• Please use real names.

    Greetings to all who have registered to OPF and those guests taking a look around. Please use real names. Registrations with fictitious names will not be processed. REAL NAMES ONLY will be processed

    Firstname Lastname

    Register

    We are a courteous and supportive community. No need to hide behind an alia. If you have a genuine need for privacy/secrecy then let me know!
  • Welcome to the new site. Here's a thread about the update where you can post your feedback, ask questions or spot those nasty bugs!

PhaseOne introduces their IQ250 Digital Back

Hi Folks,

Shortly after the Hasselblad announcement of the first CMOS based "Medium Format" sensor, PhaseOne just introduced their new IQ250 Digital Back, also with a CMOS sensor (sourced from Sony), with a slightly smaller sensor size (44 x 33 mm, 5.3 micron pitch) compared to the other PhaseOne IQ2 CCD sensor backs (e.g. 53.9 x 40.4 mm for the IQ260):

IQ250 highlights:

- ISO100 to ISO6400
- 14 f-stops dynamic range
- 50 megapixels captures
- Continuous workflow
- Fluid Live View

Datasheet

None of the current IQ2 backs use anti-aliasing filters.
The IQ250 does not use Sensor+ (lower resolution with on sensor binning) technology.

An informative Skype call interview recorded with PhaseOne management, can be found here.

Cheers,
Bart
 
Last edited:

Asher Kelman

OPF Owner/Editor-in-Chief
Well, Bart, that video is really enlightening. It does show, once again, the very close relationship between Luminous Landscape and Phase One! The statements by Michael that recognizing the oft-spoken of "3D appearance of the MF, CD, digital image, is something akin to the US Supreme court Justice's claim that he knew pornography when he saw it, even though he couldn't define it.

Michael went on to state that with MF imaging, "Everyone who sees it, knows it!"

Well, I wonder about that! If one has spent $50,000 on a camera, one expects it to give "better" images. With tests of medicines, the "placebo effect" allows colored starch pills to often have the calming effect of barbiturates!! Who has ever made blind tests of photographs taken under the same conditions? Can one actually recognize a MF image out of a bunch with appropriate choice of lens aperture to mimic the same DOF.

Frankly, so far, the Hasselblad images seemed more 3D in the link given by Jerome, but I could be kidding myself.

Asher
 
Well, Bart, that video is really enlightening. It does show, once again, the very close relationship between Luminous Landscape and Phase One!

Indeed.

The statements by Michael that recognizing the oft-spoken of "3D appearance of the MF, CD, digital image, is something akin to the US Supreme court Justice's claim that he knew pornography when he saw it, even though he couldn't define it.

Yes, and while it's a nice joke, it's also a stupid remark coming for a judge ...

Michael went on to state that with MF imaging, "Everyone who sees it, knows it!"

Well, I wonder about that! If one has spent $50,000 on a camera, one expects it to give "better" images. With tests of medicines, the "placebo effect" allows colored starch pills to often have the calming effect of barbiturates!!

That's probably one aspect of it, but there can certainly be a real difference as well.

The problem is that most people are not attributing the effect to the correct origin. The real reason has to do with the larger magnification factor which results in a higher MTF response. That, combined with shallower DOF and lower output magnification should make a significant difference. There are more factors, such as lack of low-pass filters (which boosts MTF but may also be a drawback for certain subjects, such as fabric) and more selective Bayer CFA filter characteristics for better color separation.

So it depends on what is being compared, but there can be significant differences.

Cheers,
Bart
 

Jerome Marot

Well-known member
I wonder why I took the time to post in this forum a long explanation of the optical reasons for the differences in lens rendering between formats.
 

Asher Kelman

OPF Owner/Editor-in-Chief
I wonder why I took the time to post in this forum a long explanation of the optical reasons for the differences in lens rendering between formats.

You didn't waste time. However, there are, as Bart mentioned above a lot of parts to this "felt" greater presence of MF digital with beliefs mixing in, that, for example, the CCDS has an advantage in creating that "magic" which is lacking in CMOS sensors.

Asher
 
Indeed.



Yes, and while it's a nice joke, it's also a stupid remark coming for a judge ...



That's probably one aspect of it, but there can certainly be a real difference as well.

