Luminous Landscape Forum
Site & Board Matters => About This Site => Topic started by: jerome_m on January 24, 2014, 01:39:55 pm
-
The article about the new IQ250 says that the back is very good at ISO6400. That may be the case, but it also says " In the past, with CCD digital backs it was just about impossible to use any ISO over 400 and expect decent image quality." Well, "decent image quality" is open to interpretation, but my old H3D-31 gives these results at ISO1600 when open in Phocus with default noise reduction:
(http://farm9.staticflickr.com/8529/8517989251_43761cf732_b.jpg) (http://www.flickr.com/photos/jerome_munich/8517989251/)
(http://www.flickr.com/photos/jerome_munich/8517989251/)
Below a crop and comparison with the D800 (on top):
(http://farm9.staticflickr.com/8242/8519190408_1578da2f06_o.jpg) (http://farm9.staticflickr.com/8242/8519190408_1578da2f06_o.jpg)
The complaint about poor iso performance comes from backs without micro-lenses (which can be used on technical cameras but have lower base sensitivity), but manufacturers have been selling backs with micro-lenses and a smaller sensor (the same size as the IQ250's sensor) for some time.
The phase IQ250 should improve noise level about one stop over CCD backs with micro-lenses, not much more.
-
I really shouldn't look at posts about MF. The drool is not good for the keyboard.
-
Only one stop. Likely wishful thinking.
But, let's wait for some real-world comparisons. There will be several people with Hasselblad, Phase and Leica S systems on our Antarctic expedition this week, so we'll know soon enough.
M
-
An obvious practical difference between CMOS and CCD sensors is high ISO image quality. This is due to the way these sensors handle different sources of noise.
But there is another practical difference: image quality at base ISO.
Full frame transfer CCD cells and much wider, and the base ISO is lower, so the gain applied to the high quality signal also is lower. Then, you have 16 bit AD converters (not 16 "output", but 16 bits converters). This allows for wider tonal range in the shadows. And this richer tonality gamut in the shadows at base ISO is the key of the higher perceived image quality in medium format backs.
The base ISO from CMOS sensors is higher, applied to a weaker signal (smaller cells) and due to this reason usually 14 bits converters are used. The tonal gamut in the shadows is not as rich as that of CCD/16 bit systems.
If all this is correct, the high ISO comparative should be complemented with low ISO comparative (looking at tonal gamut in the shadows). This does not mean comparing at the same ISO value, but comparing at the different base ISO values (for instance, ISO 50 for the CCD and ISO 100 for the CMOS), because those systems' base ISOs are adapted to the different conditions and characteristics of each system.
So, "CMOS versus CCD" is not all the story. The idea should be "CCD/16bits/50ISO versus CMOS/14bits/100ISO" systems.
-
Hi,
I don't think you analysis is correct. Dark areas are actually a weakness of CCDs.
Best regards
Erik
An obvious practical difference between CMOS and CCD sensors is high ISO image quality. This is due to the way these sensors handle different sources of noise.
But there is another practical difference: image quality at base ISO.
Full frame transfer CCD cells and much wider, and the base ISO is lower, so the gain applied to the high quality signal also is lower. Then, you have 16 bit AD converters (not 16 "output", but 16 bits converters). This allows for wider tonal range in the shadows. And this richer tonality gamut in the shadows at base ISO is the key of the higher perceived image quality in medium format backs.
The base ISO from CMOS sensors is higher, applied to a weaker signal (smaller cells) and due to this reason usually 14 bits converters are used. The tonal gamut in the shadows is not as rich as that of CCD/16 bit systems.
If all this is correct, the high ISO comparative should be complemented with low ISO comparative (looking at tonal gamut in the shadows). This does not mean comparing at the same ISO value, but comparing at the different base ISO values (for instance, ISO 50 for the CCD and ISO 100 for the CMOS), because those systems' base ISOs are adapted to the different conditions and characteristics of each system.
