Red Edge--Which lenses?

[Jerry_R]

Advertisement (gone after registration)

It happens with wide angle lenses which have a very steep angle of incidence of light, and where lens correction is applied to reduce vignetting

If we test, measure and discuss REASON, ROOT CAUSE of color shift - forget about any software corrections performed by camera body. Turn it off. It is additional factor, that for clean of our search should be excluded.

 

I would not relate color shift to vignetting removal. I mean - removal of vignetting is not reason of color shift.

 

Just observe with which lenses and on what apertures it occures.

[Jerry_R]

Sean Reid reports more noise in the corners...

Increasing brightness of corners, related to vignetting - always had been causing increase of noise. Nothing new.

[Advertisement]

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

[frank_dernie]

Hi Frank - I think this must be something else. It happens with wide angle lenses which have a very steep angle of incidence of light, and where lens correction is applied to reduce vignetting - it's much more prevalent wide open than stopped down, and there is little or no vignetting on the 50 1.4 asph.

 

Yet another reply from me Jono, after further reflection. Any colour shift due to the angle of incidence of light through the IR filter on the M9 depends only on the exit pupil of the lens in question, which depends on the lens design, not whether it is a wide angle or not, though obviously near symmetrical lenses for traditional RF cameras will always have a closer exit pupil with WA lenses, but lenses of the same focal length but different design can have a big variation in exit pupil position, hence a significantly different requirement in colour correction. The vignetting correction, if one is wanted, will be different again. With the external filter for the M8 it was the external rays passing through the filter which caused the shift, and therefore only dependant on angle of view.

This is more complex than the M8 requirement, but doable, with limitations, in theory. If the sensitivity is so great that the difference in tolerance between different bodies and lenses can have a significant effect, that is a tough one.

Luckily for me I am not as exigent as some when it comes to colour:-)

Frank

[Jerry_R]

One thing is certain- the have to fix it in FW.

At what cost, I don't know.

So far, it seems to be UNDOABLE, IMPOSSIBLE with current sensor and design of M9. Why? Answer is very simple. It can be only minimized.

 

Ones who used CornerFix know, that best is to create a pattern photo with YOUR CAMERA. Based on this - photos are corrected. But each camera can give different pattern.

 

Also a generic M9 patterm - is not perfect and causes different results on different cameras.

 

Ohh, maybe I am wrong. Maybe next FW will store similar pattern, created each time separately for different camera. Then, yes, it could be solution.

[ho_co]

... forget about any software corrections performed by camera body. Turn it off. ... I would not relate color shift to vignetting removal. ...

Seems counterintuitive to me. Wouldn't most people use most lenses with lens ID detection turned on? So shouldn't we test doing the same?

 

And when you say you wouldn't "relate color shift to vignetting removal," hasn't there been some indication that lenses perform differently in regard to this "Italian flag syndrome" depending on whether lens ID is turned on or off?

 

I think Jono's explanation is correct.

Edited February 8, 2010 by ho_co

[Jerry_R]

All that camera tries to do poorly, and users by guessing?!? - you can do under very good control, with better engine in RAW developer - assuming you use RAWs.

 

When you estimate lens\sensor behaviour - turn OFF any software corrections. Like portals testing lenses.

 

In fact - software corrections make sense for JPG shooters. I see no sense to use them when using RAWs.

[Alnitak]

Advertisement (gone after registration)

Seems counterintuitive to me. Wouldn't most people use most lenses with lens ID detection turned on? So shouldn't we test doing the same?

 

And when you say you wouldn't "relate color shift to vignetting removal," hasn't there been some indication that lenses perform differently in regard to this "Italian flag syndrome" depending on whether lens ID is turned on or off?

 

I think Jono's explanation is correct.

 

The lenses I have that have demonstrated the problem have shown it with the coding on or off, it didn't matter. It's not just a function of the profile corrections. In fact, I would argue that all the profile correction seem to do is to reveal it better by removing the vignetting.

 

Jeff

[ho_co]

Bravo, Jerry! Making your 100th post has made you an erfahrener Benutzer. Congratulations!

 

Jeff says he feels the problem appears worse with the lens ID turned on, because the vignetting removal has made it more obvious. To me, that supports my point:

 

IMO, the way to check for IFS is to work with a camera the way you'd work with it in the field. I doubt that many users will turn off lens ID just to be able to correct in post. That you're going to have to eliminate 'Italian flag syndrome' in post doesn't seem to be a reason to correct vignetting in post as well.

