IQ 180, or microstep-16x from an old 22mpx back?

[fotometria gr]

 Is the quality of a 80mpx MFDB comparable to the 16x image created by an Imacon 528c, Hassy CF22MS or Sinarback HR54S, when shooting stills? Has anybody done the comparison? In theory the 16x image from the 22mpx back should be noticeably better. Since I own such a back and do 16X stills with it that includes ultra demanding painting reproduction, I do care to know. Thanks, Theodoros. www.fotometria.gr

[EricWHiss]

I've done the comparison between the 528 CF and AFi-ii 12 (Aptus 12) and the AFi-ii 12 vs CF 39MS.    I have managed to perform all the the tests using the same lens eliminating that as a variable.

Here's a summary of my results:
CF 528 vs Aptus 12     CF wins easily in detail, color tonality and DR in microstep mode, loses in convenience.   If you shoot say a couple peaches, apples or grapes with both, you'll easily see how much better the microstep files are in terms of color and tonality.  Areas that look just like one color with the single shot aptus 12, have all kinds of fine texture and shades in the microstep file.   In single shot mode the aptus 12 / IQ 180 obviously will record finer detail, however the 22mp backs do get more dynamic range and have a grainless file at base ISO unlike the 80mp dalsa chips. Color? The purveyors of the new backs say its improved...  I haven't scientifically tested this, but the older 9um backs have a great look.

CF 39 MS vs Aptus 12    Too close to call on fine detail when CF in multishot but maybe the Aptus 12 edges out. Obviously in single shot mode the CF is beaten by the 80mp backs but again the differences are not as large as one might expect.     CF 39MS has more DR by about .5 to 1 stop in single shot mode, and  1.5 to 2 stops in Multishot mode.

I think of the CF 528 in microstep as the best back I have ever used or seen.  But its a limited set of applications where you can shoot microstep.   I'm lucky to have both to use.   

In general all the above really compares the situations where multishot is applicable - art reproduction, food photography, product work, still life, etc  - its not going to work for everything.    The really great thing about the modern 80mp backs is how versatile they are - getting close in a single shot, and one can shoot at higher ISO's.   But still for ultimate quality in a static scene, the microstep backs are very hard to beat.

[henrikfoto]

Hi Eric!

Have you tested the CF 528 against the Sinar 54H?

Henrik

[EricWHiss]

Hi Henrik,
I haven't used the 54H but have heard very many good things.
Eric

[fotometria gr]

Hi Henrik,
I haven't used the 54H but have heard very many good things.
Eric

I guess they will prove to be equal, but obviously the versatility of the 528c/cf22ms is much higher in single shot and the flexcolor a better and more stable tool. Regards, Theodoros. www.fotometria.gr

[fotometria gr]

I've done the comparison between the 528 CF and AFi-ii 12 (Aptus 12) and the AFi-ii 12 vs CF 39MS.    I have managed to perform all the the tests using the same lens eliminating that as a variable.

Here's a summary of my results:
CF 528 vs Aptus 12     CF wins easily in detail, color tonality and DR in microstep mode, loses in convenience.   If you shoot say a couple peaches, apples or grapes with both, you'll easily see how much better the microstep files are in terms of color and tonality.  Areas that look just like one color with the single shot aptus 12, have all kinds of fine texture and shades in the microstep file.   In single shot mode the aptus 12 / IQ 180 obviously will record finer detail, however the 22mp backs do get more dynamic range and have a grainless file at base ISO unlike the 80mp dalsa chips. Color? The purveyors of the new backs say its improved...  I haven't scientifically tested this, but the older 9um backs have a great look.

CF 39 MS vs Aptus 12    Too close to call on fine detail when CF in multishot but maybe the Aptus 12 edges out. Obviously in single shot mode the CF is beaten by the 80mp backs but again the differences are not as large as one might expect.     CF 39MS has more DR by about .5 to 1 stop in single shot mode, and  1.5 to 2 stops in Multishot mode.

I think of the CF 528 in microstep as the best back I have ever used or seen.  But its a limited set of applications where you can shoot microstep.   I'm lucky to have both to use.  

In general all the above really compares the situations where multishot is applicable - art reproduction, food photography, product work, still life, etc  - its not going to work for everything.    The really great thing about the modern 80mp backs is how versatile they are - getting close in a single shot, and one can shoot at higher ISO's.   But still for ultimate quality in a static scene, the microstep backs are very hard to beat.

