The GetDPI Photography Forum

Great to see you here. Join our insightful photographic forum today and start tapping into a huge wealth of photographic knowledge. Completing our simple registration process will allow you to gain access to exclusive content, add your own topics and posts, share your work and connect with other members through your own private inbox! And don’t forget to say hi!

wide lens IQ 180 shoot out: Digitar 28XL versus Digaron 32HR-W

rupho

New member
dear all,
beware of the new Schneider 28mmXL Digitar on the IQ180.
It got uncontrollable color cast EVEN WITHOUT SHIFT!

A fellow getDPI forum member Chris Hauser and myself met in Hong Kong last weekend and had a fun shoot out comparing the Schneider 28XL and 43XL with the Rodenstock 32HR-W and 40HR-W. He was on an Arca with the Rodies and myself on an Alpa with the Schneider both equipped with IQ180s.

to simplify our findings I simply grouped those images and made screen shots. The lens and shift amount can be found below the respective images.

The lens test were as follows:
- 32HR shifted up: 0, 5, 10 , 15 mm plus corresponding LCCs
- 28XL shifted up: 0, 5, 10 , 15 mm plus corresponding LCCs
- 32HR shifted left: 5, 10 , 15 mm plus corresponding LCCs
- 28XL shifted left: 5, 10 , 15 mm plus corresponding LCCs

Please note while both the Arca and the Alpa allowed for more than 15 mm upward shift the 28XL due to its proximity to the sensor pane can't be shifted higher than 15mm on the Alpa. The 32HR peaked at about 19mm when we reached the image circle of that lens.
Also the camera were right next to each other and there we minimal light changes during the 15 minutes of shooting these test.
I deliberately did NOT make any changes to the LCC or the image itself (no adjustments, straight out the camera).
All were shot with the same IQ180 on f11 1/8sec (1/4 for LCC) and WB set to Daylight.
Despite the lack of any scientific procedure I think the images speak a clear message:
the 28 Digitar as great as it is on other backs is NOT recommended on the IQ. I had to buy the bullet and sold it right away.

On another note: thanks to Jack and Guy for this forum plus everybody's else's accumulated wealth of information & not to mention the fact that you bring folks together as well.
Especially in this part of the world (China) its hard to initiate a shoot out with such rare equipment specs.
Special thanks also to Christopher for making me loose a few thousand bucks this weekend on the trade in :)
Hope some of you can avoid this costly mistake.

Grischa
 

archivue

Active member
"the 28 Digitar as great as it is on other backs is NOT recommended on the IQ."

at least on the IQ 180...

Anyone test it on the two others IQ ?
 

Terry

New member
I guess I'm not sure I understand what you are showing for the LCC's. These look unprocessed. Are you saying when you use C1 and process the LCC using the wide angle analysis even at zero shift the 28mm doesn't clean up and can't be fully corrected?

Would you please post the 28 no shift image after the LCC is applied.

Thanks,
Terry
 

rupho

New member
Rupho, where do you buy (and sell) your Alpa gear?

Paul
Paul, being in Hong Kong I buy my Alpa gear though Beijing. Fabulous and extremely knowledgeable team there. so far I have not sold a thing except for the Schneider 28mm lens after 3 month of use.:angry:



I guess I'm not sure I understand what you are showing for the LCC's. These look unprocessed. Are you saying when you use C1 and process the LCC using the wide angle analysis even at zero shift the 28mm doesn't clean up and can't be fully corrected?

Would you please post the 28 no shift image after the LCC is applied.

Thanks,
Terry
Terry, what I am showing in regards to the LCCs is the fact that the 28XL displays extreme color casts even unshifted . I think uncorrected LCCs tell the story of what where the limits are. I attached the 28XL unshifted and corrected. I think unshifted this lens works but even then the LCC makes the edges "lifeless" there is not much color left. Also I pushed the Light Falloff to 50% otherwise the image would have been rather bright.
I also attached a 100% screen shot that shows the side of that very same image displaying bending even unshifted.
The point I am trying to make is: this lens is only correctable unshifted and the stated image circle of 105mm is unrealistic on an IQ180.
I guess everybody can draw their own conclusions of whether this lens works for them. In terms of size , sharpness and built quality its absolutely great , but for some reason its causing color cast havoc to an extend where its not worth to bother with IMHO.
Grischa

Gazwas,
we also shot on the 43XL and 40 Hr. the differences were not nearly as severe compared to the 28XL / 32HR.
Though there was more color coast creeping in on the 43mm at around 15mm on the long side (again with an IQ180) so I did not bother posting it

Grischa
 

yatlee

Member
Hi Grischa,

Very informative test. I'm surprise to see so many member are actually from Hong Kong! I met Chris on Tuesday for a shoot as well.