The problem is that most people are not attributing the effect to the correct origin. The real reason has to do with the larger magnification factor which results in a higher MTF response. That, combined with shallower DOF and lower output magnification should make a significant difference. There are more factors, such as lack of low-pass filters (which boosts MTF but may also be a drawback for certain subjects, such as fabric) and more selective Bayer CFA filter characteristics for better color separation.

So it depends on what is being compared, but there can be significant differences.

Cheers,
Bart
I guess that the new sensor is here to make the "DSLR user" path towards MF easier… Now he can advance to MF by retaining his "habits"… he can see "the light" later on...
 

Jerome Marot

Well-known member
You didn't waste time. However, there are, as Bart mentioned above a lot of parts to this "felt" greater presence of MF digital with beliefs mixing in, that, for example, the CCDS has an advantage in creating that "magic" which is lacking in CMOS sensors.

I respect Bart, but I don't agree with some of his arguments:

-larger magnification factor which results in a higher MTF response: I suppose that Bart means lens MTF. I have discussed the issue of lenses, and it is a bit more complex than "higher MTF".

-shallower DOF: 24x36 is the format which can attain the thinner dof, because of the availability of very fast lenses.

-lower output magnification: yes, but not for pixel peepers

-lack of low-pass filters: indeed, unless one compares to a D800e or A7r

-more selective Bayer CFA filter characteristics for better color separation: actually, Phase One backs fare relatively poorly on that metrics.

All in all, a 6µm pixel is a 6µm pixel, whether it is in a 24 mpix 24x36 sensor or in a 50 mpix MF sensor (the ones already on the market, the IQ250 has smaller pixels), so there is little reason why we would see differences in noise or color at the pixel level. There are differences, but with the exception of the differences coming from lenses, they have more to do with the particular marketing choices of the manufacturer than with inherent advantage of a sensor barely twice as big. In a nutshell: 24x36 manufacturers optimise their systems for low light at the detriment of color separation and highlight headroom, MF manufacturers know their market shoots with plenty of light and needs pleasant skin tones.

But there is one thing we can agree about: MF cameras are very nice machines with fabulous lenses. They are worth a try. They can be rented for reasonable prices and second hand models can be had for sums noticeably lower than what Phase One is asking for the IQ250.
 
I respect Bart, but I don't agree with some of his arguments:

-larger magnification factor which results in a higher MTF response: I suppose that Bart means lens MTF. I have discussed the issue of lenses, and it is a bit more complex than "higher MTF".

Hi Jerome,

I agree, and actually it's always a combination. A sensor alone rarely makes an interesting image. A lens alone only produces a virtual image. Both have an MTF response which will need to be combined (usually multiplied) to achieve the 'system MTF' and an image on a sensor.

However, the sensor alone (e.g. sampling density) can already make a huge difference, even with the same lens. With a longer focal length, needed for the larger image circle of a larger sensor array, the image magnification has a similar effect as a denser sampling pitch, both boost the MTF response for a given level of detail.

Now add a better lens, and the total system MTF will increase even further.

-shallower DOF: 24x36 is the format which can attain the thinner dof, because of the availability of very fast lenses.

Sure, but the fact remains that a longer focal length with the same aperture and exposure time will produce a shallower DOF. Many studio photographers use a fixed aperture setting and regulate exposure by adjusting flash power. When switching lenses, they tend to keep the same lighting setup.

That doesn't mean that there are no other possible approaches, exactly why a lens like the 55mm Otus seems so popular, the price not withstanding.

-more selective Bayer CFA filter characteristics for better color separation: actually, Phase One backs fare relatively poorly on that metrics.

Not according to some of the (usually reliable) DxO scores, e.g. the 2 best Portrait (Color Depth) scores are for PhaseOne backs.

How that will pan out for the new Sony sourced sensor, has yet to be evaluated.

Cheers,
Bart
 

Jerome Marot

Well-known member
However, the sensor alone (e.g. sampling density) can already make a huge difference, even with the same lens. With a longer focal length, needed for the larger image circle of a larger sensor array, the image magnification has a similar effect as a denser sampling pitch, both boost the MTF response for a given level of detail.

