Q2: 24mm & 51M Pixels?

[Chippy_boy]

Advertisement (gone after registration)

I was very surprised to read Thorsten Overgaard saying the Q2 is essentially a 51MP camera with a 24mm lens, but the image is cropped down to 28mm/47MP.

He says, "In Adobe Lightroom you see the cropped frame, but if you click on the photo with the cropping tool, you will see the entire 24mm frame with the 28mm crop."

Does anyone know how to do this in Lightroom?  When I click on the crop tool, nothing changes.  Am I missing something?

[acg69]

I have the same question. I tried it both in Capture 1 and Lr and i didn’t see anything more than what the 28mm lens took. Maybe this is the case (24mm and 51 mp) but while it was evident on pre-production models, they may have decided to take it off completely due to the severe vignetting. That said, I would like to have the option to tap into that extra territory in case I have made a mistake while composing the picture.

[Advertisement]

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

[M11 for me]

First the lens is 28mm.

When you shoot in JPGs then in Lightroom you see the cropped JPG (35, 50 or 75mm) or no crop when you use the native focal length of 28mm. If you shoot in DNG then you still have the whole picture available of course. But in the Lightroom previews you see the crop (as you took your photograph) but that crop you can change/undo at any time in Lightroom as you still have the full picture available as DNG. You can as well go back to the native 28mm. 

There is there an interesting detail: In the Exifs you see hw the crop was set in the camera. So when you crop to 50mm then in the Exifs you will always see that information. Its is a bit queer though if you undo your croping now and go back to 28mm. Then the Exif entry of 50mm will stay in there.

Edited March 6, 2020 by M10 for me

[M11 for me]

In addition to the first answer:

The initial croping from something lower than the 28mm focal length has to do with the SDC (Software Distortion Correction). You can emulate the same with any of your pictures when you correct plunging lines. Just try it in your Lightroom (even without plunging lines;  just use the respective tool). Then as a result you will have an image that is a bit smaller than the original one. You have a croping. The same applies when the camera corrects distortion by software. As a result not the whole sensor is finally used.

And as a matter of fact the lens of the Q2 has quite strong distortion. On the other hand the lens is small, light and rel. cheap. As you can not change that lens anyway the SDC is acceptable. There is actually software with which you can eliminate the effect. In case of landscape lens distortion very often is of no relevance. As y result you get more detail. The SDC is a bit destructive. It is a compromise. 

 

Edited March 6, 2020 by M10 for me

[Chippy_boy]

31 minutes ago, M10 for me said:

First the lens is 28mm.

Or is it?

How do we know this?  How can we know this?  All we know is it says 28mm on it.  But the Q2 is an integrated system and all we really know is that it produces images which have the equivalent FOV of a 28mm lens on a FF camera.

If Thorsten Overgaard says it's actually a 24mm lens, I am inclined to believe him.  Seems to me extremely likely even, that it is a 24mm lens with the image cropped down to 28mm equivalent.

[M11 for me]

vor 3 Minuten schrieb Chippy_boy:

Or is it?

How do we know this?  How can we know this?  All we know is it says 28mm on it.  But the Q2 is an integrated system and all we really know is that it produces images which have the equivalent FOV of a 28mm lens on a FF camera.

If Thorsten Overgaard says it's actually a 24mm lens, I am inclined to believe him.  Seems to me extremely likely even, that it is a 24mm lens with the image cropped down to 28mm equivalent.

Has to do with SDC. Thorsten did probably not measure

Edited March 6, 2020 by M10 for me

[ajmarton1]

Advertisement (gone after registration)

I don't know if I would take everything Thorsten says as the final word.  I would check with a Leica store.  

[Chippy_boy]

4 hours ago, ajmarton1 said:

I don't know if I would take everything Thorsten says as the final word.  I would check with a Leica store.  

I doubt they would know to be honest. What is clear is that the sensor is capturing 9005 x 5715 pixels, not 8368 x 5584 as reported in Lightroom. And the image if viewed in some raw converters is wider angle than shown in LR.

There seems little doubt to me that the lens is indeed wider than 28mm. Whether any of the image wider than 28 is useable?  I don't know.

[M11 for me]

According to ReidReviews through SDC there is a crop of 1.131. There is no comment re the angle of view.

[Archiver]

9 hours ago, ajmarton1 said:

I don't know if I would take everything Thorsten says as the final word.  I would check with a Leica store.  

A European fellow at a certain Leica store that will go unidentified told my companion that the C-Lux is a 'real Leica', even though I had my M9 in plain view and had already discussed rangefinder shooting with him. I wouldn't necessarily trust what I'm told in a Leica store, although others at that store were more honest/knowledgeable.

