XCD 21mm for architecture instead of tech cam

[drevil]

just out of curiosity, wouldnt the rodenstock 23mm be much sharper than the GF23 or XCD21? i mean much much sharper?

 

[diggles]

How are you adapting these lenses to a 907x?

WRE-CA - Cambo

ACB-CA - Cambo

just out of curiosity, wouldnt the rodenstock 23mm be much sharper than the GF23 or XCD21? i mean much much sharper?

While I can't speak for the XCD21, I do have both the 23HR and GF23. The only comparison I've done so far is an image of my living room using a GFX100s and a CFV II 50c. Both lenses performed very similarly in that situation. I'm planning on photographing a building exterior with both set ups as soon as I get a chance. I'll post it in this thread once I do.

 

[rdeloe]

While I can't speak for the XCD21, I do have both the 23HR and GF23. The only comparison I've done so far is an image of my living room using a GFX100s and a CFV II 50c. Both lenses performed very similarly in that situation. I'm planning on photographing a building exterior with both set ups as soon as I get a chance. I'll post it in this thread once I do.

I look forward to seeing your comparison. I won't be surprised if your images show that the sharpness difference is minimal, and that the 23HR is not necessarily any sharper.

 

[Godfrey]

just out of curiosity, wouldnt the rodenstock 23mm be much sharper than the GF23 or XCD21? i mean much much sharper?

Why do you expect the Rodenstock 23 lens to be "much much sharper" than the XCD 21mm? I've done a decent amount of shooting with the XCD21mm on the 907x and it is a satisfyingly extremely sharp lens in everything I've seen out of it, presuming I set the focus properly.

I've not used anything for tilts and shifts as yet, those capabilities are generally not my need so far.

G

 

[onasj]

MTF curves suggest the Rodi 23 HR and the XCD 21 perform similarly across the area of a 44x33 mm sensor.

Below: XCD 21 (top), Rodi 23 HR (bottom). Note that the lp/mm curves are 10, 20, 40 for the XCD (black) curves, and 10, 20, 40, 80 for the Rodi 23 HR (red) curves. So you are comparing the first three set of black lines with the first three set of red lines from an X-axis value of 0 to 27 mm. Which look more-or-less equivalent to me, with a slight edge to the XCD 21 in acuity and astigmatism.

 

[MGrayson]

The modern Fuji and Hassy MF lenses out-resolve their sensors wide open. By giving up optical distortion control and fixing it digitally, extreme sharpness and light weight are possible. The tech camera lenses, designed for movements, have to be as perfect optically as they can be.

 

[Godfrey]

This may be true, however I haven't seen that the lens correction added to the XCD 21mm actually changes its output by all that much if anything.

It reminds me of the Pana/Leica Macro-Elmarit-DG 45mm f/2.8 ASPH designed for Micro-FourThirds, in a way: Many people dissed the lens as "needing the lens corrections to produce a decent image" but when I took the lens parameters shipped into the DNG files apart, it was obvious that there were actually NO corrections being performed as all the parameters were set to 0.

G

 

[MGrayson]

This may be true, however I haven't seen that the lens correction added to the XCD 21mm actually changes its output by all that much if anything.

It reminds me of the Pana/Leica Macro-Elmarit-DG 45mm f/2.8 ASPH designed for Micro-FourThirds, in a way: Many people dissed the lens as "needing the lens corrections to produce a decent image" but when I took the lens parameters shipped into the DNG files apart, it was obvious that there were actually NO corrections being performed as all the parameters were set to 0.

G

Don't get me wrong. I think correcting aberrations that CAN be corrected in post is a fantastic idea. And some of the µ43 lenses are jewels. I was just pointing out that the designer of a shift lens - well, not that it can't be corrected in post, but the shift parameters would have to be known to the camera quite precisely - has a harder job.

M

 

[Godfrey]

I'm not sure that the job is any harder than designing any lens to perform well, aside from giving the design a wider image circle which sustains the quality out to farther edges off the optical center. The camera doesn't have to know anything, the lens just has to perform well through the range that you use it (which depends on format and your personal evaluation of what's enough and what you need for a given scene). If you're not shooting film, image correction post-exposure is a given nowadays, regardless of single or multiple exposure solutions, etc. The technology to implement it is now well understood.

The XCD series lenses existant at present aren't designed for a tilt or shift, that's all. But multiframe capture and stitching with corrections can work extremely effectively. My question was why somdone might automatically consider the Rodenstock to be *that much* better performing, modulo the tilt/shift world. The notion doesn't make sense to me, so I seek to understand the assumption.

G

 

[diggles]

Since it stopped raining yesterday I was able to get out to compare the 23HR on a CFV II 50c to the GF23 on a Fuji GFX100s. I'm sure different cameras and sensor size/tech complicates the comparison, but hopefully it will help answer the question–is the 23HR sharper than the GF23?

For the comparison I wanted to find a building that is rectangular, fairly tall, and a good example of the types of buildings I photograph in northern Colorado.

Here is the scene:


The tripod remained in the same spot for both shots.

