voidshatter
Member
Q: Are you a pro, a dealer, or trying to sell your gear at a higher price?
A: None of the above. I am a tech geek, and by being honest to reveal all drawbacks I even risk selling my gear for less should I need to sell in the future. For such reason, dealers tend to hate me.
Q: Why CMOS?
A: Because:
a) Currently only the Sony CMOS (H5D-50C, H5D-200CMS, CFV-50C, IQ250, IQ150, Credo50, 645Z) can match up against the Nikon D800E / Sony A7R in terms of dynamic range. At pixel peeping level for high contrast scenes, the IQ280 and the IQ260 gets inferior image quality in the shadow when compared against the Nikon D800E / Sony A7R. If you are used to Canon you may not appreciate the Sony CMOS, but if you are used to Nikon / Sony you would definitely like the Sony CMOS!
b) Live View is usable in both bright and dark conditions for critical focus and composition without harassment.
c) The fab process of the Sony CMOS is mature enough to avoid tiling issues.
d) The Sony CMOS can turn off darkframe NR for long exposure shots with virtually no impact on image quality, allowing twice the chance to capture great light during sunset / sunrise when compared against all CCD sensors.
e) Higher usable ISO means handheld shots would be possible under certain situations.
Q: I am a fan of larger sensors. Why should I bother with a small crop sensor?
A: I am not persuading you to buy the small crop sensor, but you may find useful information here to predict what would happen if a fullframe 645 CMOS sensor is ready - what would happen to the current Rodenstock HR Digaron lenses?
Q: Canon has announced their 50 MP 5DSR. Why should I bother with something like an IQ250?
A: If you stitch by movements with the IQ250 you get the same / similar pixel count as if you stitch with the IQ280 (i.e. in the >100 MP territory). It is not easy to do parallax free stitching with a Canon TS-E lens. According to the dpreview interview the 5DSR would perform similarly as the previous Canon sensors in terms of dynamic range and SNR. The Rodenstock HR lenses also perform at levels higher than or equal to the Zeiss Otus lenses, especially for wide angles. It is also easier to sort out filters for the 23HR, 32HR and 40HR than for the Canon 17mm f/4 L TS-E lens when you need perspective control by movements for long exposure shots (the Fotodiox filter is a huge monster to carry around).
Q: Why don't you spent more time taking real pictures rather than testing gear?
A: While I do shoot a few images I am not a pro. I am a gearhead and I enjoy the process of pursuing ultimate image quality, as well as shooting something that the others cannot easily shoot due to gear constraints. The process of experiments and discovery is fun, and the advancement of technology is exciting!
################################################################################
Lenses tested this time (courtesy of Linhof & Studio!):
Rodenstock HR Digaron-S 23 mm f/5,6 (on ALPA 12 MAX with centerfilter)
Rodenstock HR Digaron-W 32 mm f/4 (on Linhof Techno)
Rodenstock HR Digaron-W 40 mm f/4 (on ALPA 12 MAX)
Rodenstock HR Digaron-W 70 mm f/5,6 (on Linhof Techno)
Rodenstock HR Digaron-SW 90 mm f/5,6 (on Linhof Techno)
################################################################################
First of all here are the blue sky desaturation tests (white balance: Daylight, Kelvin 5000, Tint 0; highlight -33, shadow +33):
I would say that the desaturation does not bother me at all. Within the manufacturer specified image circle all lenses seem to hold strong and if there is any observed desaturation personally I would just fix it in post processing with a few mouse clicks. It may not be suitable if you do scientific images and color fidelity is critical, but for landscape work I would assume it to be OK. The wide angles seem to have a bit of desaturation in the corners when shifted. The 70HR-W seems to have some uncorrected magenta cast when shifted to the extreme. The 90HR-SW can do well even without LCC!