[hdmesa]

Do we really know which of these is true?

1. Leica is capturing an image using a 25mm lens onto a sensor that is the resolution they say it is. The image is enlarged while correcting for distortion (corners stretched out) and cropped to the original resolution/pixel dimensions resulting in an image with a 28mm FOV.
2. Leica is capturing an image using a 25mm lens onto a sensor that is higher resolution than they say it is. The image is sized down while being corrected for distortion (sides pushed in) resulting in an image with a 28mm FOV.

Has anyone ever disassembled the Q/Q2 and physically measured the sensor? What if its slightly larger to accommodate the "crop" and the lens really is a 28mm lens? 

[hdmesa]

This is an example using the original Q, but the results should be the same for the Q2 only with different pixel dimensions.

In Capture One, you can retain the distortion correction while regaining some of the missing image area that Leica has marked in the DNG to be cropped.

Here is how the image looks normally in Capture One when you switch to the crop tool. The pixel dimensions of the crop are 6000x4000 and match the Q's advertised 24mp resolution. Notice the darkened image area available to "uncrop" to:

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!

 

In Capture One, I can uncheck "Hide Distorted Areas" in the Lens Correction panel to see the distortion correction at work. Notice how the center and long sides of the image are least affected by the distortion correction while the corners and short sides are most affected:

 

Here is the image after I use the crop tool to expand the image frame to use the full available image area while still retaining the distortion correction. I've not tested the results at infinity – the edges toward the corners may be soft, but at close distance, it's nice to regain the wider field of view. This still keeps the hard-vignetting out of the frame. Note the resulting image is 6523x4118px (26.9mp). This is the reason I tend to think the Q and Q2 have the advertised sensor resolution, and the image is being enlarged before cropping as opposed to using a higher resolution sensor and the image being sized down before cropping:

 

Now watch what happens when I set the Capture One Distortion correction slider to zero. The full image reverts back to 6000x4000px. I think the original image as recorded by the sensor is 6000x4000px and Leica is instructing C1/LR to enlarge the image to apply the corrections while keeping a 6000x4000px crop frame:

 

Finally, I've not tested this to be sure, it it's possible the vignetting in the corners of the full uncorrected image might go away if I removed the lens hood and clear filter. The full image would be a nice option to have should the distortion itself not be too distracting in a given image. Might also need to make sure to keep OIS off to be sure there are no shifts in the image circle that could darken one or more corners.

[nicci78]

It is not 24mm. It is 26mm. Thorsten got its math wrong.
You should look for previous posts about this.
It was talk about several times now. 
To sum up Q2 works like original Q. It needs a bigger than full frame sensor to overcome the Summilux-Q corner flaws : massive vignetting (dark corner) and massive distortion. 
Q2 will crop its 26mm image = 9005x5715 pixels or 24,50x38,75mm size sensor into an effective 28mm of 8368x5585 pixels or 24x36mm sensor. 
 

The image is processed with software correction in camera for JPEG and in Lightroom for DNG. Cropping allowed to throw away the worst part of the image after all corrective means. 
 

In video mode, the Q and Q2 use the full width of the sensor. Resulting into a 26mm image instead of 28mm. Which is very fortunate. Thanks Leica. 

Edited March 8, 2020 by nicci78

[hdmesa]

2 hours ago, nicci78 said:

It is not 24mm. It is 26mm. Thorsten got its math wrong.
You should look for previous posts about this.
It was talk about several times now. 
To sum up Q2 works like original Q. It needs a bigger than full frame sensor to overcome the Summilux-Q corner flaws : massive vignetting (dark corner) and massive distortion. 
Q2 will crop its 26mm image = 9005x5715 pixels or 24,50x38,75mm size sensor into an effective 28mm of 8368x5585 pixels or 24x36mm sensor. 
 

The image is processed with software correction in camera for JPEG and in Lightroom for DNG. Cropping allowed to throw away the worst part of the image after all corrective means. 
 

In video mode, the Q and Q2 use the full width of the sensor. Resulting into a 26mm image instead of 28mm. Which is very fortunate. Thanks Leica. 

If you want to see massive distortion, you should see what Canon did with the RF 24-240 at 24mm. Insane.

I would call the vignetting substantial at infinity and minor at close distance. I've also never seen someone do a test shot without the hood and a filter to see if the vignetting is reduced or even eliminated.

It sounds like you're saying the Q uses a larger than full frame sensor and crops to full frame – is that the case? Do you have a source for this? Maybe something is lost in translation with how I'm reading your post, though.

[hdmesa]

^ In any case, it's astounding that Leica was able to use this much distortion correction and still have a resulting file with such great IQ.

