M10 and editing skin tones

[ggill1313]

Advertisement (gone after registration)

9 minutes ago, LBJ2 said:

See my note above, I found the same red splotch on my Sony RAW file skin tones too when pushing the Red Saturation slider to +100.

Indeed, I can make it apparent under the most extreme circumstances in my other systems. But the gradation/banding is much more natural. In truth, I'm much more concerned about the purples and magentas than reds and oranges. I expect that reds and oranges should respond in skin tones when pushing them. It's the uneven gradation in the purples and magentas which are causing the most strife. I'm compiling some shots/examples across other systems (Fuji, Canon, iPhone) in similar lighting which will illustrate the differences.

[Advertisement]

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

[LBJ2]

It is mysterious. I am not able to replicate across all caucasian skin tones in either M10 DNG, or Sony RAW. I suspect it has to do with over-exposed highlighted areas of the skin...guessing

The good news, is I don't see any of this until near +100 and all these years I've never seen the need to push anything to Red Saturation +100. But if pushing Red Saturation +100 is part of some of your styles/edits then there could be a problem with some areas of the skin that might be hot spots or near overly exposed that will not show up otherwise.

Edited January 4, 2021 by LBJ2

[Dr No]

3 minutes ago, ggill1313 said:

Absolutely, as you said, it's to illustrate effect. Though to be clear, the effect is still apparent under less extreme circumstances. The real-world application was trying to bring a bit of color back to her face. Raising the oranges didn't do it, and the areas it did affect were blotchy. Likewise the story for the reds. I had to incrementally raise all Red, Orange, Purple, and Magenta in order to even begin to lift the colors in the face. The blotchy "banding" was apparent at almost all levels of adjustment. To be clear, I was able to get the file to somewhere close to something I'd like, but it was woefully tedious and any adjustment required several other adjustments to make it appear cohesive again.

Alternatively (or perhaps more accurately) I will generally raise the warmth of the shadows by grading them separately using a shadow tone adjustment with an orange/yellow, and adjust the midtones to somewhere adjacent in color. To offset this warmth in the skin tones and ensure the skin tonality is where I'd like it to be, I use the HSL for adjustment. I tried a myriad of looks to get what I wanted. Attempting to desaturate when I was more aggressive on the color grade to warm produced banding regardless of the level of adjustment, and it still required a conspicuous adjustment in the purple slider, even though the apparent color was "orange." The uneven and harsh gradation between the Red, Orange, Purple, and Magenta is what caused the issue to be so apparent. I have used this general methodology without issue on several systems, of which two I own and shoot regularly. 

While I can't expect that my workflow will be identical as you said, I have worked on other system's files before and have never had such a vastly different experience as this. In fact, in other less demanding conditions, the sensor behaved normally and it gave no issues with a workflow similar but adjusted for the system vs. my other systems. 

I understand that many do enjoy C1 for color work, I have never had an issue with LR. Likewise I've never had an issue with HSL across all of my systems of experience. I don't mean to sound argumentative, but please do understand that this isn't my first time having worked on another system, and I'm comfortable making amendments to my workflow as need be and have done so in the past. This isn't a workflow issue. The issue here is the poor gradation in skin tones of which I've never seen exhibited in my 13 years of shooting which only seems to happen in backlit, sunny circumstances. I've given the RAW to several other professional photographers and each one has said something to the effect of, "Yeah, that's definitely weird."

I do appreciate the frustration of working with new systems and have been here before. And no, none of this is argumentative and I don't take you posts as such either.

The variables can be quite varied unfortunately. In my opinion it is less a sensor issue than a programming and profiling one, both camera and processing, de-bayerring and profiling. Creating a custom profile may help or at least smooth things out.

IR contamination may be in the mix of things but I feel less so with these tones and colours in my opinion, though the lens flare may certainly be contributing and introducing some.

When you have your camera back, as it's already been said, reset the camera; this has been known to remedy some colour issues and is acknowledged by Leica. In my opinion avoid auto-white balance and/or test with on and off. The way the camera mixes colour, and then how it get re-mixed by the raw processor, and then how you re remixing things again with your processing can create issues in my experience. Especially so in mixed lighting conditions and so much so that I leave auto white balance off.

I would also acknowledge that that finger print of lens flare is also very different between lenses, cameras and how the colour is processed. It is really quite a unique pattern in the image you've shared here. Some lenses have a greater colour shift in their flaring. It may be worth to test in a few different lighting scenarios here just as a means to trouble shoot and I would also try another lens if possible.

Given the original seems reasonably clean I would consider using curves rather HSL. Lightroom is fine as a raw developer but personally I very rarely have 100% of colour edit and grades through lightroom alone. HSL can be really quite rough and personally while it is convenient I generally find it only acceptable for drafting, and preliminary client edits and getting things to a certain point.

It sounds from your description that you could find a simple work around in the curves palette which is far more gentle in use and organic in output. Also consider a local brush and grading treatment.

 

[ggill1313]

7 minutes ago, Dr No said:

I do appreciate the frustration of working with new systems and have been here before. And no, none of this is argumentative and I don't take you posts as such either.

The variables can be quite varied unfortunately. In my opinion it is less a sensor issue than a programming and profiling one, both camera and processing, de-bayerring and profiling. Creating a custom profile may help or at least smooth things out.

IR contamination may be in the mix of things but I feel less so with these tones and colours in my opinion, though the lens flare may certainly be contributing and introducing some.