################################################################################
Secondly, here are the mazing artifacts tests:
What are mazing artifacts and when does it happen for the Sony CMOS sensor? According to my experience to date (at the time of writing all LCC were done with Capture One v8.1), they most likely only occur when you shoot interior and there is a white wall (or other high key / simple texture) right within the region where the LCC shot has the purple-blue cast. It can be observed if you do pixel peeping at 100% pixel level. I have never observed such problem when I shoot outdoors (e.g. the blue sky). Below is so far the most stressful stress test that I can find for this phenomena:
How do movements affect these mazing artifacts? Below shows that movements along the longer edge of the Sony CMOS sensor do not affect the amount of mazing artifacts. (Actually later you would see that only movements along the shorter edge of the Sony CMOS sensor would have an impact):
So, what's the safe range of movements along the longer edge of the Sony CMOS sensor to avoid these mazing artifacts? Here are the tests results of the wide angles and you could evaluate the usable image circle using Digital Transitions' Tech Camera Visualization Tools by yourself:
Personally I would say that the safe range of rise (i.e. fall of IQ250 back for landscape orientation) is 6mm for the 23HR, 12mm for the 32HR and 16mm for the 40HR if you are picky with interior shots. However, the mazing artifacts on the 32HR are not prominent and it might even be usable beyond 16mm rise.
The 23HR on the IQ250 is like 17mm equivalent as of the 35mm format, which is even a bit wider than the 32HR on the IQ260. It is still the widest solution for long exposure shots with perspective control by movements without having to carry huge monstrous Fotodiox filters.
The 32HR on the IQ250 does not really need a centerfilter desperately as the dynamic range of the Sony CMOS sensor is really impressive. The lens is a beast and may continue to hold strong value if the fullframe 645 CMOS sensor is going to be based on the same technology as of the IQ250. (We do not consider the microlens array offset yet.) The only possible drawback is the mazing artifact issue if one is picky, but hopefully future software would be able to better compensate for this.
The 40HR on the IQ250 performs strongly and is versatile. If you own the 23HR then the 40HR is a great complementary. Personally I would assume it to be a safer bet for future fullframe 645 CMOS sensors.
The 70HR is sharp in the center (on par with the 90HR-SW) but the image quality degrades with movements towards the edge of the image circle and needs to be stopped down. Also it has color casts on the IQ250 when moved to the extreme. However given the size, weight and price I would still say it is a very reasonable choice!
The 90HR-SW is perhaps the new standard. It's a beasty lens as the 32HR. It has virtually no color cast on the Sony CMOS sensor as I was unable to hit the edge of the image circle with a crop sensor given the range of movements offered by the camera body. You can't go wrong with it except that it is bigger in size, heavier and more expensive than the 70HR.
So, what to expect next? Wide angles and sensor performance have been spear and shield for long. Symmetric design of wide angles is surely appreciated but for a sensor to achieve better dynamic range and high ISO performance there is a compromise that the lightwell has to be deep and the light shield has to be short, hence making the sensor more susceptible to crosstalk issues (color casts and mazing artifacts). Crosstalk is just one of the very many (digital) optical defect such like vignetting, MTF degradation towards the edge, distortion, chromatic aberration etc. With the Sony CMOS sensor you lose a bit of color or movement range but you gain the dynamic range, SNR etc, and when compared against the Canon / Nikon tilt-shift glasses you gain resolution as well (i.e. corner sharpness). Personally I would assume that this Sony CMOS + Rodenstock HR setup is of least compromise for landscape shots. For interior shots or other applications you might still prefer the larger CCD sensors for a generation or so.
The Samsung NX1 has introduced the first APSC back-illuminated CMOS sensor which is ray-angle friendly but I would assume it to be unable to achieve the same dynamic range as the Sony CMOS sensors do. This is one route for the 645 fullframe CMOS sensor (wide angle friendly, but less DR and SNR).
The D7000, D800E, A7R, IQ250 (similar Sony IMX094 sensor tech) tech is another route for a 80 MP 645 fullframe CMOS sensor (similar wide angle compatibility as tested this time, but great DR and SNR).
The NEX7, D5300, A7R-II tech is another route for a 120 MP 645 fullframe CMOS sensor (great DR and SNR, but will probably be incompatible with all current ultra wide angles).
