Alternative Method for correcting 486 cyan drift

[dhoelscher]

Advertisement (gone after registration)

...

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)

 

The only thing is, Jono - when you turn off lens detection for a mounted coded lens - you do see a change, right???

 

Wouldn't this imply that the firmware or hardware ROM or whatever really is doing something about the luminance vignetting ???

 

DH

[Advertisement]

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

[Jamie Roberts]

@ Jono--

 

Let's have no BIG sleeps around here

 

There is a lens table built into the firmware, so I do think Leica is making different changes to the vignette based on the lens detected. I suppose it might just be a calculation from the sensor, but if so, it's still effective.

 

And they can obviously change luminance in the DNG for the vignette, so I expect they could correct color as well (though I don't know that).

 

@ Scott--yes, we want wider and wider lenses but my point is that with the crop they all vignette less, probably, than they would if you were using them "full-frame" is all... in other words, we get the benefit of using the best part of the glass.

[scott kirkpatrick]

There's no question that it's there, but it seems to be wildly variable so far. Maybe it will get tuned somewhere around firmware 1.15. Read Sean's articles on the different lens families. He's found some that get undercorrected to date, and some that are overcorrected. I am not sure that Solms has received their ultimate light spheres yet. maybe they are outside under grey mittel-Europaische cloudy skies with their Pringles lids trying to get good data to make the corrections with.

 

scott

[jonoslack]

There's no question that it's there, but it seems to be wildly variable so far. Maybe it will get tuned somewhere around firmware 1.15. Read Sean's articles on the different lens families. He's found some that get undercorrected to date, and some that are overcorrected. I am not sure that Solms has received their ultimate light spheres yet. maybe they are outside under grey mittel-Europaische cloudy skies with their Pringles lids trying to get good data to make the corrections with.

 

scott

 

 

You've been watching too many editions of star Wars!

I stand (mostly) corrected . . . . but that left foot is still looking a little green.

[dhoelscher]

Hmmmm .... seems that Leica should now also provide M8 purchasers with their own "red dot" branded Pringles can lid ..

 

All Hail Pringles !!!

 

DH

[scott kirkpatrick]

 

@ Scott--yes, we want wider and wider lenses but my point is that with the crop they all vignette less, probably, than they would if you were using them "full-frame" is all... in other words, we get the benefit of using the best part of the glass.

,

 

True, but I sometimes shoot film with them, too. I think we are buying wide angles that would be very hard to control in full frame because of the 1.33. I never got much use out of my 19mm when I used it on an M2, and now as a sedate 25.3 mm-e lens, it feels much more comfortable.

 

scott

[tashley]

Advertisement (gone after registration)

Dave, I tried this for my CV15 (I can't get an old stlye bayonet adaptor with a full flange so I can't self-code, but I do have a filter) and it works a treat though I did find I had to reduce the exposure of the layer file by around two stops.

 

I used an aperture of f5.6 and focus at infinity since nearly all my use of this lens is at or around that combination.

 

Thank you!

 

Tim

[ChrisC]

I tried this for my CV15 (....[no].... self-code, but I do have a filter) and it works a treat .......

 

Tim - Given that many can't or won't self code lenses, I am intrigued and encouraged by your post. Are you able to post a suitable example? I am certain I am not the only one who would like to see your results.

 

...................Chris

[dhoelscher]

Dave, I tried this for my CV15 (I can't get an old stlye bayonet adaptor with a full flange so I can't self-code, but I do have a filter) and it works a treat though I did find I had to reduce the exposure of the layer file by around two stops.

 

hey thanks Tim!

 

I'm glad that it worked .... I'm still experimenting with the "best" way to take the mask picture .... it's difficult (so far) to come up with an exposure method that works for every situation ... I'm currently trying to fix pictures that have "dark corners," which causes problems with this method as it currently is detailed.

 

It seems for the most part, for most (relatively) evenly illuminated scenes, the "mask" should be exposed (or "developed" in the raw processor software) much as Michael Reichmann has said - that is, "to the right" such that the histogram bunches up at the right. Maybe I should say that the histogram should "kiss" the right, because the center of the cyan mask should be pure white (RGB 255,255,255).

 

More experiments to come.

[tashley]

Tim - Given that many can't or won't self code lenses, I am intrigued and encouraged by your post. Are you able to post a suitable example? I am certain I am not the only one who would like to see your results.