 

Just my opinion. But I may have been lulled by the M8's pretty reliable in-camera correction; I don't have an M9 yet, and my opinion may change.

 

But we're way off topic, and that's at least partially my fault.

Edited February 9, 2010 by ho_co

[Jerry_R]

;-)

I come outside of M system. Many topics related to optics, cameras and photography are not new to me. Still some aspects of M system are But am recovering quickly. And subject of 'Red Edges' is important to me, as I love wide angles.

 

Anyway - small sum up - many different lenses (not only M system) vignette whith wider apertures. So far - removing vignetting in post processing was just causing increase of noise (like other functions brightening some areas, called 'Dynamic Range Optimizer', 'Intelligent Exposure', 'Highlight Tone Priority', etc. called differently by manufacturers).

 

With M9 - we additionally to noise increase - observe color shift appearing, too. No matter, if vignetting is removed by camera software correction or RAW developer in post processing. And often it is so large, that is visible even without removing of vignetting, right?

[frank_dernie]

;-)

I come outside of M system. Many topics related to optics, cameras and photography are not new to me. Still some aspects of M system are But am recovering quickly. And subject of 'Red Edges' is important to me, as I love wide angles.

 

Anyway - small sum up - many different lenses (not only M system) vignette whith wider apertures. So far - removing vignetting in post processing was just causing increase of noise (like other functions brightening some areas, called 'Dynamic Range Optimizer', 'Intelligent Exposure', 'Highlight Tone Priority', etc. called differently by manufacturers).

 

With M9 - we additionally to noise increase - observe color shift appearing, too. No matter, if vignetting is removed by camera software correction or RAW developer in post processing. And often it is so large, that is visible even without removing of vignetting, right?

 

Pretty well all lenses vignette, it is a characteristic of rectilinear lenses. Certain optical designs can be used to reduce it. Eliminating it altogether is impractical in most lenses since it would increase the size and aberrations significantly. The element diameters would have to be much bigger than normal and the peripheral rays are the most difficult to correct. On most fast lenses, including the Nocti the light dropoff starts immediately foff centre, that is only the centre of the image is exposed at f1. If you look into a Nocti set at f1 and tilt it the visible aperture reduces immediately and the whole f1 "hole" can only be seen from straight ahead.

All manufacturers fast lenses are like this.

Frank

[adan]

C'mon folks, remember Noah's original plea:

 

"I wanted to start a thread to discuss which lenses exhibit this behavior and which do not. Let's avoid speculation about the cause and when/if we'll get a firmware fix for now."

 

There are plenty of speculation and theory threads running - if you absolutely, positively have something new to say in that vein.

[noah_addis]

C'mon folks, remember Noah's original plea:

 

"I wanted to start a thread to discuss which lenses exhibit this behavior and which do not. Let's avoid speculation about the cause and when/if we'll get a firmware fix for now."

 

There are plenty of speculation and theory threads running - if you absolutely, positively have something new to say in that vein.

 

Well, it's a public forum and I would never want to impede the free flow of information. Though I do hope this thread can continue to document which lenses do and do not exhibit this phenomenon.

 

I'm wondering if anyone with an M9 and a 21 or 24 'Lux can do a test for me--I'm wondering if you could do a test with either a neutral sky or white paper, and check for the red edge wide open, at f/5.6 and at f/11.

 

I plan to do the same with my 21 Elmarit Asph. I seem to be noticing that the red appears mostly in my outdoor photos. Seeing as how I often shoot in overcast light, and how my outdoor photos tend to be at around f/5.6 or f/8 and my indoor ones tend to be wide-open, I'm trying to decide if aperture has an effect on the red edge. I realize there are other variables--the color of the light as well as the indoor scenes I'm shooting may mask the effect to some extent.

 

The consensus seems to be that the new 'Luxes don't have the problem, but it could be because the people using them are possibly tending to shoot them wide-open. Or maybe it has more to do with the lens designs.

 

I'm not saying I plan to upgrade to a wide 'lux, but it's just a curiosity:rolleyes:

[Jerry_R]

Let's avoid speculation about the cause

I think all speculations are result of not precise question. What Noah admitted in previous post.

When? At what aperture? Outside? With sky or without? JPG, RAW? With coding? Without coding? Etc.

[Alnitak]

I think all speculations are result of not precise question. What Noah admitted in previous post.

When? At what aperture? Outside? With sky or without? JPG, RAW? With coding? Without coding? Etc.

 

In my experience, as I stated before, turning correction on or off hand no effect at all. The was either there with a lens, or not. The same with stopping down, no effect. It's there or not.