If you were able to increase the size of a camera and lens by 50% in all linear dimensions (which is about the difference between 24x36 and MF), you would realise that geometric optical aberrations increase in the same proportions. The MTF per pixel would not change by this operation.

Sure, but the fact remains that a longer focal length with the same aperture and exposure time will produce a shallower DOF. Many studio photographers use a fixed aperture setting and regulate exposure by adjusting flash power. When switching lenses, they tend to keep the same lighting setup.

In a typical studio setting, photographers indeed work at a fixed aperture, but they also rarely alternate between cameras of different sensor sizes. Moreover, if you look at, say, fashion photography, you will realise that MF users, in average, do not tend to use extremely thin d.o.f. As one of them said to me once "portraits where one eye is sharp and the other fuzzy look strange". The idea that MF has better bokeh does not mean that it allows thinner d.o.f., but that out of focus areas do not exhibits artefacts that make them difficult to read. I should probably write an article with samples about this subject.

That doesn't mean that there are no other possible approaches, exactly why a lens like the 55mm Otus seems so popular, the price not withstanding.

"Popular" here only means that photographers active on internet forums phantasise about owning one.


Not according to some of the (usually reliable) DxO scores, e.g. the 2 best Portrait (Color Depth) scores are for PhaseOne backs.

Check their measurements of sensitivity metamerism index according to ISO 17321, a well defined standard.
 

Doug Kerr

Well-known member
Hi, Bart,
The real reason has to do with the larger magnification factor which results in a higher MTF response.

For a sensor with a given pixel pitch (assuming that the pitch defines its properties), a greater image magnification (for an object at a certain distance) "improves" the sensor MTF only if we think of the frequency scale as being in cycles per degree (rather than, for example, cycles per mm).

Recall that a greater focal length gives a greater image magnification for an object at a given distance. I Imagine that this is what you were referring to.​
Of course we get this benefit with a sensor of a given size when, for whatever reason, we choose to use a greater focal length. (It's that old "reach" thing!)

To carry on a bit further, for a sensor with a given pixel pitch (assuming that the pitch defines its properties), a sensor of larger dimensions "improves" the sensor MTF only if we think of the frequency scale as being in cycles per picture height.

Best regards,

Doug
 

Asher Kelman

OPF Owner/Editor-in-Chief
..............

To carry on a bit further, for a sensor with a given pixel pitch (assuming that the pitch defines its properties), a sensor of larger dimensions "improves" the sensor MTF only if we think of the frequency scale as being in cycles per picture height.

IOW, Doug, better lens for large magnifications?

That's the metric we need in this instant, all other things being equal.

Asher
 
For a sensor with a given pixel pitch (assuming that the pitch defines its properties), a greater image magnification (for an object at a certain distance) "improves" the sensor MTF only if we think of the frequency scale as being in cycles per degree (rather than, for example, cycles per mm).

Recall that a greater focal length gives a greater image magnification for an object at a given distance. I Imagine that this is what you were referring to.​
Of course we get this benefit with a sensor of a given size when, for whatever reason, we choose to use a greater focal length. (It's that old "reach" thing!)

Hi Doug,

That's correct, altough the physically larger sensor array (more sensels) also allow to maintain the same total field of view (within limits of differing aspect ratios). One has just to look at the MTF curve, at lower spatial frequencies (due to larger projection size on the sensor), the same detail will have a larger MTF response, subject contrast is retained better. In addition, because it's larger, it's easier to post-process for local contrast.

To carry on a bit further, for a sensor with a given pixel pitch (assuming that the pitch defines its properties), a sensor of larger dimensions "improves" the sensor MTF only if we think of the frequency scale as being in cycles per picture height.

Indeed.

Cheers,
Bart
 

Doug Kerr

Well-known member
Hi, Asher,

IOW, Doug, better lens for large magnifications?

That's the metric we need in this instant, all other things being equal.

Well first remember that a given setup (sensor, lens of a certain focal length) does not have a certain image magnification unless we contemplate an object at a certain distance.

So I think (image) magnification is not the parameter that you have in mind, and so to continue a discussion revolving around that property is doomed.

It's all well and good to call bison "buffalo", until the one you ordered from Exotic Animals 'R' Us arrives at the door.​

What is the parameter that is created by the combination of sensor dimensions and lens focal length?