So, "CMOS versus CCD" is not all the story. The idea should be "CCD/16bits/50ISO versus CMOS/14bits/100ISO" systems.
-
You got it all wrong !
In the end it will all come down to the weight of the system including the sensor.
The heavier the better.
- If the system is heavy enough to keep the photographer at home,
if he/she is not really in the mood to take images that's good.
Great system! - If the system is heavy enough to force the photographer to really look at the scene,
feel and explore it and deeply think about composition, viewpoint, depth of field, color and light and
not to forget the message and care about each single image as much as possible that's good.
Great system! - If the system is heavy enough to keep the photographer from taking a crappy scene because its not worth the hassle that's good.
Great system!
In the end the advantage of one system above another should not be measured in f-stops but pounds and inches.
Maybe some gifted specialist could calculate a "bulkiness factor" for each system to get a real comparison.
So what is bulkier?
CCD or CMOS?
Have a nice weekend
Cheers
~Chris
P.S. Going out now and pointing my photocopier along with a diesel aggregate at some random crap for sheer fun.
-
There will be several people with Hasselblad, Phase and Leica S systems on our Antarctic expedition this week....
M
Why does this fail to surprise me?
:D
Roy
-
Hi,
On the other hand, carrying is a Hasselblad 555 with a P45+ back with five lenses and a Sony Alpha 99 plus two "professional" zoom lenses, an Arca D4 and an RRS TVC 33S makes for some good exercise needed by elderly overweight gentlemen like me.
Best regards
Erik
You got it all wrong !
In the end it will all come down to the weight of the system including the sensor.
The heavier the better.
- If the system is heavy enough to keep the photographer at home,
if he/she is not really in the mood to take images that's good.
Great system! - If the system is heavy enough to force the photographer to really look at the scene,
feel and explore it and deeply think about composition, viewpoint, depth of field, color and light and
not to forget the message and care about each single image as much as possible that's good.
Great system! - If the system is heavy enough to keep the photographer from taking a crappy scene because its not worth the hassle that's good.
Great system!
In the end the advantage of one system above another should not be measured in f-stops but pounds and inches.
Maybe some gifted specialist could calculate a "bulkiness factor" for each system to get a real comparison.
So what is bulkier?
CCD or CMOS?
Have a nice weekend
Cheers
~Chris
P.S. Going out now and pointing my photocopier along with a diesel aggregate at some random crap for sheer fun.
-
So, "CMOS versus CCD" is not all the story. The idea should be "CCD/16bits/50ISO versus CMOS/14bits/100ISO" systems.
Almost everything Nemo states is inaccurate and I would advise readers to ignore the post.
Does he think PhaseOne people are crazy. Why else would they offer a new camera with a smaller and inferior sensor at a US$10,000 premium over prior models?
Bill
-
Hi Bill,
Who is crazy? Phase One, cudstomers, Nemo, You, me and Bobby McGee? Countless permutations ;-)
:-) Sorry for the bad joke :-)
Erik
Almost everything Nemo states is inaccurate and I would advise readers to ignore the post.
Does he think PhaseOne people are crazy. Why else would they offer a new camera with a smaller and inferior sensor at a US$10,000 premium over prior models?
Bill
-
The "about one stop better" comes from an educated comparison of CCDs and CMOS cameras having similar pixel size. For example, the Pentax 645D and the Nikon D600 have about the same pixel size and noise at the pixel level is about one stop better on the D600. The IQ250 uses pixels of 5.3 µm, a size intermediate between the D800 and the D600. These cameras exhibit noise reduction artefacts from about iso 800 when examined at the pixel level, I don't see why the IQ250 would be any different.
In any case, the only way to know is indeed to try. I posted a test design here: http://www.luminous-landscape.com/forum/index.php?topic=86523.0 (http://www.luminous-landscape.com/forum/index.php?topic=86523.0). If someone could point an IQ250 at the same colour chart and banknotes, the results would be interesting.
