Who Does DXO Rate the Leica APO-Summicron-SL 35mm f/2 ASPH so Poorly?

[chrismuc]

Advertisement (gone after registration)

Let’s assume you have a sensor 10.000 pixels wide and the lens has 10% barrel distortion. That means that the corners of the image are at every corner 5% of the image width too far inside. In order to achieve an image without distortion, the algorithm has to scale/ enlarge/ stretch (call it what you like) these 9.000 recorded pixels to 10.000 (very simplified) in a non linear way. As mentioned before that obviously leads to a reduction of resolution/ sharpness of the amount of distortion plus the interpolation softness. 

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[Kreeshp]

On 3/25/2024 at 9:43 PM, chrismuc said:

I don’t believe DXO, I think they were not really able to deactivate the automatic distortion correction. 
0.1 % sounds completely unreasonable. 
Lloyd Chambers showed an image without distortion correction, it looks nearly like a fish eye lens. 

What you've written about the distortion in the Q series 28 mm lens is well known in the Leica community, as are the bits about in-camera correction of distortion, but I challenge you to corroborate your claims that the 35mm Apo-Summicron-SL has "very strong (about 6-8%) distortion."  I've cited a source for the 0.1% and your response is you don't believe it.  Where are your sources for "6-8%" distortion?

The 35 APO-SL is touted as a benchmark lens due to its optical performance.  Here's another source claiming its "near-zero" distortion: https://diglloyd.com/prem/s/LMOUNT/LeicaSL/LeicaSL-35f2_APO-distortion.html

The 35 APO-SL is also Peter Karbe's favorite design, because it is so technically superior (source: see 16:25 of this Youtube interview)    Do you really think he would be this proud of a lens with 6-8% distortion?

 

[chrismuc]

Hi, my claim about the SL 35 Apo Asph was from Lloyd Chamber’s text and sample image he shows, if you scroll down his main website. 
Now as I read again, the text information is a bit confusing. He mentions first the 21 Apo Asph, then the 28 Apo Asph, then the ‘35f2’ without specifying which of the two he is referring to. Below the image is just written 35f2 Asph, not clear if the Apo version or the non-Apo Panasonic derivate. So, yes, sorry, I can be wrong, maybe he talks about the non-Apo 35f2. 

[chrismuc]

I downloaded some SL lens raw files and opened them in Iridient but contrary to any other raw file I opened there, I did not succeed to deactivate the baked in distortion correction, so I was not able to verify anything regarding that matter. 

[Kreeshp]

1 hour ago, chrismuc said:

Hi, my claim about the SL 35 Apo Asph was from Lloyd Chamber’s text and sample image he shows, if you scroll down his main website. 
Now as I read again, the text information is a bit confusing. He mentions first the 21 Apo Asph, then the 28 Apo Asph, then the ‘35f2’ without specifying which of the two he is referring to. Below the image is just written 35f2 Asph, not clear if the Apo version or the non-Apo Panasonic derivate. So, yes, sorry, I can be wrong, maybe he talks about the non-Apo 35f2. 

 

Yes, it appears that it is the non-APO 35mm that exhibits poor distortion performance:  https://diglloyd.com/blog/2024/20240320_1400-LeicaSL35f2-distortion.html

If ever one needed more reasons to buy the Panasonic version instead of the Leica "rebadge" this should be compelling.

[BernardC]

20 hours ago, chrismuc said:

In order to achieve an image without distortion, the algorithm has to scale/ enlarge/ stretch (call it what you like) these 9.000 recorded pixels to 10.000 (very simplified) in a non linear way. As mentioned before that obviously leads to a reduction of resolution/ sharpness of the amount of distortion plus the interpolation softness.

It leads to a reduction in sharpness compared to the same lens, uncorrected. What about compared to a different lens that is corrected optically?

 

The optical principle is similar to fisheye lenses. They typically have very even illumination and are sharp all the way to the edge of the frame. It's hard to measure resolution with a fisheye, but good ones are pixel-sharp all across. Lens designer can use this effect to build a lens that is sharper than it might otherwise be, knowing that they'll lose a small portion of that sharpness when the image is re-mapped to be rectilinear. Consumers aren't complaining, they are paying a fraction of what they used to pay for lenses that had worse performance. Seriously, a not-so-great 35/2.8 or 2.5 cost about $300 40 years ago. That's for a Tamron or a Vivitar, an OEM lens cost nearly twice that.

[chrismuc]

Advertisement (gone after registration)

Just as an example the visible effect of distortion correction in a Q3 image:

https://www.dpreview.com/sample-galleries/9356330178/leica-q3-sample-gallery/3635322490

bottom left corner at 200% (I brightened both images a bit that the text on the sticker is better readable)

top is dng, opened in Iridient Developer, distortion correction OFF, saved as tiff

bottom is dng opened in PS, incl. distortion correction

from the widths of the sticker, one can see that the bottom image is stretched in that corner area by about 25%, the reduction of sharpness is obvious, the lens with distortion correction cannot utilise the full resolution of the 60 MP sensor 

the MTF is reduced accordingly, and I am curious to know if the MTF curves published by Leica (for Q and SL lenses) are with or without distortion correction ON, they don't mention it (I asked Leica by email, but no reply yet)

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

Edited March 30, 2024 by chrismuc