Field of view.

Now another important parameter is cycles of (sensor) resolution per degree. This is related to a valid concept of what is the "reach" of a camera with a certain focal length lens.

Of course, to fully exploit the resultion potential of a sensor requires a lens whose MTF (expressed on that same frequency scale) "supports" that sensor resolution.

Where we get into trouble is when we start comparing cameras with two different sensor sizes.

Suppose camera A has a 36 × 24 mm sensor, and is equipped with a 50 mm focal length lens. The sensor has a certain MTF (in terms of cycles per mm).

Now we shift to camera B with a 54 × 36 mm sensor, with the same MTF (in terms of cycles per mm).

Assume that for consistency in field of view (that is, to use camera B for the same composition, etc.) we will use a 75 mm lens on camera B.

Now the question is, "to 'support' the sensor in setup B as well as the lens did in setup A, will the lens in setup B have to be 'better' ?"

Well, since we are speaking of lenses of different focal length, we have to be a little careful about what we mean by "better".

For example, if our 75 mm lens (on camera B) has the same MTF (in the usual terms of a frequency scale in cycles per mm) as our 50 mm lens (on camera A), then (simplistically at least) the two lenses will equally "support" the potential resolution afforded by the sensors in the two cameras.

Does that mean that our 75 mm lens, if it meets that criterion, is a "better" lens than our 50 mm lens with the same MTF?

Best regards,

Doug
 

Asher Kelman

OPF Owner/Editor-in-Chief
Thanks Doug, for the, (customary), erudite discussion taking into account field of view and different lens focal length

Let me alter the perspective, so to speak, LOL! I wish to print large. So I'd like to choose sensor-lens combinations to deliver 16x20 or 32x40 or even 64x80 prints, viewable at 12" as "perfect".

To print, let's assume that the subject is optimally framed in any format and with whatever lens one chooses. To print the image well enough to view it "accurately" at that fixed distance of 12", then one needs a certain number of cycles total for that size of print along it's longest axis. That, is what I was grabbing at from your analysis. Perhaps that was a mistake on my part!

We actually need a certain number of cycles total to achieve our goal of being able to look at the print from 12" distance to the print, anywhere over the print and get pretty well the same sense of perfection. Lens effects such a Bokeh, distortions and illumination we shall not consider as a defect here for this argument.

If you wish, you can choose another viewing distance, but it should be such that some lenses can meet this requirement and others would fail.

IOW, one figure, total cycles per longest dimension of image would define whether or not the lens-sensor combination can deliver the required detail.

Asher
 

Doug Kerr

Well-known member
Hi, Asher,

IOW, one figure, total cycles per longest dimension of image would define whether or not the lens-sensor combination can deliver the required detail.

Basically that's it.

It doesn't matter whether we quote cycles per picture width or cycles per picture height or cycles per picture diagonal, so long as we make it clear which we are stating. The custom is cycles per picture height.

So what is the benefit of a larger format size?

Well with respect to the above, if a certain pixel pitch is desirable from a standpoint of manufacturability, noise performance, and the like, and results in a sensor resolution of a certain number of cycles per mm, then a larger sensor will (insofar as the sensor is concerned) lead to a greater number of cycles per picture height.

Now, a property of a larger format size is:

For almost any reasonable set of conditions of comparison, then for a certain f-number, the depth of field will be less (which can be better or worse depending on what you are trying to do).

Best regards,

Doug
 

Jerome Marot

Well-known member
Let me alter the perspective, so to speak, LOL! I wish to print large. So I'd like to choose sensor-lens combinations to deliver 16x20 or 32x40 or even 64x80 prints, viewable at 12" as "perfect".

When I got access to a 24" / 60cm printer 3 years ago, I had to deal with this particular question. I found it easier to answer when making abstraction of the camera lens at first.

Even if I normally measure dimensions in metric units, I found that it is easier to use inches in that particular case, because everybody is used to resolution being given in dot per inches (dpi).

What is the needed resolution for large prints in dpi? For small photographs or generally continuous tone images about 300 dpi is the maximum needed, as most humans cannot see anything finer resolved without the aid of a magnifier. Interestingly, this is also the resolution of fine art magazines and retina displays. So, we will use the value of 300 dpi as a maximum limit for small prints.