-
Using pixel size as a metric really makes no sense, especially when comparing different generations of sensor, different technologies, different manufacturers, and different generations.
Michael
-
I am talking about comparing pictures at 1:1 on screen here. I don't see why 5.3 µm pixels in a Sony MF sensor would be less noisy than 5.3 µm pixels in a different Sony sensor.
Or do you believe we should be comparing cameras of the same sensor size between generations on the final print? That is fine with me. What is the difference between the noise of the IQ250 and the IQ140 at iso 800?
-
Using pixel size as a metric really makes no sense, especially when comparing different generations of sensor, different technologies, different manufacturers, and different generations.
Michael
That is right, mostly because CMOS sensors' active surface is smaller than CCD's active surface.
I would compare two systems at their base ISOs, even if those values are different.
Noise is not the thing to look at, when we compare at the lowest ISO value, but tonal gamut in shadows.
-
That is right, mostly because CMOS sensors' active surface is smaller than CCD's active surface.
With the pixel sizes of "SLR" and "MF" sized sensors and good modern gapless micro-lenses, sensors both gather light falling over almost the entire photo site area, whether CCD or CMOS. Also, with recent designs, well capacities are about the same for either CCDs or CMOS sensors at the same pixel pitch. The differences you talk of were relevant for older CMOS designs, and perhaps still are for the far smaller photosites in compact cameras and phones, but are probably not relevant to these big, modern sensors.
But let us not sweat the theory: I am fairly sure that tests will soon enough show that Phase One is not lying when it says that the CMOS sensor of the IQ250 has about one stop more dynamic range than even the CCD of the IQ260 with its larger 6 micron pixel pitch (14 stops vs 13.): http://www.phaseone.com/en/Camera-Systems/IQ2-Series/IQ2-Specifications.aspx
-
With the pixel sizes of "SLR" and "MF" sized sensors and good modern gapless micro-lenses, sensors both gather light falling over almost the entire photo site area, whether CCD or CMOS. Also, with recent designs, well capacities are about the same for either CCDs or CMOS sensors at the same pixel pitch. The differences you talk of were relevant for older CMOS designs, and perhaps still are for the far smaller photosites in compact cameras and phones, but are probably not relevant to these big, modern sensors.
But let us not sweat the theory: I am fairly sure that tests will soon enough show that Phase One is not lying when it says that the CMOS sensor of the IQ250 has about one stop more dynamic range than even the CCD of the IQ260 with its larger 6 micron pixel pitch (14 stops vs 13.): http://www.phaseone.com/en/Camera-Systems/IQ2-Series/IQ2-Specifications.aspx
There is a substantial difference in photos gathering. It is not a question gapless microlenses.
-
and perhaps still are for the far smaller photosites in compact cameras and phones, but are probably not relevant to these big, modern sensors.
you forgot that innovations in terms of getting every single photon in the bucket are exactly in smaller sensors and then trickle to big ones ;)
-
What substantial difference?
Best regards
Erik
There is a substantial difference in photos gathering. It is not a question gapless microlenses.
-
The heavier the better.
- If the system is heavy enough to keep the photographer at home,
if he/she is not really in the mood to take images that's good.
Great system! - If the system is heavy enough to force the photographer to really look at the scene,
feel and explore it and deeply think about composition, viewpoint, depth of field, color and light and
not to forget the message and care about each single image as much as possible that's good.
Great system! - If the system is heavy enough to keep the photographer from taking a crappy scene because its not worth the hassle that's good.
Great system!
P.S. Going out now and pointing my photocopier along with a diesel aggregate at some random crap for sheer fun.
This is so damn right: my best photos so far come from my Mamiya RZ67 system whit no prism finder.