The workflow for the 23HR is as follows:

• f8 at 1/250
• 5mm camera fall focused towards the edge of the frame (I also took another one focused in the center of the frame, RAW file is included with link below)
• No CF
• LCC reference for processing in Phocus
• Landscape color correction profile in Phocus
• Default sharpening
• Output to 16 Bit Tiff
• Correct distortion with Alpa Photoshop plugin
• Convert to smart object
• Final color correction and geometry adjustments in ACR
• ACR Sharpening (sharpening 40, radius .5, detail 100)

The workflow for the GF23 is as follows:

• f8 at 1/200
• 5° camera tilt with the Arca-Swiss Core 75 Leveler Classic
• Color correction and geometry adjustments in ACR
• ACR Sharpening (sharpening 70, radius .5, detail 100)

One variable that I am unsure about is the sharpening. I'm assuming that Phocus does apply some amount of default sharpening, but I don't know that for sure. If it does, then there were two rounds of sharpening applied to the 23HR.

Here are side by side 100% crops from both images–left, center, and right along the bottom and top of the building. GF23 on the left, 23HR on the right. The GFX files are bigger which is why those images are zoomed in more. This is another variable that may influence how the sharpness is perceived, but I don't know how to fix that.

I won't be surprised if your images show that the sharpness difference is minimal, and that the 23HR is not necessarily any sharper.

Based on my copies of these lenses and this test, the sharpness difference is minimal and the 23HR is not necessarily any sharper.

The 23HR lens was calibrated by Rodenstock about two months ago so I am assuming that it's as sharp as my copy can get. The GF23 is a new copy that I got a couple weeks ago.

Here is a link to the RAW files if you would like to download them and take a closer look:

Sync | Secure cloud storage that protects your privacy

ln3.sync.com

Included in the RAW files are images focused at center and towards the edge, LCC reference images for the 23HR, and a LCC image for the GF23 because I was curious what it would look like.

 

[rdeloe]

Warren, this is incredibly helpful. Thanks for sharing the RAW files.

I examined the files closely and noticed a few things. I'm curious if you saw this too:

• The "edge" frame is overall stronger than the centre frame in the Rodenstock pair. The GF seems to be about the same in the edge/centre comparison.
• The Rodenstock lens seems to have moustache distortion. That was surprising. The GF does not appear to have that form of distortion (or it's well-corrected in camera).
• If my main requirement was maximum resolving power, I'd choose the GF 23. I compared the "edge" Rodenstock to the "edge" GF. The GF is clearly sharper across the whole frame. I can read the larger lines of text on the sign on the doors in the GF picture, but not in the HR. Of course the GF file has a lot more pixels! So I down sampled the GF file it to the resolution of the Rodenstock file and checked again. The GF frame still looks a bit sharper overall to my eye, but the difference is not as obvious as at full resolution.
• Of course the Rodenstock can shift, where the GF cannot.

In sum, they both seem to be excellent lenses, so it really comes down to what you need to do.

Cheers, Rob

 

[diggles]

Hey Rob,

The "edge" frame is overall stronger than the centre frame in the Rodenstock pair. The GF seems to be about the same in the edge/centre comparison.

Yes, I saw that too. When focusing on the center of the frame with the 23HR the edges are not sharp–they almost look jittery.

The Rodenstock lens seems to have moustache distortion. That was surprising. The GF does not appear to have that form of distortion (or it's well-corrected in camera).

The Alpa plugin did a pretty good job of cleaning that up, but yeah it does seem to have a bit of mustache distortion. It would be interesting to see what the GF23 looks like coming straight out of the camera.

If my main requirement was maximum resolving power, I'd choose the GF 23. I compared the "edge" Rodenstock to the "edge" GF. The GF is clearly sharper across the whole frame. I can read the larger lines of text on the sign on the doors in the GF picture, but not in the HR. Of course the GF file has a lot more pixels! So I down sampled the GF file it to the resolution of the Rodenstock file and checked again. The GF frame still looks a bit sharper overall to my eye, but the difference is not as obvious as at full resolution.

This is where I was struggling with the comparison. The GF is clearly sharper across the whole frame, but it is difficult to tell how much of that is the sensor size. I'm wondering how well the 23HR holds up on a 100mp sensor. I've read on the forum that it struggles with the IQ4 150.

Of course the Rodenstock can shift, where the GF cannot.

The Rodenstock clearly wins on this point! Although, I was expecting to see more issues when tilting the camera up at the building and fixing in post. The end result is really very similar. For me, the shift on the 23HR maxes out at 10mm before becoming quite ugly. I tried another shot with the GF23 where I got closer to the building and tilted the camera 10°. It corrected nicely as well. Although I'm not sure I would go much further than that.



They both are excellent lenses to my eye as well. In terms of overall IQ I believe the GF edges out the HR by a nose, but if you want in camera t/s then the HR can't be beat.

 

[tcdeveau]

• Of course the Rodenstock can shift, where the GF cannot.