For large prints, that value is relaxed a bit. My feeling is that a value of about 150 dpi sufficient. The prints still stay apparently sharp when examined close up. Note that while about 300 dpi is relatively well defined as an upper bound (it comes from the characteristics of the human vision), that value is more fuzzy. It depends on the particular picture printed and one's particular taste as to what constitutes a "sharp print". Still, 150 dpi is about in the center of what is used for large fine art prints that look quite sharp when examined up close. For comparison: computer screens are about 80-120 dpi, large 4K TVs are about the same.

Taking the value of 150 dpi, what camera resolution do we need for our prints?

The first question to be asked is: what size can we print? Fine art inkjet printers and papers are available in the following approximative width: 17" (40cm), 24" (60cm), 44" (110cm) and 60" (150cm). For bigger sizes, we would need a C-print, the Océ LightJet500XL printer is 76" (2m) wide (think Andreas Gursky). These machines are not manufactured any more, the only remaining manufacturer is ZBE Chromira, and their wider machine is 50". Larger printer sizes are available, but not in this kind of quality. That gives us the maximal height of a landscape print, the other dimension is given by the aspect ratio. We will choose 3x4 as an aspect ratio.

The number of pixels necessary is thus given by the following formula: (150 x w)^2 x 4 /3. For the available widths:
17: 8.5 mpix
24: 17 mpix
44: 58 mpix
60: 108 mpix
76: 173 mpix

But please keep in mind that these figures are approximative and derived from our 150 dpi value. And because that value is squared in the formula, small modifications in the dpi value will result in large modifications in the mpix value. Conversely, having a bit less or a bit more mpix, will only lower or increase the dpi value a little bit. taking for example the 24" print, at what dpi can we print it with various cameras? The formula here is:
(√(mpix x 3 /4)) / 24, so:

12 mpix: 125 dpi
24 mpix: 176 dpi
34 mpix: 210 dpi

and, in my experience, the difference between a 150 dpi and a 210 dpi print is barely noticeable. 120 dpi, on the other hand, will be problematic for a landscape with fine details but may be ok for a soft portrait.

Coming back to cameras and lenses, we see that most 24x36 cameras will be sufficient for 24" prints. The question is: are they sufficient for the next size up: 44"?

Finding out the resolution given on the prints by the highest resolution cameras: the D800 and A7r is quite easy. The frame size of the sensor is 7360 × 4912, so we just divide 4912 by 44 to find out the resulting dpi: 112 dpi. If we need to print at 150 dpi with that camera, the maximum print height will be about 32" (85cm).

What about the lenses on the camera, then? Well, because of the vast amount of pixel peepers on the Internet, answering that question is relatively easy: printing at the maximum size allowed by the sensor resolution and expecting a sharp print examined close is the same as examining a 1:1 detail of an image on-screen (the computer screen dpi resolutions are comparable to average prints, remember?). And the answer is: there are very few 24x36 lenses which are perfectly sharp corner to corner on the D800 / A7r especially if you need a wide-angle lens. So here again the practical answer may be to limit yourself to 24" prints. On medium format, the answer may be different since there are many lenses which are sharp corner to corner on the highest resolution MF cameras, including wide angle lenses.
 
When I got access to a 24" / 60cm printer 3 years ago, I had to deal with this particular question. I found it easier to answer when making abstraction of the camera lens at first.

Even if I normally measure dimensions in metric units, I found that it is easier to use inches in that particular case, because everybody is used to resolution being given in dot per inches (dpi).

What is the needed resolution for large prints in dpi? For small photographs or generally continuous tone images about 300 dpi is the maximum needed, as most humans cannot see anything finer resolved without the aid of a magnifier. Interestingly, this is also the resolution of fine art magazines and retina displays. So, we will use the value of 300 dpi as a maximum limit for small prints.

For large prints, that value is relaxed a bit. My feeling is that a value of about 150 dpi sufficient. The prints still stay apparently sharp when examined close up. Note that while about 300 dpi is relatively well defined as an upper bound (it comes from the characteristics of the human vision), that value is more fuzzy. It depends on the particular picture printed and one's particular taste as to what constitutes a "sharp print". Still, 150 dpi is about in the center of what is used for large fine art prints that look quite sharp when examined up close. For comparison: computer screens are about 80-120 dpi, large 4K TVs are about the same.