Yat
 

chrismuc

Member
This color cast topic with the iQ180 and wide angle lenses - symmetrical design like Schneider but also asymmetrical designs like Rodenstock - seems to me a terrible design flaw of the iQ180 back.

In comparison, the Canon sensors of the 5D2 and 1Ds3 with very wide angle lenses like the 24TSE and 17TSE at full horizontal shift show no color cast at the shifted edges.

The optical construction in front of the Dalsa sensor of iQ180 seems too deep (maybe it was also not such a good idea not to use micro lenses like Canon does) and at steep angles not all frequencies of the light are transmitted in the same way to the sensor pixels. With such strong color cast, a pp software solution is not very appropriate because noise is added and the cast compensation shifts the range of each color channel into a more non-linear area.

I'm wondering what Phase One/Dalsa is explaining regarding this serious topic and I would like to know what in detail they changed in sensor design from the iQ160/P65 to iQ180/AptusII 12 that make them work worse with wide angle lenses.
 

dougpeterson

Workshop Member
This color cast topic with the iQ180 and wide angle lenses - symmetrical design like Schneider but also asymmetrical designs like Rodenstock - seems to me a terrible design flaw of the iQ180 back.

In comparison, the Canon sensors of the 5D2 and 1Ds3 with very wide angle lenses like the 24TSE and 17TSE at full horizontal shift show no color cast at the shifted edges.

The optical construction in front of the Dalsa sensor of iQ180 seems too deep (maybe it was also not such a good idea not to use micro lenses like Canon does) and at steep angles not all frequencies of the light are transmitted in the same way to the sensor pixels. With such strong color cast, a pp software solution is not very appropriate because noise is added and the cast compensation shifts the range of each color channel into a more non-linear area.

I'm wondering what Phase One/Dalsa is explaining regarding this serious topic and I would like to know what in detail they changed in sensor design from the iQ160/P65 to iQ180/AptusII 12 that make them work worse with wide angle lenses.
The Canon 17TS and 24TS use strongly retrofocus designs which brings the lens away from the sensor and thus avoids most of these issues. They must, since neither Canon nor Nikon make a pro body without a mirror so the lens must be mounted in front of a mirror box. Even so a 7D with a 17TS will show color cast in the corners with large movements- don't take my word for it, just go shoot through a piece of white plexi and auto-level the resulting image. Notably AFAIK neither Canon nor Nikon have a software tool to deal with this (other than Capture One or other 3rd party software), and it will only become a larger issue when Canon and Nikon release next-gen bodies with full frame sensors with smaller pixels. Phase and Leaf have been aware of this fact of physics for a long time and have developed the LCC tool to address it as early as Capture One version 3.

Also, generally speaking Microlenses exacerbate the issue of color cast, not rectify it. Of course that's only generally speaking as "microlenses" can be pretty varied in design, composition, and purpose. In other words the lack of microlenses on the IQ180 is a good thing - not a bad thing - when it comes to color cast.

By the way you can in fact, can use a Canon 17TS or 24TS or any other such Canon or Nikon lens with a IQ180 (e.g. with the Hartblei Cam).

For those interested Capture Integration's testing should be posted by the end of the day tomorrow and includes the 28XL, 35XL, 43XL, 60XL, and the 70HR. It agrees with this test - the 28XL was definitely the lens in that group that performed the least well (no surprise there given how extreme a design that lens is). Otherwise it's mostly good news.

The opposite side of this argument of course is that the IQ180 gives the best dynamic range and color performance I've seen, while also giving the option for ISOs as low as 32 but also the best ISO3200 I've seen on any digital back. There is rarely a free lunch and a smaller usable image circle on the extreme wide angle XL lenses is one downside of the IQ180.

Watch for that test tomorrow!

Doug Peterson (e-mail Me)
__________________

Head of Technical Services, Capture Integration
Phase One Partner of the Year
Leaf, Leica, Cambo, Arca Swiss, Canon, Apple, Profoto, Broncolor, Eizo & More

National: 877.217.9870 *| *Cell: 740.707.2183
Newsletter | RSS Feed
Buy Capture One 6 at 10% off
 

chrismuc

Member
The Canon TSE 17mm with 12 mm shift on a 36 mm wide sensor relates to an image circle of 67 mm, the Rodenstock HR 32 mm with 13 mm shift on a 54 mm wide sensor relates to an image circle of 90 mm.