 

...................Chris

 

 

Here we go Chris, four images: first the DNG file I want to work on, then that DNG file sent to photoshop and layered with the mask in Color Dodge mode, then the mask I created by inverting my test shot, and finally the REALLY quick version made by opening the DNG in Lightroom and using the Targeted Adjustment tool to click in the cyan top left corner and drag the saturation down. It's not as good but it is VERY quick!

 

Hope it's interesting, unlike the photos themselves - I've been confined to the house today awaiting various deliveries of all the stray bits and pieces required to make an M8 work...

Tim

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!

Original

-------

Using Dave's Method

-------

The Mask

____

The down and dirty method (good for sky, not so good bottom left!)

[Jamie Roberts]

Here's a down and dirty LAB adjust of your pic, Tim (hope that's ok!)

 

This was done by converting the shot to LAB, making an elliptical selection and feathering it to about 130 px (this is much smaller than the original, of course, so you need to make a bigger feather to make this happen on an original).

 

Since LAB works independently on Luminance and Color, this is a single adjustment curve, which you could simply save as any other curve.

 

1. Convert to LAB mode.
   
2. Make selection. David's templates would really, really help here!
   
3. Feather (200-300px for a full-size shot)
   
4. Invert selection. Now you have the vignette and cyan shift selected.
   
5. Make a selection point with the eyedropper near the edge of the frame and at the worst cyan / vignette point.
   
6. Now, if you have a template, you simply add an adjustment layer and adjust the luminance (L), magenta / green (A) and blue / yellow ( till the LAB values match the middle. Voila; correction is done.
   
7. Apply the mask and curve in a new layer over the main photo. Mess with mask using normal PS tools, and adjust transparency to suit.
    
   If you don't have a mask (which I didn't in the shots below), then you need an unshifted reference from something in the middle of the frame. I used the sky, which gave me the luminance and AB values I needed at the edges.

That's it! Convert back to RGB and go from there. Vignette and cyan shift taken care of.

Honestly, once you do this a couple of times it takes longer to describe than to do. And if you have the templates to make your selection, then you can automate this adjustment with an action. You may have to tweak for critical adjustments, of course.

 

(sorry for the pixelating here, the screen shot program I'm using wasn't set quite right--but you get the idea)

 

Here's the original:

 

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!

 

Here are the curves applied:

 

 

 

 

And here's the final, adjusted file; not perfect, but very quick in PS:

 

[LJL]

Jamie,

This seems to be an approach that has a awful lot of merit....fix the vignetting in the L channel and fix the cyan drift in the color channels. I was thinking this same approach while reading your reply to Bill Parsons in another thread.

 

O.K., here is where I sort of cross the line between supportive and a bit annoyed, so bear with me please.....

 

Your entire approach to this is predicated on being able to do these corrections in something like PS. That is fine, and that is a tool that most of us probably are more than familiar with, as well as having and using. However, why cannot a series of these sorts of Lab layered curves be made by Phase One for use in C1 (Pro AND LE), since this is the "preferred RAW conversion app" that Leica is promoting and shipping?

 

While one of the best solutions may be to have everything working without filters, and such, we know that is not going to be the case. Also, as Leica continues to talk about the ability of the M8 to use older M lenses, some of which may not be able to be coded, or may take incredibly long to get coded, there should be some available method to use the M8 as easily as Leica promotes. (BTW, the use of non-coded, or non-Leica, non-coded lenses, such as the CV15/4.5, etc., would immediately benefit from having these sorts of curves built into a simple selection menu....first in the camera for directly working on the RAW files at time of capture, and/or secondarily in the preffered conversion app that is shipping with the camera.)

 

I know all of this has been hashed and rehashed time and again, and I really do not want to revisit that. And yes, we are all eagerly awaiting firmware 1.10 to arrive and have lots of great solutione within. That does not mean that there should not be some sort of fairly simple and elegant solution like this built into C1 (and hackable to other RAW conversion apps such as RAW Converter, LR, Aperture, ACR, LightZone, Bibble, etc.) to permit users to work on files as their workflow and preferences demand. This should NOT be all that hard to do.

 

O.K., enough of my mini-rant here.

 

I like this sort of solution, but honestly, it needs to be nearly automated for folks that are shooting lots and lots of images for clients, etc., and cannot afford to be making Lab layer masks for each image. It is hard enough trying to remember which lens was used, at times, so providing some drop down menus in C1 or wherever to test various effects like this would be a wonderful solution in my thinking.

 

LJ

 

P.S. Jamie, please recognize that my "rant" was NOT toward you at all. I am grateful for your efforts and all the help. My "rant" is directed toward Leica and Phase One, since they are the folks supposedly working together on this. Sorry for any confusion.

