LCT
ex-newbie
6-bit coding does not correct for distortion on the M240 whereas it does on the digital CL.
sepiareverb
genius and moron
That was beautifully explained Godfrey. Thank you.
WJJ3
Well-known
The lens corrections in the profiles in the M bodies is a half measure that doesn't really work. The body can't know the aperture or focus distance so how can it correct for any aberrations?
M bodies have a light detector that, in conjunction with the internal light meter, provides an estimate of the aperture selected. The detector is located on the front surface of the top cover in front of the hot shoe.
WJJ3
Well-known
It seems this system was done away with on the bodies from M10 forward because it doesn't work very well...
Godfrey
somewhat colored
Imaging aberrations between a lens optical system and the optical system of a specific camera/sensor are a known property that does not deviate based on aperture or focus distance, so Leica optical and software engineers can calculate the corrections required (because they have specific, intimate knowledge of each Leica lens's optical design and properties) without precise information on the aperture and focus distance used in shooting.
The external light detector is an aid to the user to allow the lens opening in use to be dynamically estimated in the viewfinder display and embedded in the captured image metadata because M-mount lenses have no direct connection to the body through which to report lens opening. It does not affect the exposure evaluation/settings of the camera in Auto mode at all ... those are determined by the light falling on the metering sensors inside the camera and are time based per the ISO setting and actual amount of light falling on the sensor due to the actual lens opening.
This external-sensor driven aperture display system was meant purely as an aid and can't be 100% accurate. Leica has gone from including its estimate in the EXIF data to excluding it because it has proven to be a source of confusion for users who, for example, may know that they had f/2.8 set as the aperture and the estimation system might produce a result of f/1.4 to f/3.5 depending upon variability in the conditions of shooting and what's falling on the external sensor. For me, I'd rather have no data than data which is not useful due to variability, so removing it in later bodies and firmware revisions simply reminds me that the number it reports isn't entirely reliable.
The only interaction between this aperture estimation system and the six-bit coding is that, with a lens profile selected, the estimation system knows that a particular lens cannot be open wider than its defined maximum aperture value defined in the injected EXIF data.
(Note that L mount lenses have a direct electronic readout of aperture and focus distance that is communicated to the bodies they are used on, so these lenses don't use an external aperture estimation sensor at all.)
G
WJJ3
Well-known
The profiles in my M9 apply a correction for falloff, an aspect of lens performance which surely you know is dependent on aperture.
charjohncarter
Veteran
My DSLR has turn off corrections (4), but two are on by default. Raw Therapee has easy sliders to correct in post.
https://rawpedia.rawtherapee.com/Lens/Geometry/de
And also has this to me confusing statement:
This "Chromatic Aberration Correction" tool in the Transform tab works on the image after demosaicing. The Chromatic Aberration tool in the Raw tab works on the image before demosaicing.
https://rawpedia.rawtherapee.com/Lens/Geometry/de
And also has this to me confusing statement:
This "Chromatic Aberration Correction" tool in the Transform tab works on the image after demosaicing. The Chromatic Aberration tool in the Raw tab works on the image before demosaicing.
Godfrey
somewhat colored
Most lenses show variant fall off depending upon different apertures and it is normal. That's not what Leica is addressing with a lens profile: Leica is NOT trying to 'correct' all imaging characteristics of their lenses. They are providing a solution so that the lens's imaging behaviors are consistent on all the sensors/cameras they provide.
The light falloff they are correcting is invariant, the result of interaction between the lens design and the ray trace to edges and corners of the format combined with the optical geometry of the different sensors/cameras that lens might be used on.
The natural falloff of illumination at different apertures is a part of image rendering that photographers take advantage of, thus there is no reason to try to force all apertures to produce perfectly even illumination across the imaging plane. To do corrections like that across the board would eliminate the often-praised "character" of a Leica lens design.
G
Godfrey
somewhat colored
In the post-demosaic case, the image is already organized into R,G, and B layers, so correcting lateral chromatic aberration is a matter of "rubber sheeting" the three primary color layers so as to align the pixels and eliminate the red/green edging.
In the pre-demosaic case, the image is not yet organized into R, G, and B layers and driver factors in the implementation that the demosaic operation uses during the conversion process can be tweaked to pre-align the color layers' pixels.
So RawTherapee is simply telling you that this correction is implemented in these two different ways dependent upon whether the image data you have as original is raw or already demosaiced. I presume that if your original image file is a JPEG or TIFF, the Raw tab is not accessible, whereas if the original image file is a raw file, you can manage lateral chromatic aberration with either algorithm.
However, this has nothing to do with Leica six-bit lens codes and lens profiles.
G
charjohncarter
Veteran
Thanks, I had forgotten this was Leica six-bit post. As I use RAW Therapee and only post process my DNG files I guess I can use either adjustments (algorithm).
WJJ3
Well-known
Not sure why you post like you are writing scripture for a holy text, but I kinda see what you are trying to say. I think it’s fine that the lens profile leaves some of the falloff that results from using larger apertures. What’s not fine is that the profile over-corrects falloff at smaller apertures. This is why I think the lens corrections in the profiles in the M bodies are a half measure and don’t really work well.