Said so:
I will now make a statement and question, I'm newbie on image quality, and a photographer professionally for just 5 years. But i really don't understand why everybody use the D800 as comparison meter. I own a d800 with a nice park of prime lens, i calibrate it with color passport and do my work on C1. I also own a Leaf Valeo 22.
The Valeo is 10 ( ! ) years older technology, super low to sensibility to light ( is best is iso 25, is 50 is just ok ) and incredibly slow. Also it work only with Leafcapture ( which looks designed from north korea military ).
Even so, if i shoot in my studio the image quality ( color and contrast ) is way superior to the D800. The D800 always end up with resolve complex areas in a muggy way, even when i shoot with the nikon 14-24, on a tripod, with mirror up. I mean is a great sensor, and is way more useful as portable, cheap, and usable with capture1. But if it come to Quality, real image quality, what i feel i can get from my Valeo is just superior ( and client feels the same to ).
How can the d800 chip be compared, i presume, with the IQ250?
humble opinion.
-
What substantial difference?
Best regards
Erik
The full-well capacity of the CCD of the Leica M9 is 60.000 photons per pixel. The CMOS of the M (type 240), with slightly smaller "pixels", 40.000. This means a difference of 0,6EV.
The dynamic range is a combination of signal and noise though.
-
Even so, if i shoot in my studio the image quality ( color and contrast ) is way superior to the D800. The D800 always end up with resolve complex areas in a muggy way, even when i shoot with the nikon 14-24, on a tripod, with mirror up. I mean is a great sensor, and is way more useful as portable, cheap, and usable with capture1. But if it come to Quality, real image quality, what i feel i can get from my Valeo is just superior ( and client feels the same to ).
Either you are using higher iso on your D800 (I posted a few samples yesterday in the MF forum that shows that the D800 noise reduction is quite visible on fine details at iso800) or what you are seeing is the effect of the low-pass filter. Looked at 1:1 on screen a camera without a low pass filter will appear sharper. Trying a D800e or an A7r would be fairer.
Moreover, the 14-24 is a tour de force in optical design and probably sharpest than any Nikon wide-angle prime, but it is not perfect. You will definitely see the limits of that lens on the D800.
-
The full-well capacity of the CCD of the Leica M9 is 60.000 photons per pixel. The CMOS of the M (type 240), with slightly smaller "pixels", 40.000. This means a difference of 0,6EV.
That is for a CMOSIS CMOS sensor, and CMOSIS is a bit of an outlier. I would like to see data for Sony Exmor sensors since that is what Phase One and Hasselblad will be using, but most of the data I can find are at
http://www.clarkvision.com/articles/digital.sensor.performance.summary/#data
http://www.clarkvision.com/articles/digital.sensor.performance.summary/#full_well
which unfortunately is rather lacking in recent Sony CMOS sensors. It does show that Canon CMOS sensors fairly consistently have higher full well capacity than Kodak CCDs of similar pixel size. (The unmarked blue squares on figure 2 are mostly for Kodak CCDs; blue squares are numbers from manufacturer's data sheets, which Kodak and Dalsa provide while Canon, Sony and Nikon do not for their CMOS sensors.)
-
That is for a CMOSIS CMOS sensor, and CMOSIS is a bit of an outlier.
Indeed. Even my Canon EOS 1Ds3 from 2007 has a higher 'well' capacity per surface area than those. I have calculated that the 1Ds3 has a capacity of approx. 57000-58000 electrons per pixel, which equates to some 1400 photons per square micron. I believe that more modern fabs can achieve something like 1800 photons per square micron.
I would like to see data for Sony Exmor sensors since that is what Phase One and Hasselblad will be using,
As soon as we can interpret the Raw data of the IQ250 (with RawDigger or similar), it will be possible to determine such metrics, independently from e.g. DxO (who usually have virtually identical results for 'screen' metrics as independent tests do).
Cheers,
Bart
-
Hi,
I don't have either camera and I neither shoot in studio nor people. I am a landscape type of photographer. So I have just a few general comments.