[Guest guy_mancuso]

I agree this needs to be in C1 and is automated or the use of sliders to make this so MUCH easier to deal with. OR NOT THERE AT ALL as we patiently wait for 1.10

[Jamie Roberts]

@ LJ and Guy--

 

Hey, I couldn't agree more. You *shouldn't* have to do this kind of thing in PS, and I'd like it to be in FW in the camera.

 

And in a way, that's also what I'm trying to show; in theory, you can make a fairly simple curves adjustment in the right colour space if you know where the vignette is occuring, and you can get rid of an awful lot of it!

 

Evidently, with coded lenses, Leica has a pretty good idea about where the vignette and shift is...so it's definitely possible that they can correct this (some people were thinking they couldn't).

 

So that's one thought. It would also be nice if Leica put a manual look-up table in the menu, too.

 

The other thought is that for older lenses and uncoded lenses, yes, this should absolutely be in C1. A lot of it *is* in ACR already; I just don't want to use ACR

 

IIRC, C1 also has this exact feature in its support for MF backs, where this happens, too. So it would be nice if they added it particularly for M8 support. I know they've said they're going to have "lens correction support" in C1 V4.

 

BTW--LightZone would also be a natural to fix this in RAW, but you're back to making selections on a per shot basis. At least in PS, you can make this into an action and batch a bunch of shots. Of course, if you can't remember what lens you used, or can't organize them, that's going to be tricky.

[tashley]

I agree this needs to be in C1 and is automated or the use of sliders to make this so MUCH easier to deal with. OR NOT THERE AT ALL as we patiently wait for 1.10

I agree also. Much as I love my M8, the amount of 'glue and string' needed to get the best out of it is stupid in this day and age. I've been trying to hand-code, cutting out templates, using marker pens for Lord's sake, buying different adaptors for lenses with different flange widths, learning and relearning new and ever more arcane PP methods, buying a 1.35X scre in magnifier (which I have to take on and off) just so I can focus, dealing with poor quality control on backfocusing lenses, trying to work with a flash gun that doesn't have a swivel head, blah blah blah. When it's all over and i look at the images I am very very happy. When I use the camera, I love it. Everything in between is farcical!

 

Tim

[dhoelscher]

Make selection. David's templates would really, really help here!

 

Hi Jamie!

 

I think that I finally understand better what you've been proposing ...

 

With regard to templating, would it be of any help for me to make a project of posting the cyan masks of various lenses at various apertures? I would guess that you could make a selection based on color, right? And then apply your methodology.

 

What would be the best way to select the cyan in a mask? selective color? magic wand?

[Jamie Roberts]

@ David--

 

Yes, if we can come up with the perfect way of selecting the vignette, and of mapping it in the mask, then we can perfect the curves to nullify it.

 

Personally, I've been eyeballing it and playing with the feather! I'm thinking a gradient mask mightbe a very good way of dealing with this. As for selecting the cyan, I'd boost the saturation and play with the magic wand, maybe?

 

So if you want to post one DNG with the vignette, we could play a bit.

 

I'm sure other people have good PS ideas too. It's a cute correction puzzle, if you like those kinds of things.

 

@ Tim--I agree on the hoops you've got to go through here; seems a bit much! Let's hope the RAW converters or the firmware have a way of automating this!

 

BTW--from what Rob Stevenson has posted, I think Panotools does a great job here.

[tashley]

Thanks Jamie - I will give this a try though Jono and I have an old joke that if you find yourself using Lab mode, you're not getting out enough! ;-)

 

Seriously, what are the advantages of doing it this way? It seems more complex and at screen res at least the results seem similar. My idea would be to de-vig the luminance to the same degree in LR on both the actual and the mask files, then go with Dave's method.

 

Any thoughts

 

Tim

 

ps alright I've never really understood Lab mode....

[Advertisement]

Hallo,

Look here: Interesting Leica gear

[dhoelscher]

What would be the best way to select the cyan in a mask? selective color? magic wand?

 

Duh .. to answer my own question, all you would have to do is desaturate the cyan mask, and now you have an instant layer mask for selection!

 

OK, looks like it's time for me to get busy on this project ... what shall we call it .... Project Pringle Fairy?

[dhoelscher]

BTW--from what Rob Stevenson has posted, I think Panotools does a great job here.

 

It's weird though, Jamie ... try to use PanoTools on an image from the CV 12mm f/4.5 Heliar ..... I really couldn't get acceptable results .... don't know why.