I first noticed that phenomenon when working with stills, and looking at the various histogram channels in Photoshop.

When I work with footage in Apple's "Color" application, the parade waveform often lets me see if I'm about to push something beyond the edge.

Whenever I'm out shooting, I generally keep an eye out for skin washouts. In fact, the EX1 has so much dynamic range that I'm often watching out for white areas and making sure they maintain detail. About the only time I let something get to IRE 100 is if it's a light source or a bright specular.

Would this be a solution :
- put a colored filter in front of the lens in order to have less red coming into the CCD
- do your WB with the filter placed in front ?

If you did that you'd shift your image toward cyan.

You might adjust the matrix a bit to reduce red saturation. But the problem is otherwise easily avoided by checking skin highlights while shooting.

Most human skin tones naturally contain a lot of red and magenta. It's not obvious to our eye but it is apparent to both film and digital cameras.

Greg,

That is exactly what I'm noticing, fake, plastic looking skin! Good description. What is troubling me is, to keep those skin tones under 70, which is where it starts to compress, the rest of the image looks slightly under exposed.

I still can't upload pictures (?), but to see this "compression effect", shoot skin tone in bright sunlight, turn the iris up and with histogram on, you'll see the skin tone region reach about 3/4 toward the right, then squish flatter before proceeding up to the far right side and 100%. This means you need to iris down to retain those highlights in this upper midtone region, thus darkening the rest of the image. I've noticed this in most all the Gammas, though the Cines more as Leonard pointed out. If you aren't shooting skin tones (mostly in bright sun), it's not as noticeable.

It's a compression that kicks in, in the upper mid-range, not the top range like the knee usually effects. Thats the best way I can describe it. Sorry if my terminology is confusing some. I was using "Clipping" because to me "Clipping" is a loss of detail regardless of IRE level, I guess everyone got confused and thought that only meant "over 100-110". I was talking about loss of detail in the skin tones, it's definately "Compressing" them so thats a better term.

At least I'm glad to see I'm not the only one experiencing this and will just have to adjust when shooting. Setting my Zebra 1 to 60 or 65 will be the first step, thanks for that tip Greg, I've always set them to 70 with most other professional cameras I've used.

Leonard, your style of bringing your whites to 100 is what I do too. You're right on, that bringing your whites to 100 can overexpose other mid-tones....so I'm wondering what GAMMA you are using? I found STD 4 and 3 do it to some effect too but haven't tried them all.

An example of the red channel clipping...
 

Here's a screen grab of FCP showing the clip and the scopes. As Dean suggested, the Red channel is indeed clipping and he is correct about skin having a lot of red component.

The white balance is correct however, as you can see most of the skin tones falling on the line in the vectorscope that represents typical caucasian skin tones.

I'm wondering what sort of Matrix setting I could use to avoid (or at least limit) the risk. How do I get a little bit less red in my final image without affecting the other colors too much ?

I noticed this immediately when I tested the cine gammas, and my solution is that I just use the standard gamma 3 and it works fine for me. Maybe I'll figure out how to use the cine's someday when I have time though, obviously lots of people are happy with them.

I showed this to Adam Wilt and last I talked to him as I recall he still liked the cine gammas but just chose to slightly underexpose to avoid the skin compression.

Its not a gigantic problem though folks, just be careful where you put your flesh tones.

I just shot a bunch of interviews in a variety of settings. Outdoor cloudy and sunny, indoor flourescent, incandescent and studio lighting. In all settings I manually white balanced correctly and in all settings the skin tone of the subject was too red even though whites were correct. I was using Cine4 PP with some minor adjustments including blacks at -6. Whattup wit dat???

Mick Haensler
Higher Ground Media

I think I'll go for a try to white balance slightly cooler than recommended (i.e. try to find a reddish card to balance on, or do it manually in the Picture Profile menu) so that in the RGB parade, blue will come up a bit and keep the reds in a reasonalbe limit. I suspect it'd be be easier to warm the whole thing up in post and have as much detail as possible than trying to get those details back in a "plastic" skin area (in any case less hair loss in my case :-))

Attached is a quick attempt I did from my window on the street below today (grey overcast day you'd more expect to WB by 7000k). I had a try at 4300k and another at 6000k. On the RGB parade (veescope live) you can easily notice the difference. Plus : at 4300k, the blue is less "glued" to the bottom of the signal, which means you also gain some details in the dark areas I think.

Did some rough "indoor" tests, and I decided to do a "real world" test tommorow with "offset white" of my Picture Profile setting at -9. I just set my zebra level at 67.
With an effort to keep the skin tones within the range of my zebra setting (67 plus or minus 5 if I'm right) and having the offset white at -9 should (I hope...) give me fairly good results.
I'll have to wait for tomoorow to shoot some faces again, but have good hopes.

Fine for you Leonard, but in the F350, there is only one standard gamma curve (well, accessible from the user menus anyway).

-gb-

Mick, I don't have experience with cine 4 , but reddish skin tones sounds pretty weird to me with this camera unless you've played with the matrix


Likewise Jean Dennis - you guys should not be that far off with standard matrixes


What are you monitoring with?

don't know anything about the 350 Greg

Results of my tests from yesterday (see attachments) :

Setting and operation mode were :
- Color Matrix based on the one by Bill Ravens ("true colors")
- zebra 1 set to 67
- White offset at -3
- expose for medium lights : try to manage that skin tones remain within the zebra 1 level
- then crush or stretch the blacks if needed
- then adjust the knee point if required

...all that to obtain a nice histogram with a bit of everything ranging from real black (but not too much, just a bit) and real white (but not too much).

So far it seems to work well for me. But it was indoor at night, under artificial light (so preety much controllable). I'd have to run another test in bright sunlight to see how to handle greater contrasts.

I'm using both the LCD and EFV, unfortunately can't afford an external monitor right now. I should clarify when I said skin tones were too red, They aren't extremely off, just enough to raise an eyebrow. I haven't viewed the footage back in the studio yet, so things might change when I view things on a studio monitor.

Mick Haensler
Higher Ground Media

I have had a chance to review some of the footage back in the studio and it looks awesome. Interestingly I haven't touched the monitor adjustments on camera at all. But the reddish skin tones I'm seeing on the LCD and EVF are not showing on my studio monitor, just a beautiful well balanced picture. Which says a lot for the EX's WB. Much better than the Z1 which I found to be almost useless. I would usually have to dial in WB and adjust in post.

Mick Haensler
Higher Ground Media