- The D800 is taken for comparison, because it is the most high resolving full frame 135 camera. Now it got competition from the Sony A7r.
- In general larger pixels look better than small pixels at actual pixels view, but the number of pixels is important
- Best way to compare images is often print
- Colour rendition may be different
As I said, I don't shoot Nikon D800 (or any other Nikon) nor Leaf, but I shoot Phase One P45+ and Sony Alpha 99 (full frame 135, 24 MP) and I don't see a horrible lot of difference in my type of shooting.
Here is an article describing my experience, with a lot of real world samples in the end: http://echophoto.dnsalias.net/ekr/index.php/photoarticles/80-my-mfd-journey-summing-up
Best regards
Erik
This is so damn right: my best photos so far come from my Mamiya RZ67 system whit no prism finder.
Said so:
I will now make a statement and question, I'm newbie on image quality, and a photographer professionally for just 5 years. But i really don't understand why everybody use the D800 as comparison meter. I own a d800 with a nice park of prime lens, i calibrate it with color passport and do my work on C1. I also own a Leaf Valeo 22.
The Valeo is 10 ( ! ) years older technology, super low to sensibility to light ( is best is iso 25, is 50 is just ok ) and incredibly slow. Also it work only with Leafcapture ( which looks designed from north korea military ).
Even so, if i shoot in my studio the image quality ( color and contrast ) is way superior to the D800. The D800 always end up with resolve complex areas in a muggy way, even when i shoot with the nikon 14-24, on a tripod, with mirror up. I mean is a great sensor, and is way more useful as portable, cheap, and usable with capture1. But if it come to Quality, real image quality, what i feel i can get from my Valeo is just superior ( and client feels the same to ).
How can the d800 chip be compared, i presume, with the IQ250?
humble opinion.
-
The "about one stop better" comes from an educated comparison of CCDs and CMOS cameras having similar pixel size. For example, the Pentax 645D and the Nikon D600 have about the same pixel size and noise at the pixel level is about one stop better on the D600. The IQ250 uses pixels of 5.3 µm, a size intermediate between the D800 and the D600. These cameras exhibit noise reduction artefacts from about iso 800 when examined at the pixel level, I don't see why the IQ250 would be any different.
I went to DxO with this question. When I compare the 645D with the D610, I get two stops improvement in DR in "print" mode. When I look at "screen" (pixel level) mode, the difference is even wider, with the D610 getting a 2.5 stop advantage in DR.
-
I will start saving my $$$ for the full frame CMOS replacement for my IQ180!
Maybe next year? I'm excited about the potential IQ
Marc :)
-
Hi Marc,
Nice to hear about your enthusiasm.
Right now it may seem that Phase has a lot of emphasis on high ISO, possibly to much, you need to save a lot so I guess Phase has some time to sort of things.
Clean air and no turbulences up there?
Best regards
Erik
I will start saving my $$$ for the full frame CMOS replacement for my IQ180!
Maybe next year? I'm excited about the potential IQ
Marc :)
-
I went to DxO with this question. When I compare the 645D with the D610, I get two stops improvement in DR in "print" mode. When I look at "screen" (pixel level) mode, the difference is even wider, with the D610 getting a 2.5 stop advantage in DR.
I was discussing noise level and not dynamic range. This is what I get in "screen" mode. When I draw an horizontal line to compare the two cameras at the same noise level, the intersect points are separated by one stop. Try it for example at 30 dB, where there already is an horizontal line: about iso 600 for Pentax and about iso 1000 for the Nikon. The pixel pitch is the same between the two cameras: 5.9 µm.
-
In screen mode for 18% SNR the D70 does better than the 645D.
The significance of this metric is a bit hard to parse. Does it point to midtone response? How much is chroma noise and how much is luma noise? How much is pattern noise? It certainly isn't saying that the D70 is a better camera than the 645D and about as good as the D800.