A: None of the above. I am a tech geek, and by being honest to reveal all drawbacks I even risk selling my gear for less should I need to sell in the future. For such reason, dealers tend to hate me.
Q: Why CMOS?
A: Because:
a) Currently only the Sony CMOS (H5D-50C, H5D-200CMS, CFV-50C, IQ250, IQ150, Credo50, 645Z) can match up against the Nikon D800E / Sony A7R in terms of dynamic range. At pixel peeping level for high contrast scenes, the IQ280 and the IQ260 gets inferior image quality in the shadow when compared against the Nikon D800E / Sony A7R. If you are used to Canon you may not appreciate the Sony CMOS, but if you are used to Nikon / Sony you would definitely like the Sony CMOS!
b) Live View is usable in both bright and dark conditions for critical focus and composition without harassment.
c) The fab process of the Sony CMOS is mature enough to avoid tiling issues.
d) The Sony CMOS can turn off darkframe NR for long exposure shots with virtually no impact on image quality, allowing twice the chance to capture great light during sunset / sunrise when compared against all CCD sensors.
e) Higher usable ISO means handheld shots would be possible under certain situations.
Q: I am a fan of larger sensors. Why should I bother with a small crop sensor?
A: I am not persuading you to buy the small crop sensor, but you may find useful information here to predict what would happen if a fullframe 645 CMOS sensor is ready - what would happen to the current Rodenstock HR Digaron lenses?
Q: Canon has announced their 50 MP 5DSR. Why should I bother with something like an IQ250?
A: If you stitch by movements with the IQ250 you get the same / similar pixel count as if you stitch with the IQ280 (i.e. in the >100 MP territory). It is not easy to do parallax free stitching with a Canon TS-E lens. According to the dpreview interview the 5DSR would perform similarly as the previous Canon sensors in terms of dynamic range and SNR. The Rodenstock HR lenses also perform at levels higher than or equal to the Zeiss Otus lenses, especially for wide angles. It is also easier to sort out filters for the 23HR, 32HR and 40HR than for the Canon 17mm f/4 L TS-E lens when you need perspective control by movements for long exposure shots (the Fotodiox filter is a huge monster to carry around).
Q: Why don't you spent more time taking real pictures rather than testing gear?
A: While I do shoot a few images I am not a pro. I am a gearhead and I enjoy the process of pursuing ultimate image quality, as well as shooting something that the others cannot easily shoot due to gear constraints. The process of experiments and discovery is fun, and the advancement of technology is exciting!
################################################################################
Lenses tested this time (courtesy of Linhof & Studio!):
Rodenstock HR Digaron-S 23 mm f/5,6 (on ALPA 12 MAX with centerfilter)
Rodenstock HR Digaron-W 32 mm f/4 (on Linhof Techno)
Rodenstock HR Digaron-W 40 mm f/4 (on ALPA 12 MAX)
Rodenstock HR Digaron-W 70 mm f/5,6 (on Linhof Techno)
Rodenstock HR Digaron-SW 90 mm f/5,6 (on Linhof Techno)
################################################################################
First of all here are the blue sky desaturation tests (white balance: Daylight, Kelvin 5000, Tint 0; highlight -33, shadow +33):
I would say that the desaturation does not bother me at all. Within the manufacturer specified image circle all lenses seem to hold strong and if there is any observed desaturation personally I would just fix it in post processing with a few mouse clicks. It may not be suitable if you do scientific images and color fidelity is critical, but for landscape work I would assume it to be OK. The wide angles seem to have a bit of desaturation in the corners when shifted. The 70HR-W seems to have some uncorrected magenta cast when shifted to the extreme. The 90HR-SW can do well even without LCC!