IMHO this is the main reason to go for a tech cam with Rodenstock lenses over the XCD/GF lenses, although the 23 Rodie is more limited movement-wise than some of the others.

I never did any direct comparisons between the 40HR and 45mm XCD when I briefly had the CFVII (the most apples-to-apples comparison I could've done) but am not sure I saw a huge difference in resolution between the 40HR and 45mm XCD. The 40HR maybe seemed a tad sharper, but that could've just been my brain telling it was based on the price tag.

Unless one needs/wants movements, or wants the tech cam joy (they are a lot of fun), the XCD and GF lenses are plenty sharp enough.

 

[PSS]

I am not a big fan of the attitude to fix it in post but I have also shot a ton with 4x5 cameras and I loved my fuji 680. which really makes me appreciate the XCD 21 and all possible adjustments in Phocus. I am still itching to put the CFV II 50 on a tech camera but probably more to play with tilt.
If I have the feeling that I might run out of image area with severe adjustments there is still the option to stitch a few frames as well.
If I had to shoot high end architecture for huge print every day, I would probably look at the XT.

 

[diggles]

I am not a big fan of the attitude to fix it in post but I have also shot a ton with 4x5 cameras and I loved my fuji 680. which really makes me appreciate the XCD 21 and all possible adjustments in Phocus. I am still itching to put the CFV II 50 on a tech camera but probably more to play with tilt.
If I have the feeling that I might run out of image area with severe adjustments there is still the option to stitch a few frames as well.
If I had to shoot high end architecture for huge print every day, I would probably look at the XT.

Traditionally I have not been a fan of "fix it in post" either. Getting the shot in camera is optimal. Working with Tech cams is a lot of fun and I really enjoy the WRS and CFV II 50c, but if you can get equal/better results by fixing in post with the GFX 100s then it's worth considering.

For now, I'm not convinced that in all situations the GFX and GF23 will provide equal/better results compared to a tech cam with the 23HR. Tilting your camera up to fix converging lines may fix one problem, but what if you need to avoid an obstruction and fix converging lines. With a tech cam you can shift to the left or right and up or down. Maybe swinging your camera to the left and tilting it up is fixable in post, but I'm not sure because I haven't tried it.

 

[rdeloe]

My strong preference is to solve as many problems as I can before I make the exposure, so camera movements are central to how I work. Furthermore, there are also things you simply can't do in post. However, it's clear to me that camera movements are niche and getting more niche as sensors improve. The cost of entry to the world of camera movements (financial, time, brain sweat) is simply too high for many people relative to what they're able to see as the benefits. That last point is in no way meant to be condescending. Sometimes we don't know what we don't know. Someone starting out who looks at the images Warren posted is going to have a hard time understanding why it's worth going through all the expense and awkwardness of using a technical camera when the GFX 100S plus GF 23mm seems to be able to produce an image that is "just as good" for a fraction of the bother.

Anyway, I shall keep doing my thing.

 

[vieri]

just out of curiosity, wouldnt the rodenstock 23mm be much sharper than the GF23 or XCD21? i mean much much sharper?

Why do you expect the Rodenstock 23 lens to be "much much sharper" than the XCD 21mm? I've done a decent amount of shooting with the XCD21mm on the 907x and it is a satisfyingly extremely sharp lens in everything I've seen out of it, presuming I set the focus properly.
...

I think we need to take sensor size & pixel size into consideration when doing such a comparison, as well as pixel count. Until we'll be able to test them both on the same, larger sensor, rather than a matter of absolute sharpness it's mostly a matter of "is what you see with your system sharp enough for you". Which, incidentally, is what counts in the end

Best regards,

Vieri

 

[rdeloe]

I think we need to take sensor size & pixel size into consideration when doing such a comparison, as well as pixel count. Until we'll be able to test them both on the same, larger sensor, rather than a matter of absolute sharpness it's mostly a matter of "is what you see with your system sharp enough for you". Which, incidentally, is what counts in the end

Best regards,

Vieri

You are correct Vieri, in that an apples-to-apples comparison is always best. However, that's not always possible. In this case, I normalized the resolution and compared. There are issues with that method too, but it's a reasonable way to compare across systems like this I think.

A print of the same size would be another excellent way to compare. Comparing prints would reveal how good all these lenses and cameras are.

 

[Robert Salisbury]

Hi there, this is the exact dilemma I face, but I fear with far less knowledge than the people posting here. I have a Cambo actus db2 with 50cv11 back. What is the best option to obtain the widest angle for architecture? Currently I use the 40mm CF fle hass v lense. I require wider and more room to correct with movements. My head hurts with trying to understand the large format lenses, and do I need a center filter with these options. Any help / advice on what I should be looking at would be awesome appreciated!!

 

[mristuccia]

I may be wrong but, at least on the paper, in addition to the tossing of lateral pixels, with keystoning we loose resolution as well due to the stretching+interpolation of what still remains in the frame (the same applies with digital lens distortion correction).
I didn't measure the degree of this smearing, and from what I see here none of you seems to complain about it, but instinctively I'd say that I prefer a tech cam with shift to digital keystoning.