Benjamin Marks
Veteran
LOL. He's posting that way because you tried to pick a fight. But he didn't bite. Let's go take some photos.
WJJ3
Well-known
Fighting is something you do with your partner, your family, or your fists. Since this is the internet fighting isn’t possible. Do you think I’m trolling with my comments? I’m genuinely interested in this topic.
Nice to see I wasn’t the only one who thought his posts were written that way though.
CVickery
Established
It may be a half measure, but it's the best that can be achieved without manual intervention since no M body is capable of knowing what aperture is set on the lens.
Godfrey
somewhat colored
WJJ3,
You can make whatever assignations you want or interpret my "tone" in writing any way you like. Just remember that it's all in your perception. There is neither intent for tone nor combat in what I wrote.
I'm simply trying to provide objective and correct information based on studying and using Leica digital cameras for about a decade now, and having fifty-one years of experience with using Leica equipment, and 36 years of doing digital imaging work, for photography beyond that.
If you consider all the excellent stuff that Leica has provided to be a "half measure", that tells me more about you than it does about Leica lens profiles and six bit codes, and the affect of these things on DNG files.
G
WJJ3
Well-known
Fair enough, I guess the fact that we’re making comments on a hobbyist camera gear forum means we’re all a bit out there anyway. But back to the topic, if all the highbrow explanation you made in this thread about Leica’s goal with the lens corrections in the M bodies are true, then why do I see overcompensation of falloff at smaller apertures, and lateral fringing on my M9?
My impression is that Leica might have been trying to find a solution to M bodies not knowing shooting aperture e.g. extra sensor on earlier bodies, but it doesn’t work well, so they abandoned it. So what we have are lens profiles with corrections that don’t really do a great job Of what they should do. I really don’t buy the bit that they preserve a lens’ rendering across bodies and mediums, but definitely interested in some evidence of what you are explaining.
Godfrey
somewhat colored
Just because they are trying to do something doesn't mean that they always do it perfectly. Bugs exist. The M9 is now obsoleted by almost eight years worth of advancements in later models. The SL, btw, had horrendous problems with several of the firmware v1 lens profiles, which were solved with later firmware updates.
The intent of their efforts was explained to me by a couple of engineering folks close to Leica engineering (Working with or for them on contract) when I was working and had a business relationship that included Leica. I retired four years ago, but don't feel comfortable citing names.
It is also evident in my personal tests comparing rendering qualities of the dozen or so M and R lenses I have at my disposal when images made with the same lenses fitted to M9, M-P 240, M-D 262, SL, CL, M4-2, R6.2 and Leicaflex SL bodies are compared.
No, I'm not going to put together "evidence."
I'm not trying to prove anything, just reporting what I've been told and what my own observations have led me to understand as true. I think I have plenty of experience with the equipment and a range of different lenses to draw reasonable conclusions based on what I was told, but don't vouch for it being definitive ... I learn new things all the time as different situations and different uses demonstrate/express different facets of the equipment in use.
You need a range of different Leica bodies AND lenses to be able to compare things in this effort. Having just one body and a couple of lenses is an insufficient base upon which to build the data you need.
BTW: "fringing" is usually caused by longitudinal chromatic aberration, not lateral chromatic aberration. The latter is an easy correction to make, the former is difficult without the full vector information of the capture available; fixes for fringing are usually patch up work done on the post-mosaic RGB data since today's cameras generally do not allow capture of the vector information required.
G
Godfrey
somewhat colored
Treating this part of your comment separately: The two notions in it are disjoint.
The external sensor solution to calculating the lens opening was intended to provide something in the user interface. It didn't work all that well and they've dropped the notion.
That external sensor has nothing to do with the six-bit coding and lens profiles, and resultant image corrections. The fact that the external sensor was dropped when it didn't work out for its intended purpose does not lead to any conclusion about the efficacy of the lens profiles or corrections applied to an image.
G
WJJ3
Well-known
G,
Thanks for all your explanation. From what you’ve explained it sounds like the corrections in the profiles are an even more inadequate solution than I was giving Leica credit for. A fixed amount of falloff correction applied regardless of aperture is kinda lame. And an extra sensor for guessing aperture, only so it can be included in EXIF etc., that was so ineffective they don’t even include the value anymore is also pretty lame.
In my view Leica should provide a raw converter that has a comprehensive set of lens profiles tuned to address their lenses at different apertures and focus distance. A user who cared about this stuff could take notes while they are shooting and plug those numbers into the converter in post. This would be a good solution that could compliment their ‘invariant’ approach.
Thanks for all your explanation. From what you’ve explained it sounds like the corrections in the profiles are an even more inadequate solution than I was giving Leica credit for. A fixed amount of falloff correction applied regardless of aperture is kinda lame. And an extra sensor for guessing aperture, only so it can be included in EXIF etc., that was so ineffective they don’t even include the value anymore is also pretty lame.
In my view Leica should provide a raw converter that has a comprehensive set of lens profiles tuned to address their lenses at different apertures and focus distance. A user who cared about this stuff could take notes while they are shooting and plug those numbers into the converter in post. This would be a good solution that could compliment their ‘invariant’ approach.