Your findings are about what i suspected, thanks for sharing. The higher resolution must have a lot to do with the fact, that the lens analysis can obviously cope better with a 22mpx sensor rather than stretch to "feed" an 80mpx sensor. The true 88mpx file when in microstep is a sensor outcome, the resolution is not required from the lens, isn't it? OTOH, when high resolution is needed in single shot, it's usually because of the large print required, which can be solved by using film. The undeniable finer detail of the 80mpx MFDB may be welcome, but surely no more than other aspects of photography, like DR or "weight" of the scene, that film may give. I feel that low-light shooting is better done with a top DSLR, since it will be accompanied by faster lenses and even more sensitivity latitude. I don't deny that there are photographers who need to shoot high resolution all the time and thus take full advantage of such a HighRes back, but surely they can't be that many (especially if cost is added to the equation). Usually HighRes is needed on stills where detail, color and dynamic range is of most importance, thus at least multishot (4x) should be a must for such kind of work, with 16x being the ultimate as you state. It seems that recent backs don't do 16x anymore, because engineers must have great difficulty to control a smaller than 4.5μm move that is required, but I wonder why they didn't apply that superb technology to the Dalsa 22mpx sensor (which is still in production). I guess that Hasselblad's new H4-50MS (or 200MS whatever the name is) solution with both multishot and microstep must be the ultimate solution currently that keeps a balance that serves both single-shot and multishot High-Res requirements in stills. It may be best for them to introduce the back as a CF version, for the non H users to share its qualities. Regards, Theodoros. www.fotometria.gr

[EricWHiss]

Yes, I think you are correct that the multishot backs are easier on the lenses.  Each frame is still collected with the 9um wells, prior to being processed.    One very important outcome of this that I left out in the earlier post is that the photographer can stop down much further with multishot and microstep. 

with the 80mp backs - diffraction can be seen after f/8 though not really bad until f/16 and smaller but with the 22mp backs diffraction isn't getting bad until f/22 on a single frame and when using multishot one can shoot out to f/32 .  I've tested this very carefully - the multishot files have at least a 2 stop advantage in aperture settings which can be important for macro work or product work where you need all the DOF possible.  Now because the 22mp backs have one stop over the 80mp backs, you really would have 3 stops advantage when using multishot or microstep.   It's quite amazing.

[fotometria gr]

Yes, I think you are correct that the multishot backs are easier on the lenses.  Each frame is still collected with the 9um wells, prior to being processed.    One very important outcome of this that I left out in the earlier post is that the photographer can stop down much further with multishot and microstep. 

with the 80mp backs - diffraction can be seen after f/8 though not really bad until f/16 and smaller but with the 22mp backs diffraction isn't getting bad until f/22 on a single frame and when using multishot one can shoot out to f/32 .  I've tested this very carefully - the multishot files have at least a 2 stop advantage in aperture settings which can be important for macro work or product work where you need all the DOF possible.  Now because the 22mp backs have one stop over the 80mp backs, you really would have 3 stops advantage when using multishot or microstep.   It's quite amazing.

Oh come on Eric! is that all of it? I like things hot! Where are the rest of the answers? we don't play safe here, we try to push things to improve our photography isn't it? Its the 75% of the post up there that remains unsupported! You can help Eric! Theodoros. www.fotometria.gr

[EricWHiss]

Well, what else do you need to know?   My answer is one might need both options, always test yourself in your own conditions.   I have both cameras and mostly use the AFi-ii 12.  I know I can get the best results with the microstep, but sometimes out of convenience I use the AFi-ii 12.   For the studio, a really great setup is the CF 528 plus a technical camera and great digital lenses in electronic shutter.  I have the X-Act2 and rollei electronic shutters.

[fotometria gr]

Well, what else do you need to know?   My answer is one might need both options, always test yourself in your own conditions.   I have both cameras and mostly use the AFi-ii 12.  I know I can get the best results with the microstep, but sometimes out of convenience I use the AFi-ii 12.   For the studio, a really great setup is the CF 528 plus a technical camera and great digital lenses in electronic shutter.  I have the X-Act2 and rollei electronic shutters.

Sorry Eric, I thought you could comment on my other remarks or suggestions, please forgive my insist! I do like things hot. Regards, Theodoros. www.fotometria.gr

[EricWHiss]

One other point to be made.  Non interpolated resolution can't be compared straight across to a bayer interpolated resolution.  It's not 3x the file resolution like Foveon suggests but then again its not 1x either.  The 88mp non-interpolated (microstep) file just has a lot more there than the 80mp interpolated file.  How much more, I'll leave to the likes of Erik K, B Janes, or others to quantify.   Maybe if I have time next week, I'll post a few comparison images with crops because this will help illustrate the differences better.