################################################################################
Secondly, here are the mazing artifacts tests:
What are mazing artifacts and when does it happen for the Sony CMOS sensor? According to my experience to date (at the time of writing all LCC were done with Capture One v8.1), they most likely only occur when you shoot interior and there is a white wall (or other high key / simple texture) right within the region where the LCC shot has the purple-blue cast. It can be observed if you do pixel peeping at 100% pixel level. I have never observed such problem when I shoot outdoors (e.g. the blue sky). Below is so far the most stressful stress test that I can find for this phenomena:
How do movements affect these mazing artifacts? Below shows that movements along the longer edge of the Sony CMOS sensor do not affect the amount of mazing artifacts. (Actually later you would see that only movements along the shorter edge of the Sony CMOS sensor would have an impact):
So, what's the safe range of movements along the longer edge of the Sony CMOS sensor to avoid these mazing artifacts? Here are the tests results of the wide angles and you could evaluate the usable image circle using Digital Transitions' Tech Camera Visualization Tools by yourself:
Personally I would say that the safe range of rise (i.e. fall of IQ250 back for landscape orientation) is 6mm for the 23HR, 12mm for the 32HR and 16mm for the 40HR if you are picky with interior shots. However, the mazing artifacts on the 32HR are not prominent and it might even be usable beyond 16mm rise.
The 23HR on the IQ250 is like 17mm equivalent as of the 35mm format, which is even a bit wider than the 32HR on the IQ260. It is still the widest solution for long exposure shots with perspective control by movements without having to carry huge monstrous Fotodiox filters.
The 32HR on the IQ250 does not really need a centerfilter desperately as the dynamic range of the Sony CMOS sensor is really impressive. The lens is a beast and may continue to hold strong value if the fullframe 645 CMOS sensor is going to be based on the same technology as of the IQ250. (We do not consider the microlens array offset yet.) The only possible drawback is the mazing artifact issue if one is picky, but hopefully future software would be able to better compensate for this.
The 40HR on the IQ250 performs strongly and is versatile. If you own the 23HR then the 40HR is a great complementary. Personally I would assume it to be a safer bet for future fullframe 645 CMOS sensors.
The 70HR is sharp in the center (on par with the 90HR-SW) but the image quality degrades with movements towards the edge of the image circle and needs to be stopped down. Also it has color casts on the IQ250 when moved to the extreme. However given the size, weight and price I would still say it is a very reasonable choice!
The 90HR-SW is perhaps the new standard. It's a beasty lens as the 32HR. It has virtually no color cast on the Sony CMOS sensor as I was unable to hit the edge of the image circle with a crop sensor given the range of movements offered by the camera body. You can't go wrong with it except that it is bigger in size, heavier and more expensive than the 70HR.
So, what to expect next? Wide angles and sensor performance have been spear and shield for long. Symmetric design of wide angles is surely appreciated but for a sensor to achieve better dynamic range and high ISO performance there is a compromise that the lightwell has to be deep and the light shield has to be short, hence making the sensor more susceptible to crosstalk issues (color casts and mazing artifacts). Crosstalk is just one of the very many (digital) optical defect such like vignetting, MTF degradation towards the edge, distortion, chromatic aberration etc. With the Sony CMOS sensor you lose a bit of color or movement range but you gain the dynamic range, SNR etc, and when compared against the Canon / Nikon tilt-shift glasses you gain resolution as well (i.e. corner sharpness). Personally I would assume that this Sony CMOS + Rodenstock HR setup is of least compromise for landscape shots. For interior shots or other applications you might still prefer the larger CCD sensors for a generation or so.
The Samsung NX1 has introduced the first APSC back-illuminated CMOS sensor which is ray-angle friendly but I would assume it to be unable to achieve the same dynamic range as the Sony CMOS sensors do. This is one route for the 645 fullframe CMOS sensor (wide angle friendly, but less DR and SNR).
The D7000, D800E, A7R, IQ250 (similar Sony IMX094 sensor tech) tech is another route for a 80 MP 645 fullframe CMOS sensor (similar wide angle compatibility as tested this time, but great DR and SNR).
The NEX7, D5300, A7R-II tech is another route for a 120 MP 645 fullframe CMOS sensor (great DR and SNR, but will probably be incompatible with all current ultra wide angles).
Last edited:
