Canon 24TSE-II vs Rodenstock 23HR

[gerald.d]

Background -

I've done a set of images of the 24TSE on an HCam/IQ180 before, but received some feedback from some that perhaps I hadn't got the most out of the Canon, since I'd only gone up to f/11. You can see those earlier images here:

http://08.ae/panos/HCam/24mmTSE/24TSE.html

Bottom left are a set of thumbnails with self-explanatory tool-tips taking you through f/4 to f/11 with 0 and 5mm rise.

More recently, I attempted to shoot a comparison between the TSE and the HR, the results of which you can see here over at GetDPI:

http://www.getdpi.com/forum/487229-post36.html

That was at f/8, and again - it was commented that wasn't particularly fair on the Canon, and that the Canon shot was slightly over-exposed.

So, this morning, spurred on by Peter's thread here, I thought I'd have another go at comparing the two.

23HR on a Max, 24TSE on an FPS. Same IQ180 back used on both cameras.

Carrying out this kind of test is challenging, because there are many different variables to consider, and so many different ways of going about it. One thing I wanted to do this time around was to try to be consistent with exposing of the shots. For this reason, I chose not to use the center filter of the HR, but rather to shoot both lenses at identical apertures and shutter speeds, along with LCC's for every shot. This possibly may put the HR at a disadvantage? I don't really know, but I'm sure others will chip in with their thoughts on this approach. I also chose not to do any shifting this time around, to keep the comparison as simple as possible.

Obviously one of the key differences between the lenses is that the light fall-off of the HR is considerably more than the TSE. Since great interest will be in the corners, and since I set this up with interesting pixel-peeping areas in the top left that were well lit, I chose to go approximately +1 stop over "sunny 16", knowing that there would be light-fall off in the corners, hopefully leading to good exposure from both lenses in this area.

Both sets of images had good looking histograms prior to applying the LCC's, with limited blow-outs on straight sunlight reflections off glass and metal, but once the LCC's had been applied, there was quite a bit more over-exposing. To counter this, I simply applied Capture 1's auto-exposure to each image individually. I have not made any highlight/shadow adjustments - I want to keep the post processing on this as minimal as possible.

So, the images were shot at the following settings:

All images - ISO35, daylight white-balance as per the IQ180 (4996K, 0 tint).

f/8 - 1/60th
f/11 - 1/30th
f/16 - 1/15th

LCC's shot two stops slower.

Exposure corrections:

23HR:
f/8: -1.08
f/11: -1.19
f/16:  -1.26

24TSE:
f/8: -1.10
f/11: -1.15
f/16:  -1.13

You can judge for yourselves how similar the results are exposure wise, but I think it's pretty close. I'm assuming that by allowing Capture One to auto-correct each image for exposure, that it should also correct for any shutter speed variation between the HR leaf shutter and the FPS focal plane shutter, and also any inconsistencies when changes to shutter speed were made.

Nothing else has been touched in Capture One. Default settings have been left for sharpening - 160/1.0/1.0.

In order to provide the best pixel-peeping opportunity, I have gone back to presenting these as per my original TSE/HCam tests. This involves the following -

Exporting the files from C1 as 16-bit TIFFs.
Importing the TIFFs into Kolor Panotour Pro ("PTP"), then allowing PTP to work its magic on producing a JPG image pyramid, with JPEG quality set to 11.

I've visually checked the result on the web against looking at the original file in C1, and I believe there is a perceptible difference in sharpness, but don't believe it's sufficient to mask the comparative results.

Any objective comments on the above process would be gratefully received.

So, without further ado, the link for you to check it out for yourself:

http://08.ae/panos/TSEvsHRtake2/HRTSE.html

On the left hand side is a little collapsible drawer that allows you to select which image you want to examine. Tools for zooming and panning around should be obvious. I guess if you want to do a good A:B comparison, just open the thing up twice in different tabs, and select the thumbnails you want to compare...

Kind regards,

Gerald.

[ErikKaffehr]

Hi,

I have no issues with the methodology.

No great surprise, the test show how good the IQ180/Rodenstock 23HR can be. I actually think the 24TSE keeps up with the HR at the centers but sharpness fall apart in the corners. I guess that is what we would expect using 135 lens on a 645 FF sensor.

It is also interesting to see how much sharpness is lost by stopping down to f/16 (just checked the HR image).

Best regards
Erik

[gerald.d]

Hi,

I have no issues with the methodology.

No great surprise, the test show how good the IQ180/Rodenstock 23HR can be. I actually think the 24TSE keeps up with the HR at the centers but sharpness fall apart in the corners. I guess that is what we would expect using 135 lens on a 645 FF sensor.

It is also interesting to see how much sharpness is lost by stopping down to f/16 (just checked the HR image).

Best regards
Erik

Interesting.

Which lens and f-stop would you say this crop was from, Erik, and how do you come to your conclusion?



(original link temporarily password protected)

[ErikKaffehr]

Hi,

I would guess it is the 23HR at f/16. At f/8 it was very sharp. As far as I remember the TSE was much softer at f/8 I never checked the TSE at f/16 or f/11, so it is quite possible that TSE would improve at 8/11 or f/16.

I'm pretty sure the crop is not HR 23 at 8 (would be much sharper), I am also pretty sure it is not the TSE at f/8 would be much softer. So OK, either 23HR at f/16 or TSE at f/11. That's my guess. Thinking about lateral chroma but I don't see much of that , I would expect a bit more on that on the TSE.

Best regards
Erik

Interesting.

Which lens and f-stop would you say this crop was from, Erik, and how do you come to your conclusion?



(original link temporarily password protected)

[gerald.d]

Hi,

I would guess it is the 23HR at f/16. At f/8 it was very sharp. As far as I remember the TSE was much softer at f/8 I never checked the TSE at f/16 or f/11, so it is quite possible that TSE would improve at 8/11 or f/16.

I'm pretty sure the crop is not HR 23 at 8 (would be much sharper), I am also pretty sure it is not the TSE at f/8 would be much softer. So OK, either 23HR at f/16 or TSE at f/11. That's my guess. Thinking about lateral chroma but I don't see much of that , I would expect a bit more on that on the TSE.

Best regards
Erik

With respect, you kinda missed the entire point of the test if you didn't bother to check the TSE at f/11 and f/16

That crop was from the TSE at f/16.

Kind regards,

Gerald.

/link re-opened.

/edit
I think your observation and comments raises a couple of interesting points. Firstly, without being presented with a clear a/b test, it's not easy to make a call as to which lens was being used and at which f-stop.

Secondly, you would appear to expect little corner sharpness performance difference between a 23HR at f/16 and a 24TSE at f/11. To me, that seems quite remarkable given the price and design differentials between the lenses considering we're just looking at a single f-stop difference.

I must say that given my previous experience with these two lenses, I was stunned with the 24TSE performance here. I'm not claiming the test was perfect, perhaps I didn't get the best out of the 23HR, but it's a damn fine performance from the Canon!

[samueljohnchia]

Thank you for doing these tests. I have the TS-E24mm II but still I am stunned by how well it performs in this test. It is simply phenomenal. It also produces more uniform color across the sensor plane.

[Stefan.Steib]

There is one essential thing that needs to be discussed in the public:

My theory (supported by some years of doing this stuff with the HCam now) is now that diffraction at larger F-Stops is not as big a fault
as an LCC and vignetting from a "real" or so called wideangle design.
First some informations about this : The Schneider 24mm in my opinion (and also ALPA´s) is completely useless on 80 Mpix backs.
The back lens is 12mm above the chip and this produces immense color casts and vignetting. removing this with an LCC would in fact reduce the resolution and quality of an 80 Mpix sensor on the borders of the format to something like 10 Mpix with heavy noise
Rodenstock was smarter, their 23mm HR is already a retrofocus design with a factor of about 2x (focal lenght to mains distance).
This allows them to get a usable result, but still needs LCC and vignetting control, reducing the quality of the optical result on a digital sensor.
On Contrast the Canon 17 and 24mm TSE use about a factor of 4(24mm)-6(17mm)x retrofocus design.
This as theory proofs is reducing the overall sharpness (Microcontrast ) but still has the info AND gives much more uniform image information.
Now the BIG mistake of Rodenstock  and even more Schneider is they think as optics companies, they want maximum CENTER sharpness from an optical
product, giving away usability and uniformity for top Microcontrast.
This results in a  loss in net information after applying digital image processing on a digital file which can be sharpenend significantly as long as it is UNIFORM !
Same thing is valid to distortion and vignetting.
On my Opinion I would even state that if someone would analyze the Canon Files on the outer areas and compare them to an LCC processed 23HR image
the color information and Net image information will be superior to a "Pro" Wideangle Viewcamera design.