[fotometria gr]

One other point to be made.  Non interpolated resolution can't be compared straight across to a bayer interpolated resolution.  It's not 3x the file resolution like Foveon suggests but then again its not 1x either.  The 88mp non-interpolated (microstep) file just has a lot more there than the 80mp interpolated file.  How much more, I'll leave to the likes of Erik K, B Janes, or others to quantify.   Maybe if I have time next week, I'll post a few comparison images with crops because this will help illustrate the differences better.

No need to post any files Eric, at least not for me, I know what you suggest and I agree. The 88mpx file is during capture, interpolation is through process..., no comparison whatsoever! But what a great man achievement! ...ha? Regards, Theodoros. www.fotometria.gr

[TH_Alpa]

hi Theodoros,

Just a little comment concerning lenses and 16-shot mode with multi-shot DBs:

It is better to use the digital lenses (HR lenses) rather than usual lenses, this will improve the quality drastically. Don't forget that the shots in a 16-shot mode are done by moving the sensor by 1/2 a pixel each time; 4 times at o-position, 4 times to the left, 4 times down, 4 times to the right. Even the light from the space in-between the pixels are captured by moving half a pixel each time. Therefore it is important to have the best possible lens.

Best regards
Thierry

[fotometria gr]

hi Theodoros,

Just a little comment concerning lenses and 16-shot mode with multi-shot DBs:

It is better to use the digital lenses (HR lenses) rather than usual lenses, this will improve the quality drastically. Don't forget that the shots in a 10-shot mode are done by moving the sensor by 1/2 a pixel each time; 4 times at o-position, 4 times to the left, 4 times down, 4 times to the right. Even the light from the space in-between the pixels are captured by moving half a pixel each time. Therefore it is important to have the best possible lens.

Best regards
Thierry

I see the difference for the sensor, but I don't see the difference for the lens, sorry if I am wrong but the lens, IM(engineering)O still records 22mpx! Its the sensor that divides the image in four half side-size rectangles, the lens has send info for one pixel and is one pixel that records that image, the difference in the recording comes from the info from the 1/4th of the neighbor pixel, which has also recorded a 4/4ths image and shares a quarter of it. Am I wrong? Each of the shots is recorded on a 22mpxs sensor, isn't it? Please explain, I may be wrong but it doesn't feel right the way you put it. Regards, Theodoros. www.fotometria.gr

[fotometria gr]

Even the light from the space in-between the pixels are captured by moving half a pixel each time. Therefore it is important to have the best possible lens.

Best regards
Thierry

How can this be? the lens doesn't move, the sensor does! so there is no CAPTURE of light between pixels, the light between pixels may fall in the next pixel as the sensor moves and thus affect the recording of this next pixel, but the effect is still for the WHOLE pixel, so lens resolution is not affected! Do I understand something wrong here? I will really feel stupid if I am wrong on this! But it looks to me (up to now) that I may have catch you a bit ...misinformed, I've seen that on web before and thought "what?". Regards, Theodoros. www.fotometria.gr

[TH_Alpa]

Theodoros,

You misunderstood me and did not catch me misinformed, and probably I wasn't clear in my explanation: the sensor moves by half a pixel vs. by a full pixel when using the 4-shot mode. With the 4-shot mode one does not register the information in-between the pixels. When one now moves the sensor by half a pixel, each pixel will place itself over the gap between the pixels, thus registering the information there. That's the big difference between the 2 modes.

I didn't say that lens resolution is affected, but that there is a difference in quality when using "normal" digital lenses vs. HR digital lenses: make the test.

Best regards
Thierry

How can this be? the lens doesn't move, the sensor does! so there is no CAPTURE of light between pixels, the light between pixels may fall in the next pixel as the sensor moves and thus affect the recording of this next pixel, but the effect is still for the WHOLE pixel, so lens resolution is not affected! Do I understand something wrong here? I will really feel stupid if I am wrong on this! But it looks to me (up to now) that I may have catch you a bit ...misinformed, I've seen that on web before and thought "what?". Regards, Theodoros. www.fotometria.gr

[fotometria gr]

Theodoros,

You misunderstood me and did not catch me misinformed, and probably I wasn't clear in my explanation: the sensor moves by half a pixel vs. by a full pixel when using the 4-shot mode. With the 4-shot mode one does not register the information in-between the pixels. When one now moves the sensor by half a pixel, each pixel will place itself over the gap between the pixels, thus registering the information there. That's the big difference between the 2 modes.