In this case the angle of view of the 12 mm shiftet Canon TSE 17 is about 35% larger than the angle of view of the 13 mm shifted Rodenstock HR 32!

Even if the Canon lens design is stronger retrofocus than the Rodenstock design, this much larger angle of view should relate to a steeper angle of light from the back of the lens to the sensor. Which is handled very well by the Canon sensor design but seems much less perfect by the Dalsa/Phase One/Leaf 80MP sensor design.
(And one must consider that the Canon sensor even includes an AA-filter and a ultrasonic cleaning system which enhances the thickness of the sensor and therefore makes it more difficult to deal well with the steep angles of light.)

I asked Stefan Steib from Hartblei how well the two TSE lenses perfom on the Hartblei cam with a 80MP back. I also aksed him how well (or worse) they perform regarding color cast compared to the P65 60MP back.

And that's still my question for Dough: How much stronger is the color cast of the IQ180 compared to the iQ160 (especially with shifted Rodenstock 28, 32 and 40 mm lenses) and what is the difference in sensor design between the two sensor generations. The reduced pixel width of the 80MP sensor does not directly cause this problem, IMO it must be the ratio of pixel diameter to sensor depth or other geometrical design changes of the sensor.

Looking forward your test results tomorrow!

Christoph
 

Terry

New member
The Canon TSE 17mm with 12 mm shift on a 36 mm wide sensor relates to an image circle of 67 mm, the Rodenstock HR 32 mm with 13 mm shift on a 54 mm wide sensor relates to an image circle of 90 mm.

In this case the angle of view of the 12 mm shiftet Canon TSE 17 is about 35% larger than the angle of view of the 13 mm shifted Rodenstock HR 32!

Even if the Canon lens design is stronger retrofocus than the Rodenstock design, this much larger angle of view should relate to a steeper angle of light from the back of the lens to the sensor. Which is handled very well by the Canon sensor design but seems much less perfect by the Dalsa/Phase One/Leaf 80MP sensor design.
(And one must consider that the Canon sensor even includes an AA-filter and a ultrasonic cleaning system which enhances the thickness of the sensor and therefore makes it more difficult to deal well with the steep angles of light.)

I asked Stefan Steib from Hartblei how well the two TSE lenses perfom on the Hartblei cam with a 80MP back. I also aksed him how well (or worse) they perform regarding color cast compared to the P65 60MP back.

And that's still my question for Dough: How much stronger is the color cast of the IQ180 compared to the iQ160 (especially with shifted Rodenstock 28, 32 and 40 mm lenses) and what is the difference in sensor design between the two sensor generations. The reduced pixel width of the 80MP sensor does not directly cause this problem, IMO it must be the ratio of pixel diameter to sensor depth or other geometrical design changes of the sensor.

Looking forward your test results tomorrow!

Christoph
Aren't you forgetting about how close these lenses are to the sensor. If you pull the lens away from the sensor isn't the angle of light hitting the sensor going to change. If you look at the IQ180 with the DF body and lenses and compare the 28mm lens for the DF vs the Schneider 28XL they will render very differently in terms of any color cast that needs to be corrected. This is similar to the issues that Leica had when developing the M8/M9 and having lenses that sit so close to the sensor.
 
The focal length of the lens can obviously has a relationship to the angle of view from the nodal point of the lens, but of course this is varied by the optical professors. The angle of light through the rear lens elements is only _equivalent_ and does not reflect the actual final exit angle from the rear of the lens, so two lenses of same focal length, one being actual the other being retro focus, will have vastly differnt final exit angles and therefore the same focal length can cause differing degrees of the colour cast problem. The wider the sensor the stronger the problem. Hence leica originally claiming that full frame digital was impossible. Although that seems like clever marketing now.

The lenses on the sensor can be angle to compensate to a degree, but this is a compromise as what is right for one lens is wrong another.

As light falloff is also a problem and under exposure over emphasis saturation, this also magnifies the problem.
 