 

I am in the process of putting up some test photos with a 24 'lux and 21 Elmarit ASPH to demonstrate.

 

Jeff

[Alnitak]

OK, here are some demonstration images. First, the 24mm Summilux.

 

Wide open, f/1.4:

 

 

Then at f/5.6:

 

 

I see no red edge, regardless of aperture. Note the vignetting wide open; these are taken with lens detection turned on. Turned off, the vignetting gets ugly.

 

Now, the 21mm Elmarit-M ASPH. First, a set of shots with the lens detection turned on, so we have in-camera corrections taking place.

 

Wide open, f/2.8:

 

 

Now at f/5.6:

 

 

Split into the next reply to fit all the images.

 

Jeff

[Alnitak]

Personally, I see at most a tiny difference in the red edge wide open vs. stopped down. You can see red on the bottom edge most easily, as wide open the vignetting on the left side is so strong it hides the red. Note that the red is very mild with this lens. I personally don't mind it and it doesn't show up in most shots.

 

Now let's look at it uncoded. First, wide open, f/2.8:

 

 

Then stopped down, f/5.6:

 

 

To my eye, there is no major difference in the red edge coded or uncoded. There is a big difference in vignetting and cyan drift.

 

Anyway, take a look and draw your own conclusions.

 

Jeff

[adan]

Well, thanks, Noah! Since you left the door open....

 

'Way back when, when the M9 was still a pipe dream and the 21/24 'luxes first came out, two things occurred to me:

 

1. I posed the question, in that survey I did of what people's preferences were for a full-frame camera, of how much backwards compatibility (or lack thereof) would be required or acceptable. And one of the choices was "works acceptably with older wideangles - but requires new digitally-friendly, and probably bigger, SLR-like wideangle lens designs for top performance." Maybe that is exactly what we got with the M9 - although the inconsistencies from camera to camera, and the off-centeredness, still make me wonder what else is involved.

 

2. I noted that in addition to providing a fast aperture, the f/1.4 21/24 lenses also provide a large rear element, and in fact the whole design of those lenses, in cross section, looks more "telecentric" - a long tubular design with a more or less consistent internal diameter, as opposed to the pinched "two funnels tip to tip" light path in the f/2.8 lenses.

 

The lens cutaways are available through Leica's Technical Data pdfs for each lens. Links at:

 

Leica Camera AG - Photography - NEW: LEICA SUMMILUX-M 21 mm f/1.4 ASPH.

Leica Camera AG - Photography - LEICA ELMARIT-M 21 mm f/2.8 ASPH.

 

Therefore I also look forward to examples and reports regarding the 21/24 Summiluxes and red edges. (Thanks, Jeffrey!)

 

(BTW, Nikon just re-entered the f/1.4 wideangle fray with a 24 f/1.4 - Nikon releases 24mm f/1.4 G ED fast wideangle lens: Digital Photography Review - which also looks like a severely telecentric design. A 24mm lens that is 88mm long (plus camera depth) - a far cry from the Nikon Rangefinder 25/4 of the 1950's/60's that extended a bare 10mm in front of the camera body.)

Edited February 10, 2010 by adan

[innerimager]

I'm wondering if anyone with an M9 and a 21 or 24 'Lux can do a test for me--I'm wondering if you could do a test with either a neutral sky or white paper, and check for the red edge wide open, at f/5.6 and at f/11.

:

 

I'll get to it with my 21/1.4 as soon as I can Noah, probably later today using the 12-24 inches of snow due to arrive! best...Peter

[noah_addis]

I'll get to it with my 21/1.4 as soon as I can Noah, probably later today using the 12-24 inches of snow due to arrive! best...Peter

 

Yes, the snow does present a good opportunity. Don't want to kill my M9 in the wind and snow we're having right now, but tomorrow there will be plenty of snow to test for the red edge!

 

I wonder if Leica will introduce new f/2.8 wides if they can't fix the problem in firmware?

[Advertisement]

Hallo,

Look here: Leica M

[noah_addis]

Jeff--thanks for your test. I'll check my 21 as soon as the snow ends with both M9s to see if there is any difference between bodies.

 

It seems that with the 21 Elmarit in your tests the aperture doesn't matter. I think the slightly stronger vignetting in the f/2.8 photo masks the red edge to some extent.

 

Turning lens recognition off is not an option. The cyan corners are way more of a problem.

 

There seems to be a lot of in-camera processing going on to make the full-frame sensor work. Am I the only one bothered by this? The noisy corners at high-iso and the color issues are troublesome.