Alternative Method for correcting 486 cyan drift

[Jamie Roberts]

Advertisement (gone after registration)

To the best of my knowledge so far...Luminance vignetting does vary by aperture and, to a lesser extent, focus distance. Cyan drift (red vignetting), in isolation, does not seem to vary according to aperture but since both kinds of vignetting appear together in the file, one can influence the appearance of the other.

 

Cheers,

 

Sean

 

Sean, that's interesting. Only having one lens with a filter makes it tough for me to generalize, but this sounds like another reason to go to a LAB correction.

[Bob Ross]

Hey thanks, Bob!

 

Great point, cool idea .... I'm now thinking that either an ExpoDisc (if you have one big enough) or a piece of translucent plexiglas or something like that would be easier to photograph for these corrections than pointing the camera at a sufficiently large enough white background or foam core ...

 

I'll post today's experiments later this afternoon ...

 

DH

Hi Dave,

I use a 55mm Expo Disk (from way back when they were introduced), with step up rings and can use it with all my 28mm to 135mm lenses. I thought of using a light box, but mine isn't that evenly lit. Your approach has the advantage of tayloring the "filters" to your particular lens unit, rather than relying on the camera's generic correction. If your particular lens vignetts or color shifts more on one side (not unusual) there could be noticeable affects with a uniform generic correction in-camera or in PP. An alternative could be the Pringle lid idea or a Tupperware lid, in order to get something big enough.

Bob

[sean_reid]

Just move in close. It doesn't have to be in focus.

 

I would double check that. The penumbra may be different if the wall/board is out of focus. I haven't tested for that specifically (I do my board tests with focus at actual subject distance) but my guess is that penumbra roll-off for the luminance vignetting may be different when the subject is out of focus.

 

Cheers,

 

Sean

[dhoelscher]

I am scheduling myself to do a whole bunch of tests this afternoon in the beautiful Austin sunlight to try to determine if this is so ...

 

DH

[sean_reid]

Sean, that's interesting. Only having one lens with a filter makes it tough for me to generalize, but this sounds like another reason to go to a LAB correction.

 

Scott K. wrote an interesting article about this (based on my vignetting/cyan drift test examples) and its on RR (linked from the 28s article).

 

Cheers,

 

Sean

[Guest magyarman]

Best way fix up first vignetting, after fix up cyan what stays. Is better works except try to do in once.

[jonoslack]

Advertisement (gone after registration)

Well - I'll be astonished if 1.1 sorts out the Cyan drift - or any other firmware come to that.

 

Kodak had a similar problem with some lenses with the Kodak 14n. I'm not sure if the cause was the same (it was never really discussed) but the effect was exactly the same - although not nearly as obvious.

 

I did a lot of testing at the time, using spreadsheets with distance / aperture / ISO / White balance / lens example / sensor , and ALL of these things had enough of an effect to spoil a picture - especially if you knew about it.

 

With the Kodak, tables were put in the firmware for different lenses, and of course the Aperture could be recorded and used accordingly (unlike the Leica). There was a routine for setting up a lens, where you shot it wide open and at f8 (as far as I can remember) on a white card.

 

There was, in fact, a great deal more information available to the camera, and still it was never fully resolved - it was fine with most lenses, bad with others, and with some lenses one example would be much worse than another.

 

Maybe this is quite a different issue, but as I understand it, a cyan - magenta - cyan shift across a frame is going to have so many variables it's going to be very very very hard to fix it in firmware (let's hope I'm wrong).

 

I think that David's approach in post processing is likely to be the best solution.

 

My gut reaction is that it will be harder to deal with the cyan drift than it is to deal with the IR problem - in neither case can the camera hold enough information to be able to sort it, but at least the IR problem is even across the frame.

 

It's not like me to be a misery - but I really wouldn't get one's hopes up too much about firmware 1.1.

[dhoelscher]

Hey Jono!

 

It won't be miserable .... just a few things to consider ... thanks very much for your valuable input ... I hope to have an "alternate" alternative workflow posted this afternoon that should make it easier for everybody to do this mask thing ...

 

DH

 

PS - Wow ... maybe we'll still do something like this even for coded Leica lenses and firmware 1.10 ... who knows ???

[dhoelscher]

Ok ... after some more tests, I'd like to share what I've found so far .... this is far from being complete ... but so far, so good!