Taking the value of 150 dpi, what camera resolution do we need for our prints?

The first question to be asked is: what size can we print? Fine art inkjet printers and papers are available in the following approximative width: 17" (40cm), 24" (60cm), 44" (110cm) and 60" (150cm). For bigger sizes, we would need a C-print, the Océ LightJet500XL printer is 76" (2m) wide (think Andreas Gursky). These machines are not manufactured any more, the only remaining manufacturer is ZBE Chromira, and their wider machine is 50". Larger printer sizes are available, but not in this kind of quality. That gives us the maximal height of a landscape print, the other dimension is given by the aspect ratio. We will choose 3x4 as an aspect ratio.

The number of pixels necessary is thus given by the following formula: (150 x w)^2 x 4 /3. For the available widths:
17: 8.5 mpix
24: 17 mpix
44: 58 mpix
60: 108 mpix
76: 173 mpix

But please keep in mind that these figures are approximative and derived from our 150 dpi value. And because that value is squared in the formula, small modifications in the dpi value will result in large modifications in the mpix value. Conversely, having a bit less or a bit more mpix, will only lower or increase the dpi value a little bit. taking for example the 24" print, at what dpi can we print it with various cameras? The formula here is:
(√(mpix x 3 /4)) / 24, so:

12 mpix: 125 dpi
24 mpix: 176 dpi
34 mpix: 210 dpi

and, in my experience, the difference between a 150 dpi and a 210 dpi print is barely noticeable. 120 dpi, on the other hand, will be problematic for a landscape with fine details but may be ok for a soft portrait.

Coming back to cameras and lenses, we see that most 24x36 cameras will be sufficient for 24" prints. The question is: are they sufficient for the next size up: 44"?

Finding out the resolution given on the prints by the highest resolution cameras: the D800 and A7r is quite easy. The frame size of the sensor is 7360 × 4912, so we just divide 4912 by 44 to find out the resulting dpi: 112 dpi. If we need to print at 150 dpi with that camera, the maximum print height will be about 32" (85cm).

What about the lenses on the camera, then? Well, because of the vast amount of pixel peepers on the Internet, answering that question is relatively easy: printing at the maximum size allowed by the sensor resolution and expecting a sharp print examined close is the same as examining a 1:1 detail of an image on-screen (the computer screen dpi resolutions are comparable to average prints, remember?). And the answer is: there are very few 24x36 lenses which are perfectly sharp corner to corner on the D800 / A7r especially if you need a wide-angle lens. So here again the practical answer may be to limit yourself to 24" prints. On medium format, the answer may be different since there are many lenses which are sharp corner to corner on the highest resolution MF cameras, including wide angle lenses.

What is your printer Jerome? …is it a pigment one that prints on 720, or 1440, or 2880 dpi…? Also, you obviously mean ppi when you refer to size of print don't you? (i.e. how many camera pixels you'll use) at an inch of print, …don't you?
 

Jerome Marot

Well-known member
The 24" printer I was referring to is a Canon IPF6450. Yes, I refer to the number of pixels from the camera, not to the number of ink droplets the printer uses for emulating colours for one particular element of image.
 

Jerome Marot

Well-known member
Your question cannot be answered, since the value depends on the internals of the raster image processing software used. More than one is available (Canon even delivers two with the printer) and what they are doing internally is not really well documented.
 
Your question cannot be answered, since the value depends on the internals of the raster image processing software used. More than one is available (Canon even delivers two with the printer) and what they are doing internally is not really well documented.
Never the less your printer drivers upsample internally to either 300 or 600dpi whatever the input might be in ppi... do you find this correct? ….Also, it is more probable that it is doing so at 600dpi, since it: 1. "misses" a pip as it moves the paper in y-direction (plots on x) and, 2. 300dpi sounds a bit limiting for ultra demanding work.. (painting reproduction for instance). Do you find the above correct?

So… we end up with a 2400x2400dpi printer that misses one of the vertical dpi and prints an (up)sampled 600ppi input at 2400x1200dpi… correct again?
 
Top