C1 Lens/Filter Correction

[Ken R]

Advertisement (gone after registration)

What would really be great is if Leica would get Phase One to build lens/UVIR correction into the processing step withing C1. With that functionality removed from the camera, we could apply appropriate cyan correction without worrying about a lens have the correct 6 bit code information selected.

 

I imagine Leica's resoning for not going this route is that it would make it easier for M8 users to buy other brands of lenses. On the other hand though, it would make ALL m and screw mount lenses compatible with the UVIR filters, not just the lenses on the Leica list.

 

Just a thought. Wondering if anyone on the list has brought this up. In years past I had a very good relationship with Phase One management due to my dubious disticntion of making the largest single Phase One camera purchase in their history. That was some years ago now, so my moment of influence has most likely passed.

 

Ken

[Advertisement]

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

[Guest tummydoc]

Yes I've mentioned it several times, but people are intent on arguing whether Leica is trying (or is justified) in "closing" the M8 "system" to 3rd-party lenses. The fact is they have also "closed" the system to all their own Leitz and Leica lenses not on their list. My valuable 35 Summilux-ASPHERICAL for example. Perhaps the code for the ASPH would be a perfect fit, but I'd have to be completely daft to mutilate the lens by milling code wells, and the sharpie marker rubs off too easily.

 

There is a free plug-in for Photoshop called Panotools that has a feature called Radial Luminance that quickly and painlessly corrects the cyan drift. Surely it would be easy to include such a function in C1. The problem is, each lens needs a different correction value, but without the lens being identified in EXIF, how are you going to separate the files into batches taken with each lens? What I've suggested is that Leica add a simple numeric identifier that would put a user-selected tag in the EXIF. You could assign a number to each non-coded lens, and be able to sort them into batches easily after downloading, then run each batch through the specific PS action tailored to the cyan drift of each lens.

[Jamie Roberts]

I want them to do this too. It would be very easy to do a couple of things to make this quick...

 

1) Include a slider that modifies the vignette and cyan reduction so you can eyeball it quickly

 

2) Inlcude an algorithm that measures, (in CMYK for argument's sake), the amount of C as it changes from the center to the edge of the frame and auto-correct the RAW RGB rendering. If it changes significantly (as defined by testing), then correct.

 

This is essentially what I do on an adjustment layer in LAB.

 

And yes, it will be wrong if you're taking pictures of the sky with a hot-air balloon in the middle of the frame. But cyan itself isn't, fortunately, all that natural a colour. But in cases where it doesn't work, turn this off

 

3) Have a simple way to save your preferred corrections by name (where you could put lens / type / aperture anything else you wanted.

 

4) Have 1 & 2 work together, so you could tweak and save the auto-correct.

 

Since the next C1 is rumoured to have a plug-in structure like PS, there's every chance in the world that this could happen--and C1 wouldn't even have to do it themselves!

[Guest tummydoc]

Yes, but imagine 1000 shots made with an assortment of lenses from 12mm to 35mm exchanged at random. You could be there all night trying to decide which preset to use on each shot. However if the user assigned a number to each uncoded lens that he would input to the M8 via a menu each time he changed lenses, and that number posted to the EXIF, it could be recognized by C1 to automatically call up the appropriate preset.

[Jamie Roberts]

Yes, but imagine 1000 shots made with an assortment of lenses from 12mm to 35mm exchanged at random. You could be there all night trying to decide which preset to use on each shot. However if the user assigned a number to each uncoded lens that he would input to the M8 via a menu each time he changed lenses, and that number posted to the EXIF, it could be recognized by C1 to automatically call up the appropriate preset.

 

Oh--I don't need to imagine it--I do it... every time I shoot a wedding And more than 1000, too, from 3 cameras no less!

 

However, the 1000 shots aren't exactly random--I've never taken 10 shots in a row and changed lenses on one body anything like 10 times!--and you don't have to (as you point out) do anything above 35 mm...

 

But let's stay with your supposition for a minute...

 

If I was changing lenses that randomly, well, I'll tell you that when rapidly changing many lenses quickly--the very last thing on earth I want to do is mess with a menu when I should be shooting!

 

But this is why the in-post "auto detect and adjust" for the cyan change (with a smart save) is so tempting.

 

If you can calculate an excess cyan drift "use profile" over time (eg by "teaching the program" what the drift looks like to correct, as well as having some test shots as guides in your shot "database"), then the program can get "smarter" about your lenses, and how you use them.

 

So if the efficiency got to between 80 and 90 percent, well, then I'd be happy.

 

For me--and for many others as well--I don't ever print something colour critical (like wedding album prints) straight out of the RAW converter. So it's going to get tweaked in PS anyway.

 

That's usually about 100 shots out of 1000, so if 80% of them are fixed in the RAW process, that's fabulous for an uncoded lens set.

[scott kirkpatrick]

What nobody has said is that applying corrections during the raw file generation, pre-interpolation, as leica appears to do, gives you a more accurate result, since the transformation into the 8 bit compressed format only occurs after that. At least that is my hope. Howver, the result is only about right, not perfect, and vignetting effects remain without the color shifts, even in the best case. So I think we need good corrections, and then good PS or C1 tools for the perfectionist to use in finishing the job.

 

scott

[Guest tummydoc]

Advertisement (gone after registration)

Jamie, now I get what you're saying. Have C1 detect and correct the cyan drift itself without needing to know what lens was used. That would be great. From my experience so far with the firmware, it isn't nearly as sophisticated. Someone decided how much cyan to correct for with each coded lens and fixed it into firmware, and in different light if I look closely I can see where the default correction is either too little, too much or (mostly) ok.

[Jamie Roberts]

Vinay--

 

Yes--it could build up a little database of your kinds of uses, too, sort of the way P mode on a camera has any number of pre-arranged settings for different metering patterns etc...

 

Like I said, cyan is un-natural If you detect a variable, vignette-like increase across the image edges in cyan, then well, it's probably the result of an IR filter-microlens interaction

 

@Scott--yeah, it would be nice to do this pre-RAW writing. But you know what? I've been able to correct this in LAB on an 8bit JPEG, and it looks just fine. So I'm not worried about working in the RAW converter...