Stefan Steib

Active member
HI Christoph/All

well it is interesting that all the predictions I have done about Microlens cornershading and color casts have come true to the worst intent possible.
This clearly shows that the concept of "true wideangles" as "digital lenses" (see Schneider whitepaper of some years ago) has failed. All the specialists saw this coming, Rodenstock has already reacted and made the 23mm retrofocus, so this was and is apparent.
A "correction" of this kind of color cast in Software may result in a visible usable image file, but spoken from a point of lithography view, these files will give a hell of headaches to everybody trying to convert to CMYK -correct and print these in a qualitative way. Of course if you only work for the internet or RGB that may get through somehow, but if I remember right, most of us are still getting their life paid by the printing industry products.

Ask any litho guy, he will tell you the truth about such files.

I can only invite everybody to take a look at the images we have done on our HCam-B1 with the Leaf aptus II 12R with no visible color cast nor chromas at 200% and movements with the Canon TS17mm and TS24mmII - all the people who saw this on the recent PCP Tour here in Germany were speachless - ask Yair Shahar if you don´t believe me - I have given the sample pics to him as well as I have given them to Canon Germany.

I have already contacted KN-Studiotechnik here in Munich and also Phase One to get a demo 180 IQ ASAP so we can proof this with the newest Phase One backs too.

There needs to be a change in concept for digital wideangle Pro Photography on cameras using these lenses, the classic approach is coming to a limit as we can clearly see now. It is most interesting that the solution is already worked on- it happens in the consumer areas with small chips and 16Mpix - it´s called NEX, MFT and things that people like Ricoh and Foveon do.
I´d also ask anybody to read Olympus four thirds whitepaper which was a milestone in my opinion about the future of digital photography.

The Pro Range was somehow sleeping, but I think this is a wakeup call.

Greetings from Munich
Stefan Steib - HCam.de
 

chrismuc

Member
Terry:
I only referred to the Rodenstock lenses because they also are retrofocus designs similar to the Canon TSE lenses. The Schneider lenses with their symmetrical design (which leads to a much shorter distance between back lens and sensor and therefore steeper angles of light) are obviously less suitable for digital backs.
BTW, the two Canon TSE lenses are also incredible free of distortion for a retrofocus design. Regarding this aspect Zeiss (18/21mm Distagons) and Rodenstock (HR-S/W wide angles) can still learn a lesson from their Japanese counterpart;-)
Christoph
 

cng

New member
A "correction" of this kind of color cast in Software may result in a visible usable image file, but spoken from a point of lithography view, these files will give a hell of headaches to everybody trying to convert to CMYK -correct and print these in a qualitative way. [...] Ask any litho guy, he will tell you the truth about such files.
Stefan, can you elaborate on this point?
 

yaya

Active member
There are 3 things still going for non-retrofocus lenses:

1. Lack of distortion
2. Lack of aberrations
3. Size/weight (and thus lower cost)

If retrofocus lenses can be designed so that they tick these 3 boxes then this will be a big step forward

Canon seems to have (almost) cracked it with the 24mm (the 17mm is a bit too extreme IMO) but you can only use it on an HCam
 

Stefan Steib

Active member
This is complicated but OK-I´ll try to make a short one.
Imagine the digital back as a signal to noise device. The most important part is getting a) best possible dynamics plus b) uniformity of this signal and c) getting a best match on the Luther condition (get as close to human color vision as possible).
The better these conditions are met the more information is rendered. Now if I have to partially correct color casts the information given to the image rendition is lacking best possible conditions in 1,2 and 3. An electronic correction will result in differing internal color gradients (dynamically spoken a loss of steps available for color transformations) applying image corrections will result in differing output on equal colors, gradation corrrections will result in posterisations etc.pp.This may not be that visible on the first look, but if you need to adapt a file for a difficult print color transformation you will fail (or have at least a lot of fun and time spoiled using these files).

I am sure a specialist science guy who is into image processing will be able to explain this with about 600 pages and formulas and maybe my shortening is to be discussed on several points because of simplification, but this is about it in a (well several) sentence(s).

regards
Stefan
 

Stefan Steib

Active member
Hi Yair

I think the definitions could be very clear: If Dalsa or Kodak or all the other Chipmakers would release clear data about which angle, which light falloff and which center shift for these chips/parameters would be acceptable before a file gets visible color cast like the shown above, everybody could count 1+1=2.
Rodenstock already quietly uses retrofocus designs (23mm) because they could not have built this lens otherwise ! And as it shows it is by far superior to the Schneider solutions available.
Technology marches on. Companies who will not keep pace will be left behind.
That simple.

regards
Stefan
 
Top