 

First, I'd like to thank everybody whose work I have built upon, notably Sean Reid, Guy Mancuso, Jamie Roberts, Scott Kirkpatrick, Bob Ross, Jono Slack, and many others. Really everybody builds up the base of knowledge starting from other people's hard work. I'd also like to thank the Queen, my Mom, and the probe "Nomad" from the original Star Trek series.

 

DISCLAIMER: I'm a rank amateur, I don't know how to take nice pictures, I'm just a technician posting bogus test pictures of no artistic value, Your Mileage Will Vary, this surely can't be the only acceptable approach to the problem(s), I haven't had a chance to give Jamie's workflow a thorough testing (yet), I can't be responsible if your wife leaves you because you're out in public waving around a piece of translucent plexiglas ... etc. etc.

 

Now, moving on ....

 

 

 

You're asking, but Dave, this is a PITA - why should I have to do this ??

 

Well, good question ... You certainly DON'T need to do any of this if:

 

 

 

1) You choose to shoot without a 486 filter .... a perfectly reasonable option, depending on your subject matter ... check out this cool set of pictures taken with the M8 and the Cosina/Voigtlander 12mm f/4.5 Heliar lens WITHOUT any sort of filter ... beautiful!

 

http://www.leica-camera-user.com/digital-forum/16853-m8-voigtlander-12-5-6-a.html#post176355

 

2) You intend your final image to be Black and White (either JPEG right out of the camera, or in post-processing a DNG) - check it out; it really doesn't matter !!!!

 

3) You have a 6-bit coded Leica lens with M8 firmware version 1.10 or later (we'll see eh Jono?)

 

4) You have hand coded / CNC-machined, or otherwise used black magic on a non-Leica (or Leica) lens to emulate some standard Leica lens that firmware 1.10 and later will recognize

 

5) You are shooting with a lens with a focal length greater than about 35mm (seems cyan drift isn't really a problem here)

 

 

So you say, Dave, I'm a JPEG shooter and I don't want to use a raw processor - I say fine and dandy, just have the same WB set for both your "mask" picture and your image JPEG and proceed away as below.

 

 

Now, if you still want to torture yourself ... check this out ....

 

Sean - you are right about luminance vignetting interacting (superimposing) with cyan drift, you are right about focal distance, you are right about everything.

 

Scott K. - interesting stuff on Sean's site .... thanks for sharing !

 

So, this method below takes all this into account into the correction and isn't really all that difficult once you get used to it.

 

MATERIALS AND METHODS

 

You will need:

 

1) Your trusty M8 with your lens of choice with the 486 filter mounted

 

2) A WhiBal, or neutral object to white balance with in post (OK JPEG shooters - you don't need this - neither do you people who don't really need critical white balance - after all WB doesn't really need to be accurate all the time, just "pleasing" for the particular image in question)

 

3) an ExpoDisc, or a piece of white translucent plexiglas, Lexan, or maybe even frosted glass - or maybe even just a white handkerchief or something

 

 

 

Ok, so now you're ready for a shoot.

 

STEP ONE - Take a picture of your WhiBal or neutral color reference in the major light source of your picture, e.g. sunlight (JPEG shooters can skip this, or everybody else who doesn't care)

 

STEP TWO - Using your M8 in MANUAL shutter speed mode, compose your picture, set your aperture, set your shutter speed manually to where you want it (this helps to keep the exposure settings constant for the mask exposure next)

 

STEP THREE - Take your picture(s) - chimp, and when you're happy ....

 

STEP FOUR - Put your ExpoDisc directly up against the 486 filter (or translucent white plexiglas or whatever) and point your camera directly at your major scene illumination (e.g. the sun or whatever - just like normally using the ExpoDisc) and take another exposure

 

(this step ensures that the mask is taken with exactly the same focus distance, aperture, and shutter speed as your picture to be corrected !)

 

STEP FIVE - Load up your stuff, fire up the computer, and open the second "mask" exposure that you took with the plexiglas or ExpoDisc in your favorite RAW converter and profile of choice . IMPORTANT - You must process the mask in the same RAW converter with the same settings as you will process your picture.

 

In your RAW converter, use the white balance dropper (or equivalent) to click in THE CENTER of your mask image to make it white (do not use the WhiBal reference yet) This compensates for any coloration in your ExpoDisc or handkerchief. Make sure that you click the center, to avoid any cyan drift coloration. Save out this file and call it "Cyan Mask" or something similar.