Rodenstock seems to have known that early because they changed to Retrofocus, but they did not do it consequenly enough, the actual design will max out on performance
on 60-80Mpix. Schneiders "real" Wideangles up to 32mm are not even usable for more than 40-60 Mpix anymore.

This is the digital age. The complete system has to be taken into account. A new theory in Optics design is needed.

A lens can be distorting, and have less center sharpness if in return the uniformity and vignetting are controlled down to a minimum.

I know this will  irritate many people extremely, but this is my experience from testing this now all over.

Greetings from Germany
Stefan

[FredBGG]

This is from the Canon 24mm TS II at f4 full shift, bottom right corner.



no corrections applied


Here is the corner from the previous Canon 24mm TS I lens.



From this it is very clear that Canon is making very significant lens quality progress at a pace that the declining MF sector cannot match.
These improvements make the quality difference between MF and the latest Canon and Nikon gear smaller and smaller.

Here is an example of Nikon's optical improvements

The new 85mm 1.4G at f 4



The previous model 85mm 1.4



As you can see very significant progress.

There are many reasons for this progress. Volume and competition is one of them.

However another reason is that these companies are in the motion image Canon and Fuji in particular.
Canon has allocated huge resources towards optical design due to it's move into high quality motion picture.



We have been working with this lens on our latest production



$ 25,000 lens.

The resources and know how that goes into these $25,000 TO $48,000 spills over to the dslr
divisions and the results are what we are seeing in the latest Canon lenses.

Nikon is right on Canons heels as far as entering the motion picture field and has just opened offices in the heart of Hollywood.

It is going to be very difficult for small MF companies to keep pace. Also the relevance of MF in the shifting media world.
Today for long term progress and financial solidity it's important to have a good footing in multiple media types.

I think it's safe to say that most of the investments being made today are in smaller optical designs and motion picture optical designs
 rather than larger format optical designs.

There are also probably more imaging engineers being hired by cell phone companies than MF companies.
Google recently bought Nik software. Kyochera that used to make the Contax now makes cell phones.
Nikon could easily move into the cell phone market as it already has cameras that run on the android OS.

[FredBGG]

I think your observation and comments raises a couple of interesting points. Firstly, without being presented with a clear a/b test, it's not easy to make a call as to which lens was being used and at which f-stop.

Secondly, you would appear to expect little corner sharpness performance difference between a 23HR at f/16 and a 24TSE at f/11. To me, that seems quite remarkable given the price and design differentials between the lenses considering we're just looking at a single f-stop difference.

I must say that given my previous experience with these two lenses, I was stunned with the 24TSE performance here. I'm not claiming the test was perfect, perhaps I didn't get the best out of the 23HR, but it's a damn fine performance from the Canon!

You make a good point. Quality levels have become so close that even "pixel peeping" on a crop the difference is quite hard to see.

Another thing to keep in mind is what sort of quality can be obtained with a rectalinear stitch using just the shift  of a Canon 24mm TS II with a Canon 5D III.
A virtual 36x54 40MP+ sensor for landscape or still life. Canon 5DIII and 24mm TS II for a small total of $ 3,000 plus $ 2,200. That's $ 5,200.

Compare that to over $ 30,000 for the 23HR, IQ140 and a tech camera.....  One can shoot a lot of landscape with a $ 25,000 travel budget.

[torger]

I think it is a bit sad that medium format sensor making seems to move away from being compatible with weak/no retrofocus lens designs. To keep a unique edge I would like to see further development in the 6um pixel size (found in IQ160 etc) and make it better to handle color casts rather than smaller pixels and require stronger retrofocus (i e as the IQ180).

The fact that tech cam users can accept an LCC workflow, centerfilters and small largest aperture gives us access to lens designs which are unthinkable in the DSLR world. It would be nice if that advantage/uniqueness could be kept and further developed.