting slrs.....I prefer mf and even rf.....but I really have no choice anymore....

well i havent switched yet. i use both at the moment.

[pschefz]

well i havent switched yet. i use both at the moment.

i did not want to make it sound like you switched exclusively....2 people i know have....

architectural should really be a stronghold for dmf....and even in that field things are changing.....

on a side note....never got to see that rz33 but afaik it is just a rz with the aptus....which is a great back.....but why can't they make a wlf that makes the finder image look like 6x7? is it THAT hard?
once I hold a H or mamiya (or pentax) up to my eye, it turns into a dslr....only bigger, slower, with a laughable af....
am I the only one who wants what mf used to be?

[gwhitf]

once I hold a H or mamiya (or pentax) up to my eye, it turns into a dslr....only bigger, slower, with a laughable af....
am I the only one who wants what mf used to be?

I tested the P65+ on an H2 in Atlanta last week. Now THAT is an awesome viewfinder -- the large H image, coupled with no cropping. And the H2 shoots the P65+ very very fast, in full rez mode.

[EricWHiss]

Hi,

That's just 4 MP. If you downsample a 40 MP sensor to 4 MP you would get about the same DR. The Kodak sensor has a lot of read noise, BTW, around 20 electrons that would reduce DR to around 25000:1.

So, you would be better of with a smaller pitch sensor in almost all cases. If you need a 50mm x 50mm square sensor, it's a different case, but how much are you willing to pay for that?

Just to explain how it works:

The Kodak sensor has few but large pixels, so each pixel can hold about 500000 electrons. A 40 MP digital back may hold about 60000 electrons in each pixel. If we downsample the 40 MP sensor to 4 MP we get about 600000 electrons per (downsampled) pixel. In practice DR is limited by shot noise (random variation of photons falling on sensor pixels) and that variation would be the same weather 4 MP (downsampled) or 4 MP (real).

Best regards
Erik

Yeah but don't get hung up on the theoretical DR from specs. The DR that matters to working photographers is measured differently.  It always comes up with the armchair arguments for MFDB vs DSLR and its just pointless to go down that path.   The biggest problem with that kodak sensor is not that its 4mp, but rather that its just a sensor.  We'd need some of that phase or imacon voodoo and downstream electronics to make it useful - well that and its only monochrome.  

[eronald]

Yeah but don't get hung up on the theoretical DR from specs. The DR that matters to working photographers is measured differently.  It always comes up with the armchair arguments for MFDB vs DSLR and its just pointless to go down that path.   The biggest problem with that kodak sensor is not that its 4mp, but rather that its just a sensor.  We'd need some of that phase or imacon voodoo and downstream electronics to make it useful - well that and its only monochrome.

If you want to see the *practical* DR difference, just pick up a Canon and a Leica M8 or M9, shoot at low ISO and look at the way the Leica marks the textures. Although I must say that one cannot complain about the 5D2.

With a Phase back, you can shoot a model in direct sun, and bring back the shadows in post. If the Raw software you are using is decent there will be no hue shifts while opening the shadows. Hint: the most popular prosumer Raw converter is not necessarily the one you really really want to use after spending all that money, the guys who made the back may be making decent software for it too.

Edmund

[fredjeang]

If you want to see the DR difference, just pick up a Canon and a Leica M8 or M9 and look at the way the Leica sees into the shadows and marks the textures.

With a Phase back, you can shoot a model in direct sun, and bring back the shadows in post. If the Raw software you are using is decent there will be no hue shifts while opening the shadows. Hint: the most popular prosumer Raw converter is not necessarily the one you really really want to use after spending all that money, the guys who made the back may be making decent software for it too.

Edmund

Agree 100% Edmund

Cheers.

[ErikKaffehr]

Eric,

The point I want to make is that there is no real advantage to a large pitch sensor over a small pitch sensor. Would that be the case the, Phase One couldn't charge 30 kUSD (or so) for the P65+. Actually, if you use the technical definition of DR (at SNR equal 1) there may be an advantage with large pixels, but normally it's more "shot noise" which limits the acceptable noise levels in the darks. So in theory it's sensor size that matters and not pixel size.

A disturbing fact is that it is very hard to find a single comparison of MFDBs and DSLRs using adequate methods. With adequate methods I mean:

1) Reproducible setup
2) Correct exposure to the right (so full well capacity is actually utilized)
3) RAW ord DNG images (so they could be analyzed by RawAnalyzer or other tools)

Such an experiment would be very easy to conclude for anyone having both kinds of equipment, but I have actually not seen any.

Also, it seems that there is an obsession with DR in these discussion. DR is not particularly easy to measure. There are other parameters that may be much more important.

1) Sharpness obviously, no one really doubts that there is a sharpness advantage of larger sensors
2) Color handling, this is much dependent on the Color Grid Array in front of the sensor but also on algorithms
3) Internal reflections, number of air/glass surfaces, baffling

The point I really wanted to make that making a 2500 squre mm sensor with four gargantuan size megapixels would not produce better DR (in print) than a similar sized sensor with ten times as many smaller pixels as photon counts would be the same.

Finally, even a "cheap" MF equipment is a major investment. So I'd imagine that it is important that potential buyers are well informed, armchair arguments or not. I'd also suggest that anyone considering an investment in an MF system should consider arranging an equipment for test/loan/rental before shelling out their money.

Best regards
Erik

Yeah but don't get hung up on the theoretical DR from specs. The DR that matters to working photographers is measured differently.  It always comes up with the armchair arguments for MFDB vs DSLR and its just pointless to go down that path.   The biggest problem with that kodak sensor is not that its 4mp, but rather that its just a sensor.  We'd need some of that phase or imacon voodoo and downstream electronics to make it useful - well that and its only monochrome.