 

Next, open up your WhiBal reference shot and your picture; use the dropper on the WhiBal to WB your picture - next, process your picture using your favorite RAW converter settings/profile (must be the same as the mask above).

 

STEP SIX AND A HALF - In Photoshop, open up the "Cyan Mask" that you created, and invert it (Image>Adjustments>Invert). Next, open up your picture. Go back to the Cyan Mask file, Select All, and Copy. Next, paste into a new layer above your picture. Change the blending mode to "Color Dodge."

 

Next, you may find a little layer transparency fiddling will help fine tune your image.

 

STEP SEVEN - Enjoy how clever you are.

 

 

 

See before and after below. This picture was taken with the Cosina/Voigtlander 12mm f/4.5 Heliar lens with the 77mm filter adapter fitted with a 77mm B+W 486 filter at f/8.0; mask exposure used the ExpoDisc I had lying around.

 

 

 

 

Thanks,

 

DH

 

Austin, Texas, Earth

 

"The Creation of Perfection is No Error" - the probe "Nomad" - from the original Star Trek series

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!

[jonoslack]

I can't be responsible if your wife leaves you because you're out in public waving around a piece of translucent plexiglas ... etc. etc.

 

Having seen a picture of Guy's wife, I don't think she looks like the kind of person who will put up with careless waving of plexiglass . . . . I know mine isn't!

 

David - this is a lot of work, it was also one of the most entertaining posts I've ever read hereabouts, and I suspect that in the final analysis this is going to be the right approach to dealing with these issues.

 

Whatever - you've done a grand job, and raising a smile as well? I hope your sandals never rot your socks:p

 

Congratulations

[scott kirkpatrick]

Norm Koren at Imatest - image quality evaluation software is a great resource for short introductions on how to really do optical testing right, but with limited expense. He has a module in his toolkit that studies "light falloff" i.e. vignetting. He can produce contour plots by channel or for luminance, but he doesn't offer a good way to separate the effect on the reds from the overall effect on luminance. You might want to read his discussion on vignetting anyway, because he is very clear. He favors setting the lens at infinity and using the most uniform possible source of white light at the closest possible distance. He describes several ways to create the light source. If cost is no object:

 

 

but in fact, he then argues that a sheet of opaline plastic is as smooth a diffuser as you will need, and I suspect he'd then go outdoors in indirect daylight and use the sky as the source -- in effect, Norm is on the side of the Pringles folks!

 

scott

[dhoelscher]

... you've done a grand job, and raising a smile as well? I hope your sandals never rot your socks:p

 

Congratulations

 

Hey thanks, Jono!

 

This will continue to evolve; I'm still discovering things about this ... Thanks for your encouragement!!

 

DH

[dhoelscher]

in effect, Norm is on the side of the Pringles folks!

 

scott

 

Thank goodness for that!

 

Can you imagine such a laboratory-grade setup as an accessory to your M8 ???

 

But, of course, this is the type of stuff that Solms has hidden away in their secret laboratories, so we should expect future M8 firmware updates to be far more clever than poor silly people in Austin like me!

 

DH

[Guest guy_mancuso]

Yes Jono a little paddy wagon would certainly be coming around the corner with a very nice white suit inside with very fashionable straps to pin me up and take me away. Or I would get if Leica is asking you to do this your are one crazy SOB. ROTFLMAO.

 

David seriously though excellent work here, it certainly is a way to go to solve the pressing cyan issue. Just think what Leica is doing to fix this cyan cast issue, now i may have to call the local white truck in Solms and send in a few white suits for those guys, there going to need them. ROTFLMAO

 

I think my new signature should read "Are we having fun yet"

[scott kirkpatrick]

here's a simplistic picture that might be worth keeping in mind to separate the two effects -- overall and red vignetting -- and the three factors, angle of view, angle from the exit pupil to the edge of the image, and lens design that are at work.