I would love to see Leica create a medium format Q-S:

• $9K
• 45mm lens (about 35mm full frame FOV equiv.)
• True manual focus, no AF
• No size and weight constraints, therefore:
• Highest IQ goal with no distortion correction required

[01af]

Am 6.3.2020 um 10:50 schrieb Chippy_boy:

I was very surprised to read Thorsten Overgaard saying the Q2 is essentially a 51 MP camera with a 24 mm lens, but the image is cropped down to 28 mm and 47 MP.

Being surprised proves you a sensible person. Because it's bunkum ... even though it does contain a (very) small grain of truth.

First of all, the Leica Q2's sensor effectively is 47 MP, not 51 MP. As any sensor, it does have a few more pixels in total but those aren't used for the actual image.

Then, contrary to conventional wisdom, the focal length of a lens is not just a simple number. In fact, most real-world lenses have many focal lengths because it varies across the image height. At the frame's center, the focal length usually is different from the focal length at the frame's edge — or corner. Except, of course, when the lens was perfectly distortion-free ... but most lenses aren't, and the Q2's most definitely isn't. The focal length of a photographic lens can be determined in two different ways: (1) via the magnification at the frame's center, or (2) via the angle-of-view captured across the frame's diagonal. When the lens distorts then these two ways will yield different values for the focal length. In lenses with barrel-shaped distortion, the focal length decreases from center to edge (hence the negative numbers to quantify barrel-shaped distortion); in lenses with pincushion-shaped distortion, the focal length increases from center to edge (hence the positive numbers to quantify pincushion-shaped distortion).

In the Leica Q and Q2, the Summilux 28 mm Asph lens really is 28 mm by the magnification at the frame's center. But it has severe barrel-shaped distortion so by the angle-of-view across the diagonal, the focal length is more like 24.7 mm. But then, the image-circle diameter is not 43.3 mm but 41.4 mm only. So the lens cannot illuminate the full sensor area, hence the 24.7 mm are a 35-mm-format equivalent of 25.8 mm. The extreme corners are black. (By the way, it's the same with the Vario-Elmarit-SL 24-90 mm Asph at 24 mm on the Leica SL and SL2.) The black corners disappear after the in-camera distortion correction which renders the corner's focal length equal to the center's focal length.

So ... the blunt statement that the lens was 24 mm just doesn't do justice to the complexity of how things really are. And in real life, it's the result that counts, not the loops jumped along the way.

[nicci78]

@hdmesa 
Nothing is lost in translation. Q2 has a 51.5MP sensor which is slightly bigger than full frame, but reframed into 24x36 46.7MP

28mm is a physical element. Not made up. You have to be able to measure it. 
A 28mm lens upon a slightly bigger sensor 24.50 x 38.75mm will give you an equivalent focal length of 26mm. Because its crop factor is around 0.93x

It is a really clever way to keep the size of the optics as small as possible. And also as good as possible. 
 

 

Edited March 8, 2020 by nicci78

[nicci78]

Q has a 26.3MP sensor. 
If you do the math it will match 26mm equivalent field of view and slightly bigger sensor than 24x36

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!

[jaapv]

This has been discussed ad nauseam on the introduction of the SL.
The lens is a hybrid design, like quite a few lenses nowadays. What is going on is that the lens is corrected both optically and digitally; that allows the designer to correct a large number of aberrations better by optical means, allowing the residual aberrations that can be corrected digitally  (mostly distortion) to the camera firmware, resulting in a better image overall. One cannot be separated from the other. Optical purists are quite upset about this development in lens design, but it does result in better optical quality, more compact lenses and lower prices. Most people don't seem to realize that lens design consists of shifting inevitable optical aberrations to be able to minimize them by counter-aberrations. Trying to separate the two is equivalent to removing one element from a traditional lens and complaining that image quality is impacted.
The focal length argument is quite nonsensical; most lenses cover a wider angle of view than needed for the format they are designed for, some considerably so. In that sense nearly all lenses are "cropped down". Up until now, nobody had any objection.*
As for the MP count of the sensor, all sensors have more pixels than are used for image forming for technical reasons. So the SL2 has more "technical pixels"  than average. So what?

 

*In fact, some R lenses turn out to be quite usable on an S sensor.

[Advertisement]

Hallo,

Look here: Leica Q

[01af]

vor 19 Stunden schrieb nicci78:

Q has a 26.3 MP sensor. If you do the math ...

Your math is flawed. The lens doesn't illuminate the whole 26.3 MP (or 50.4 MP in the Q2). It doesn't even fully illuminate the 24 MP (or 46.7 MP in the Q2) of the 24 × 36 mm image area. The image circle is just 41.4 mm wide. The lens still has a diagonal angle-of-view of approx. 80° due to the barrel-shaped distortion.