-
Hi Marc,
Nice to hear about your enthusiasm.
Right now it may seem that Phase has a lot of emphasis on high ISO, possibly to much, you need to save a lot so I guess Phase has some time to sort of things.
Clean air and no turbulence up there?
Best regards
Erik
Clean air yes, no turbulence some times :), I've done well turning in my Phase One backs for a healthy discount skipping generations P30 for a P45+ for an IQ180 for a IQXXX!
Marc
-
In screen mode for 18% SNR the D70 does better than the 645D.
And so it should be, since the Nikon D70 has larger pixels than the Pentax 645D. It is not as good as a camera, because it has a lot less pixels, but "screen mode" refers to the noise level per pixel.
To explain again why I compared the D610 and 645D: the question is what the change between ccd and cmos technology will bring for MF cameras. My idea is to compare the 645D, a ccd camera, with a hypothetical 645Dcmos, that is the same camera, same sensor size and same number of pixels but with a cmos sensor. That 645Dcmos should have a better low light response, but by how much?
To determine how much better the 645Dcmos could be, I check how much better the noise level per pixel is between a recent ccd camera of a given pixel size and another camera having the same pixel size but built with cmos technolgy. These two cameras are the 645D (again, but we don't have much choice as to recent ccd cameras) and the 610D. I find that the cmos pixel is about one stop less noisy than the ccd pixel. Thus, I conclude that the 645Dcmos, should it come to market, would be about one stop less noisy than the current 645D.
It is an approximation, but it should give a reasonable ballpark estimate. The end camera will probably have 50 mpix and not 40 and may have other improvements or not. Still: I would expect the cmos successor to the 645D and also the new IQ250 to be roughly one stop better than the current 645D (or H5D-40, which uses the same sensor as the 645D).
Compare that to the current press release of Phase One and a recent lula article which give the impression (granted: without actually saying it) that the cmos MF cameras are unusable beyond iso400 and that the IQ250 would be great at iso 6400...
-
Not that I'm going to be buying this new digital back anytime soon, but I do have a lot of interest in this discussion because we'll learn what it is that makes medium format digital actually tick.
-
Here is an article describing my experience, with a lot of real world samples in the end: http://echophoto.dnsalias.net/ekr/index.php/photoarticles/80-my-mfd-journey-summing-up
Great reading, i was actually already bump in that images before.
-
18% SNR in screen mode does not seem to be a blanket measure of noise per pixel. It seems to involve specific SNR in midtone. And without the SD term, it's hard to know what the qualitative effects will be. Meanwhile, dynamic range differences between these two cameras, also in screen mode (around 2.5 stops), suggest that there is not a single unambiguous "number of stops difference" that will capture the qualitative differences between these two cameras.
I'd be interested to see an image on the 645D at ISO 3200, compared with the same image on the IQ250 at ISO 6400. This would give an indication of whether the difference that you estimate at about "1 stop" makes sense in a qualitative assessment.
-
Lorenzo,
Thanks for comment, thanks for being nice to people!
Best regards
Erik
Great reading, i was actually already bump in that images before.
-
Hi,
18% SNR is essentially about shoot noise until we get into very high ISOs, and that will be roughly equal on sensors of similar pixel size if comparable CGA-s are used.
The difference will be mainly in the shadows, and that is much better described by DR.
Best regards
Erik
18% SNR in screen mode does not seem to be a blanket measure of noise per pixel. It seems to involve specific SNR in midtone. And without the SD term, it's hard to know what the qualitative effects will be. Meanwhile, dynamic range differences between these two cameras, also in screen mode (around 2.5 stops), suggest that there is not a single unambiguous "number of stops difference" that will capture the qualitative differences between these two cameras.
I'd be interested to see an image on the 645D at ISO 3200, compared with the same image on the IQ250 at ISO 6400. This would give an indication of whether the difference that you estimate at about "1 stop" makes sense in a qualitative assessment.