 

Overall (luminance) vignetting results from the fact that the light hitting the edge of the image has further to go (it spreads), hits the film or imager at a less effective angles (but offset microlenses help here), and sees the aperture obliquely, making it an ellipse instead of a circle and thus smaller. A second factor, lens design, can make luminance vignetting worse at wide apertures -- the edge of the image can't see the full aperture because it is stuck far above in a narrow tube and blocked by the lenses' own light baffles. But looking through the predicted vignetting curves that Leica offers in their extensive technical data sheets, it seems that it can't be that simple. Some lenses have wide aperture vignetting corrections that seem like a dramatic cutoff at large radius, no effect near the center, while others show a gradual correction across the whole image as they open up.

 

Red vignetting without a 486 depends only on the angle the light makes behind the lens. There is some in the M8 because there is still a (very very thin) green filter back there, and the filter gets thicker when you pass through it at an angle away from vertical. But the angles that you encounter in front of the lens are greater than behind it, and the increase in red vignetting as the angles increase with a dichroic ("cut") filter is more dramatic than what happens with absorbing green glass, so the second effect, the angle of view in front of the lens, is a bigger problem than the first effect.

 

scott

[Jamie Roberts]

{snipped}

My gut reaction is that it will be harder to deal with the cyan drift than it is to deal with the IR problem - in neither case can the camera hold enough information to be able to sort it, but at least the IR problem is even across the frame.

 

It's not like me to be a misery - but I really wouldn't get one's hopes up too much about firmware 1.1.

 

Hi Jono--

 

My gut is in complete agreement with yours, except for one maybe not-so-minor fact: Leica has vignette control built into the camera already to deal with the rangefinder vignette, and microlens capability the Kodak probably didn't have.

 

So if they were coming at this fresh, with no sense of how to correct the vignette and the amount, I'd say you were right.

 

But as it is, they have the data they need (I think), so with any luck at all it will be a matter of tweaking the red response (for JPEGs anyway).

 

How they do that in RAW I have no idea, but they also fix the vignette in RAW, so maybe they can fix the cast too.

 

FWIW, I've been pretty successful in LAB just with a simple curves adjustment so far. You'd never know there was a vignette or a cast (but the technique is still in the rough stages)

[jonoslack]

 

I think my new signature should read "Are we having fun yet"

 

Hi Guy

Haven't we all been having BIG fun since early November. What starts as an irritation with mitigating circumstances ends up as a challenge with beautiful side effects.

 

As for wives - there was a look in your lady's eyes which I recognised - I'm sure she'll back you to the hilt when you need to buy a WATE - but wasting time on something silly?

[Jamie Roberts]

{snipped} A second factor, lens design, can make luminance vignetting worse at wide apertures -- the edge of the image can't see the full aperture because it is stuck far above in a narrow tube and blocked by the lenses' own light baffles. But looking through the predicted vignetting curves that Leica offers in their extensive technical data sheets, it seems that it can't be that simple. {snipped}

 

Hey Scott, as for lens correction and (common) photographic vignetting, at least we're spared the worst of that because of the 1.33 crop, no?

[scott kirkpatrick]

Hey Scott, as for lens correction and (common) photographic vignetting, at least we're spared the worst of that because of the 1.33 crop, no?

 

Well, it's all relative. I consider myself pretty conservative, but I went out and bought a CV15 to overcome that 1.33. And I hear a lot of noise about CV12's and expect to see someone lash up a 7 or 8 mm focal length any day now (with a filter of course).

 

scott

[Advertisement]

Hallo,

Look here: Interesting Leica gear

[jonoslack]

Hi Jono--

 

My gut is in complete agreement with yours, except for one maybe not-so-minor fact: Leica has vignette control built into the camera already to deal with the rangefinder vignette, and microlens capability the Kodak probably didn't have.

 

Hi Jamie

Nice to know that our guts are in agreement - I was beginning to doubt that mine would ever be in agreement with anything.

 

I'm afraid I'm no techie, and I never really understood why the cyan drift occured in the Kodak (it may be something completely different).

 

I don't think that Leica have vignetting control built into the camera at all - just that the offset on the Kodak sensor works really well (as it obviously does) - but at present there isn't even a specific lens table built into the firmware (I don't know if you used the SLR/n, but there was scope for specific lenses at specific apertures).

 

Having said that - I'm still not convinced that it's work using IR filters for Wide Angle lenses, and I wonder whether the IR won't be easier to deal with than the Cyan Drift . . . . but now I'm repeating myself! Time for a last glass of wine, a large expresso and a long (rather than BIG) sleep.