24-90mm Focus Shift (Diglloyd)

[jaapv]

Advertisement (gone after registration)

Let's wait and see how the new Sony 24-70/2.8 (better specs on paper, at less than half the price) compares to this lens.

But I am afraid that the answer to your question is that even the best people sometimes make mistakes, especially when dealing with new technologies (for them).

I don't think the techology is new for the designers of this particular system.

[CheshireCat]

There's no end of products with better on-paper specs than the Leica equivalent that sell at half the price or lower.  That's a different subject really; an important one, but different.

 

But am I understanding you correctly? Are you suggesting that the decision to use digital correction on the 24-90 was just a mistake?

 

 

I pay the extra price for Leica because I get something more. In my case, this something more is lenses with the best quality/size ratio.

For someone else, it may be something else. But I think we all agree that only fools would pay a higher price to have the same or less.

 

When I see the raw output of the SL with the 24-90 (18 elements, 4 asphericals, 11 anomalous partial dispersion), the "Emperor's new clothes" tale comes to mind. All this engineering effort for something that requires heavy software corrections similar to the ones required by $100 lenses.

Is this a mistake ? Well, at the end of the day, only pictures matter.

But is it not a mistake if consumers can have same or better pictures for less than half the price, smaller size, and much lower weight ?

[talt03]

End of the day, it's the person's money whether he spends foolishly or wisely buying sony lenses. Life is short so we better shoot outside, analyze and praise our pics and techniques instead of writing a book of debate about the gear. ✌

[Peter H]

I pay the extra price for Leica because I get something more. In my case, this something more is lenses with the best quality/size ratio.

For someone else, it may be something else. But I think we all agree that only fools would pay a higher price to have the same or less.

 

When I see the raw output of the SL with the 24-90 (18 elements, 4 asphericals, 11 anomalous partial dispersion), the "Emperor's new clothes" tale comes to mind. All this engineering effort for something that requires heavy software corrections similar to the ones required by $100 lenses.

Is this a mistake ? Well, at the end of the day, only pictures matter.

But is it not a mistake if consumers can have same or better pictures for less than half the price, smaller size, and much lower weight ?

 

 

I understand everything you say, but it still feels as if you're objecting to digital corrections as a matter of principle rather than because you are getting (or seeing) inferior photographs .

 

In practice virtually everyone seems to be saying what wonderful photos they're getting from it. So I'm still not clear about what the problem is: if it's not about the quality of the photos, what else is it about? Is it a matter of principle? Do you feel the lens is worth a lot less than the asking price, regardless of the quality of photos it can produce?

[thighslapper]

 

 

When I see the raw output of the SL with the 24-90 (18 elements, 4 asphericals, 11 anomalous partial dispersion), the "Emperor's new clothes" tale comes to mind. All this engineering effort for something that requires heavy software corrections similar to the ones required by $100 lenses.

Is this a mistake ? Well, at the end of the day, only pictures matter.

But is it not a mistake if consumers can have same or better pictures for less than half the price, smaller size, and much lower weight ?

 

Plenty of other people have more sense than me when it comes to money ...... 

 

I doubt if most of us would be on this forum if price was the main consideration when it comes to Leica products.

 

Value for money is a very subjective thing ....... and the wearing of rose tinted spectacles and blinkers is fairly endemic amongst photographers who don't make a living from it ........

[scott kirkpatrick]

Are we sure that the sensor area is exactly 6000x4000? These are suspisciously round numbers.Could it  be that the actual active sensor area is larger to accomodate for the crop, like on the Q ? (And does the crop apply always, even if another lens is mounted - that would show up on Rawdigger as well?)

RawDigger digs out a few more rows and columns, but Leica seems to like to use 6000x4000 as the recommended size.  The image with pincushion-shaped edges is C1's best effort to construct what was actually out in the field of view, transformed by the lens into the 24M pixels that are available on the chip (plus, as Tim finds, a few more that are usually just for testing and level-setting).  To construct the "virtual" image of what was really out there, the R,G, and B levels found in pixels near the edge of the chip are mapped into the extended array of pixels that are needed to describe the undistorted image.  Now if you want to have 24M RGB points to send to your screen or printer, you take the area that C1 shows not blacked out.  This is a true 24mm focal length image, and it came from fewer than 24M pixels on the chip, since the rest are left outside.  Tim estimated that only 20 MPx are being used.  My calculation was more like 22, but who cares?  My other point was that if you want to get more use out of your original pixels, you can expand (keeping the 3 to 2 ratio) and extract a new picture with, say 25 M  RGB data points, clean edges and corners and about a 23 mm focal length's view by extending the crop to the top and bottom edges.  Maybe 26 M RGB points can be captured by making the picture a little wider, but the corners begin to soften from too much extrapolation.  At that point you are probably getting a 22.5 mm focal length.

 

All this stuff depends on the specific lens, and its focal length if it is a zoom.  Leica puts these parameters into most of the R lens files that the SL creates, not just the SL, Q and T's new lenses. This means the SL can actually give a better corrected image with classic R lenses than the R's could.