When you have your camera back, as it's already been said, reset the camera; this has been known to remedy some colour issues and is acknowledged by Leica. In my opinion avoid auto-white balance and/or test with on and off. The way the camera mixes colour, and then how it get re-mixed by the raw processor, and then how you re remixing things again with your processing can create issues in my experience. Especially so in mixed lighting conditions and so much so that I leave auto white balance off.

I would also acknowledge that that finger print of lens flare is also very different between lenses, cameras and how the colour is processed. It is really quite a unique pattern in the image you've shared here. Some lenses have a greater colour shift in their flaring. It may be worth to test in a few different lighting scenarios here just as a means to trouble shoot and I would also try another lens if possible.

Given the original seems reasonably clean I would consider using curves rather HSL. Lightroom is fine as a raw developer but personally I very rarely have 100% of colour edit and grades through lightroom alone. HSL can be really quite rough and personally while it is convenient I generally find it only acceptable for drafting, and preliminary client edits and getting things to a certain point.

It sounds from your description that you could find a simple work around in the curves palette which is far more gentle in use and organic in output. Also consider a local brush and grading treatment.

 

Please see this link with examples across other systems I had on hand (Fuji, Canon, iPhone non-respectively) in somewhat similar backlit circumstances (mostly). The first set is actually of the same young lady. Take note of the lack of influence the Purple and Magenta has on her face. Indeed, the colors do turn garish in the Reds and Oranges when boosted to 100, but these are easier to manage when the gradation is smoother and less influenced by several other colors. To that end, I've noticed this effect more on XTRANS sensors (this was shot on a Fuji X100F) than other Bayer arrays. 

The set beneath that is of my wife on a Canon 6D likewise in harsh, backlit conditions. Again, note the lack of influence the Purple and Magenta have. 

The next set is likewise on Fuji X100F. And again, backlit with little to no influence in the skin tones from Purple or Magenta. I have endless examples of these because I shoot backlit very, very often. These are just some I had on hand that I could easily identify and manipulate. 

While I do perform a certain aspect of my color grading in the curves, I do also perform a large portion of it in HSL and it has never been problematic and gives finer control over independent saturation levels that Curves cannot. Likewise I do resort to local adjustments when necessary, but attempt to get it all done with global adjustments. 

I suppose my frustration could be distilled down to the handling of skin tones in backlit circumstances as they relate to what's determined to be Purple and Magenta. I'm hoping that a few fixes described here can help remedy this issue, but in my years of shooting and editing in identical light, it's never been a problem. 

I do appreciate all of the help thus far.

https://imgur.com/gallery/tBa45WA

[Dr No]

1 minute ago, ggill1313 said:

Please see this link with examples across other systems I had on hand (Fuji, Canon, iPhone non-respectively) in somewhat similar backlit circumstances (mostly). The first set is actually of the same young lady. Take note of the lack of influence the Purple and Magenta has on her face. Indeed, the colors do turn garish in the Reds and Oranges when boosted to 100, but these are easier to manage when the gradation is smoother and less influenced by several other colors. To that end, I've noticed this effect more on XTRANS sensors (this was shot on a Fuji X100F) than other Bayer arrays. 

The set beneath that is of my wife on a Canon 6D likewise in harsh, backlit conditions. Again, note the lack of influence the Purple and Magenta have. 

The next set is likewise on Fuji X100F. And again, backlit with little to no influence in the skin tones from Purple or Magenta. I have endless examples of these because I shoot backlit very, very often. These are just some I had on hand that I could easily identify and manipulate. 

While I do perform a certain aspect of my color grading in the curves, I do also perform a large portion of it in HSL and it has never been problematic and gives finer control over independent saturation levels that Curves cannot. Likewise I do resort to local adjustments when necessary, but attempt to get it all done with global adjustments. 

I suppose my frustration could be distilled down to the handling of skin tones in backlit circumstances as they relate to what's determined to be Purple and Magenta. I'm hoping that a few fixes described here can help remedy this issue, but in my years of shooting and editing in identical light, it's never been a problem. 

I do appreciate all of the help thus far.

https://imgur.com/gallery/tBa45WA

I would say that is similar to my experience of using Canon which is quite restrained in it's palette. Canon skin tones, while they can look nice are quite thin in their range. It makes it easier to edit but as said I find it quite wanting in colour depth.

Its not that HSL doesn't give you control, it's just that the files you are working with are different and HSL is working on numbers rather than selected areas and you have little control over the numbers. It's sort of like using a lasso tool with no or little feathering. OK with some files and not with others. As said, working with different cameras requires different methods of editing.

Curves in LAB mode has far greater control and better output over saturation of colour then HSL, though of course less convenient.

[jaapv]

10 hours ago, LBJ2 said:

I live in strong/bright sunshine year round and use my M10 mostly in natural light. What should I be seeing to indicate the need for a IR/UV cut filter for the M10. I've read using such a filter will not hurt...but to date I don't think I see any indication it is needed. But am very willing to investigate. 

Like most photographers, I do chase the light and try to avoid harsh conditions as much as possible, but again what should I be looking for in my images that would indicate I need a IR/UV cut filter for the M10? 

Olive greens, purple blotches in skin tones, orange shift in tropical noonday sun, magenta cast on synthetic black fabrics. 

[jaapv]

Advertisement (gone after registration)

And coming back to that "warning" on the shop's website, even if the M8 is virtually unusable without such filters and IR contamination is far less obtrusive on other cameras, they are certainly not made exclusively for that camera; they existed long before it was introduced.