I didn't say that lens resolution is affected, but that there is a difference in quality when using "normal" digital lenses vs. HR digital lenses: make the test.

Best regards
Thierry

But its still one pixel of info that is splitted in 2, isn't it Thierry? I mean this pixel (the one that splits the info and diffuses half of it to the pixel above it, while receiving diffusion from the one under it and then to the right, or left, or under) is a product of 22mpx analysis. I have no way to test anymore with my Sinar, I sold all the system but the body to a ...Horseman and have listed my P2 body for sale in the Lula sale thread (perhaps you may help here), I recently purchased a Fuji GX680, with which I share the back from my Contax, I no more require Sinar's extra capabilities. But I feel that the requirement for a good lens that you observe isn't because of the 88mpx file, but rather because we were used to use lenses with large circles in the past and then with digital, we required from them to keep their resolution down to a 37x49 image area, that's unfair don't you think? The HR lenses, are lenses aimed for 6x9 cameras that have a much smaller image circle and are in addition modern, that's why the image is better, not because of the 88mpx analysis, the 4x5 lenses found the 37x49 sensor too small and couldn't cope with the 22mpx resolution! Regards, Theodoros. www.fotometria.gr

[TH_Alpa]

Yes Theodoros, it is still 1 pixel of information, of course.

All I meant to say is that it is better to have this "in-between" information registered and the better with a better lens.

And yes, the 4x5" lenses don't have the necessary resolution power, that's why the digital lenses have ben built. And since it doesn't need the IC needed for the 4x5" format and for which LF lenses where made (to allow movements a LF camera is built for) one could concentrate on the resolution of those lenses. In a next step came the digital HR lenses, with even a higher resolution power and meant especially for multi-shot.

Best regards
Thierry

But its still one pixel of info that is splitted in 2, isn't it Thierry? I mean this pixel (the one that splits the info and diffuses half of it to the pixel above it, while receiving diffusion from the one under it and then to the right, or left, or under) is a product of 22mpx analysis. I have no way to test anymore with my Sinar, I sold all the system but the body to a ...Horseman and have listed my P2 body for sale in the Lula sale thread (perhaps you may help here), I recently purchased a Fuji GX680, with which I share the back from my Contax, I no more require Sinar's extra capabilities. But I feel that the requirement for a good lens that you observe isn't because of the 88mpx file, but rather because we were used to use lenses with large circles in the past and then with digital, we required from them to keep their resolution down to a 37x49 image area, that's unfair don't you think? The HR lenses, are lenses aimed for 6x9 cameras that have a much smaller image circle and are in addition modern, that's why the image is better, not because of the 88mpx analysis, the 4x5 lenses found the 37x49 sensor too small and couldn't cope with the 22mpx resolution! Regards, Theodoros. www.fotometria.gr

[Bart_van_der_Wolf]

But its still one pixel of info that is splitted in 2, isn't it Thierry?

Hi Theodoros,

Actually it is not a simple split. The physical center of the sampling position is shifted half a sensel/pixel, thus getting better response from signal that is originally best aligned in between 2 sensels. The effect is 2x higher resolution, although it is somewhat 'softened' by the half sensel overlap with its neigbors. This type of staggered sampling has a positive effect on the MTF curve and leads to reduced aliasing. The other shifts are used to sample R+G+B info at each sampling position, which also improves resolution, mostly for chromaticity and usually a little less for luminance.

All this can easily be demonstrated by taking an image of a resolution target that allows to quantify resolution, e.g. a target like the one I made available for free. See also this thread.
 
Cheers,
Bart

[fotometria gr]

Yes Theodoros, it is still 1 pixel of information, of course.

All I meant to say is that it is better to have this "in-between" information registered and the better with a better lens.

And yes, the 4x5" lenses don't have the necessary resolution power, that's why the digital lenses have ben built. And since it doesn't need the IC needed for the 4x5" format and for which LF lenses where made (to allow movements a LF camera is built for) one could concentrate on the resolution of those lenses. In a next step came the digital HR lenses, with even a higher resolution power and meant especially for multi-shot.

Best regards
Thierry

I am sorry I misunderstood you Thierry, in the begging of this conversation I thought you was saying that because of the 88mpx final image, the lens should be able to cope with 88mpx of resolution, which you didn't. I suppose you would also agree that part of the sharpness difference that Eric has noticed when comparing the 80mpx sensor with the 16x image of the 22mpx sensor, is because the lens he used found the big sensor much more demanding to cope with. Regards, Theodoros.