 

scott

[pop]

Advertisement (gone after registration)

.... All this engineering effort for something that requires heavy software corrections similar to the ones required by $100 lenses....

On the fact finding mission again, I see.

 

Show that what takes place is a "heavy" software correction. Fact is, that they're just a few parameters in the DNG for a bit of pixel fiddling in your house and garden PP program. Show that a $100 lens produces comparable pictures with or without a comparable software correction.

[thighslapper]

from the exif the image size is 6120x4016

active image size is 6016x4016

default cropping is 6000x4000

 

I have read Sandy's original article again (and scott's contributions) and he seems fairly sanguine about all this ..... basically they are all at it and it is now a fact of digital life..... and even more so with mirrorless cameras. 

 

Depending on your point of view, software manipulation is enhancing and augmenting the inherent lenses qualities ...... or it is making up for bad optical design.......

 

We are in a glass half full, half empty situation ..... 

[cpclee]

Not an imaginary rectangle.  I'm referring to the rectangle you use to select a crosshair when in field AF mode and it turns green when lens is in focus.   That rectangle marks the contrast detection area.  CDF reads off the whole rectangle, not just the crosshair, when performing autofocus.  

 

I think with a sloping target with uniform contrast such as what you have the focus will always end up on the bottom edge of the rectangle.  This is because depth of field behind the point of focus is always deeper than in front of.  By setting the point of actual focus on the bottom edge, the CDF rectangle covers the greatest area within the DOF and hence contrast is maximized.  Conversely, if the CDF set the point of actual focus on the top edge, the total contrast is minimized because the DOF in front of the point of focus is shallow.

 

If you reverse the direction of the sloping, I bet it will set the point of actual focus to the top edge. In fact I believe this is why, in thighslapper's test, when you turn the camera upside down the result is the same.  The AF sets nearest point covered by the rectangle to be the point of actual focus because of the target has very uniform contrast and the direction of the sloping.

 

With real life targets the contrast is not so uniform and this will rarely be an issue.

 

 

 

 

Not sure what you mean? Are you trying in a complicated way to say that the camera focuses on the nearest point inside an imaginary rectangle around the cross hair?

 

[CheshireCat]

On the fact finding mission again, I see.

 

Show that what takes place is a "heavy" software correction. Fact is, that they're just a few parameters in the DNG for a bit of pixel fiddling in your house and garden PP program. Show that a $100 lens produces comparable pictures with or without a comparable software correction.

 

Well, I just finished playing with RawDigger, and some 42MP Sony 24-70/2.8 MG raw files here:

http://www.dpreview.com/samples/2325904456/crowd-pleaser-sony-24-70mm-f2-8-gm-real-world-images

And these are my observations:

- Outresolves the sensor at least in the center.

- No vignetting.

- Some coma wide open at 24mm (visible on 1:1 crops).

- Very sharp to the extreme corners (42 MP).

- Slight distortion: barrel in the center and pincushion at the corners, so correcting it in software will not harm resolution (will actually improve it in the corners).

- Very noticeable amount of lateral CA (easily corrected in software).

- Low longitudinal CA wide open.

 

I have also played with some 24MP Leica SL 24-90 files here:

http://www.dpreview.com/articles/5300274665/leica-sl-studio-and-real-world-sample-gallery

And these are my observations:

- The heavy distortion and vignetting only happens at the wide end. It is of the barrel type, heavier than the Sony lens but less complex in nature.

- At other FL, the distortion is not as heavy.

- Corner performance is good at the wide end, but Sony looks same or better.

- The SL does not crop if it is not needed (e.g. other FL, or M lenses).

- Very noticeable amount of lateral CA (easily corrected in software).

 

NOTE: Keep in mind that these were random shots is not a direct "same-shot" comparison

 

All in all, the 24-90 is much better than I originally thought, but seems like a 28-90 stretched to cover also 24.

However, - very personal opinion - for me it's Sony all the way. The extra 20mm at the long end are compensated by the higher MP count.

[CheshireCat]

from the exif the image size is 6120x4016

active image size is 6016x4016

default cropping is 6000x4000

 

The RAW output file is 6016x4016.

The extra 8 active (image) pixels on each border are used to simplify image processing arlgorithms so they may easily use neighboring sensor elements to determine the value of a pixel (debayer, et cetera).

These simplified algorithms produce artifacts at the borders, and this is why the image is then cropped to 6000x4000.

 

The extra 104 sensor elements on each row are probably not active and used for "internal stuff".

[thighslapper]

Not an imaginary rectangle.  I'm referring to the rectangle you use to select a crosshair when in field AF mode and it turns green when lens is in focus.   That rectangle marks the contrast detection area.  CDF reads off the whole rectangle, not just the crosshair, when performing autofocus.  

 

I think with a sloping target with uniform contrast such as what you have the focus will always end up on the bottom edge of the rectangle.  This is because depth of field behind the point of focus is always deeper than in front of.  By setting the point of actual focus on the bottom edge, the CDF rectangle covers the greatest area within the DOF and hence contrast is maximized.  Conversely, if the CDF set the point of actual focus on the top edge, the total contrast is minimized because the DOF in front of the point of focus is shallow.

 

If you reverse the direction of the sloping, I bet it will set the point of actual focus to the top edge. In fact I believe this is why, in thighslapper's test, when you turn the camera upside down the result is the same.  The AF sets nearest point covered by the rectangle to be the point of actual focus because of the target has very uniform contrast and the direction of the sloping.

 

With real life targets the contrast is not so uniform and this will rarely be an issue.

 

I assumed it was something like this ....... a design 'feature' rather than an error ....... and borne out by the fact that with 'normal' subjects the 'problem' doesn't manifest itself. 

 

I suppose Leica could help by substituting a box in spot AF and correcting the slight mismatch between where it is and the actual focus point .....

[adli]

In my set up there is no rectangle, just the white cross hair which moves in steps between the 49 defined focus points.

 

Not an imaginary rectangle.  I'm referring to the rectangle you use to select a crosshair when in field AF mode and it turns green when lens is in focus.   That rectangle marks the contrast detection area.  CDF reads off the whole rectangle, not just the crosshair, when performing autofocus.  

 

I think with a sloping target with uniform contrast such as what you have the focus will always end up on the bottom edge of the rectangle.  This is because depth of field behind the point of focus is always deeper than in front of.  By setting the point of actual focus on the bottom edge, the CDF rectangle covers the greatest area within the DOF and hence contrast is maximized.  Conversely, if the CDF set the point of actual focus on the top edge, the total contrast is minimized because the DOF in front of the point of focus is shallow.

 

If you reverse the direction of the sloping, I bet it will set the point of actual focus to the top edge. In fact I believe this is why, in thighslapper's test, when you turn the camera upside down the result is the same.  The AF sets nearest point covered by the rectangle to be the point of actual focus because of the target has very uniform contrast and the direction of the sloping.

 

With real life targets the contrast is not so uniform and this will rarely be an issue.

[cpclee]

Not so much a mismatch, just the fact that under extreme conditions the contrast maximizing point of focus can fall on one edge of the CDF rectangle. The algorithm is doing what it's supposed to do but can be confusing to the unsuspecting user.  There are a few ways Leica can mitigate

 

- make the rectangle always visible once selected.  The crosshair can go away.  This won't change how the camera actually focuses but will make it less confusing if such a situation were to occur.

 

- make the CDF rectangle much smaller.  So even when the point of focus on an edge of the rectangle, it will still be pretty close to the center of the crosshair

 

- upweight the contrast of the center pixels in the rectangle and downright the peripheral pixels.  This way the algorithm will maximize the "weighted total contrast" of the rectangle rather than the "total contrast".  Extreme conditions can still pull the point of focus off center but not so much that it falls on an edge.  

 

I suppose Leica could help by substituting a box in spot AF and correcting the slight mismatch between where it is and the actual focus point .....

[cpclee]

When you use the joystick to select one of the focus points / crosshairs do you not see a rectangle moving on the screen?  That's the rectangle I'm referring to.

 

In my set up there is no rectangle, just the white cross hair which moves in steps between the 49 defined focus points.

[tashley]

Maybe the crosshairs are a sort of marketing gimmick to give the impression of extreme precision, but in fact work just like a box does? Like a volume knob that goes to 11?

[LocalHero1953]

When you use the joystick to select one of the focus points / crosshairs do you not see a rectangle moving on the screen?  That's the rectangle I'm referring to.

You have got the AF Field Size set to 1-point rather than Field, haven't you? The latter shows a rectangle, the former a cross.

Just to be sure we're talking about the same thing.

[adli]

As I said, no visible rectangle, only this white cross hair jumping between the focus points. I have my set up to AFc, Static and 1 point.

 

If you have AF Mode to Tracking or AF Field Size to Field, you will see a rectangle instead of the cross hair.

 

 

 

 

 

 

When you use the joystick to select one of the focus points / crosshairs do you not see a rectangle moving on the screen?  That's the rectangle I'm referring 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!

[profus]

Any of the issues ( be it focus shift or just missed /front / back focus ) described in this topic on previous 11 pages solved with the latest

fw  2.0  update?

 

The claim from Leica is:

 

Improved Autofocus

 Higher precision

 Improved tracking

 New AF-steps in setup-menu (1 and 5 steps)

 Improved 1Point AF: 529 AF-Points

[Advertisement]

Hallo,

Look here: Leica L-Mount

[jonoslack]

Any of the issues ( be it focus shift or just missed /front / back focus ) described in this topic on previous 11 pages solved with the latest

fw  2.0  update?

 

The claim from Leica is:

 

Improved Autofocus

 Higher precision

 Improved tracking

 New AF-steps in setup-menu (1 and 5 steps)

 Improved 1Point AF: 529 AF-Points

 

As I understand it the "focus shift" is sorted, and the AF is improved