Skin tones on the af100

I call it like I see it. The overexposure didn't help - I agree wholeheartedly, but even the edges of the subject are hard. ND definitely would have improved the shot - but wouldn't change what is clearly a characteristic of the camera.

We can go on all day about what the camera is capable of in ideal situations, but I'm very interested in the limitations of the camera in situations that are not fully controlled. If I have to run ENG, I don't have time to go ND a window, I need the shot now - and whatever detail, sharpness, knee, etc. settings that I have on - well, I don't have time to go fidgeting with them to improve the shot.

I'm aware this thread was started to discuss skin tones - but you put the frame out there, so I noted what I saw.
 
Taylor Rudd said:
What gamma are you guys using?

I like what I see, editman.
Click to expand...

I have a thread in the "Screen Grabs" section called "M4W Screen Grabs" if anyone wants to check out the skin tones I got on that shoot. I was using CineV gamma with the Norm2 matrix. Sorry I'm on my phone so I couldn't post a link.
 
oh, i think i also had some sharpening going on in post, so again, don't judge my still for anything else but the skin tone. Thanks.
 
Here is a 100% crop from a frame I shot at ASA800 in my shop.

SkinTones.png


This is natural light coming in the windows, no fill or bounce, Preset 56K.

No grading at all Canon 17-40L set to 30mm f4 I think. I am very happy with this, I can easily grade it to any number of different looks.

I think this camera is great. You do have to dial it in though. I am now thinking of dropping the Chroma to 0 or maybe -1.

The settings were:
Film Cam
Detail Level -4
V Detail Level -4
Detail Coring -3
Chroma Level +2
Master Pedestal -2
Gamma HD NORM
Knee LOW
Matrix NORM2
 
Even the highlight on the face has lots of detail here it is at 400%.

Bear in mind the reflected light is off the snow outside, none of the values are above 160 out of 255 as seen in PSD.

Skin400.png


The color here is exactly what Sharyn's face looks like in this room. Bear in mind a 50 something female w/o makeup is very real.

BTW this is from a NanoFlash video file shot at 100mb/s Long GOP it holds up very well.
 
Last edited:
HD NORM is designed to present images that comply perfectly with the Rec 709 HD standard. Using HD NORM and NORM1 matrix will result in the most accurate color reproduction and proper HD gamma.

The other gammas and matrices are for looks and preference. The reason I used HD NORM in the ISO test was because it most accurately reproduces the chart. Look at the chart on the monitor, and then in real life, and they look exactly the same. It wasn't because I was trying to dial in the "best" look, it's because I was going with the most accurate look.

HD NORM is accurate, but it also does have the typical video boost in the midtones.
 
Well, as always, there isn't really a 'best', there's "best for each specific circumstance." If you intend on doing post-processing on the footage, cine-D is designed to give a flat, linear slope, which is more like what negative film would deliver. It doesn't necessarily look the best on TV, but it is the flattest and most linear, therefore it's the most technically accurate as a place to start. Cine-V is similar, but with an eye towards making it look better on a television. It has the typical video gamma bump through the midtones, and a lot steeper curve at the top and bottom. HD NORM is designed for direct video shooting, making pictures that look appropriate for HDTV shooting. Put the camera in 1080/60i or 720/60p and HD NORM gamma and you'd have a really accurate live video look.
 
"We can go on all day about what the camera is capable of in ideal situations, but I'm very interested in the limitations of the camera in situations that are not fully controlled. If I have to run ENG, I don't have time to go ND a window, I need the shot now - and whatever detail, sharpness, knee, etc. settings that I have on - well, I don't have time to go fidgeting with them to improve the shot."

thank you viddovation, that does explain it better than i did.

what i have learned is that i need to bring my 17" panasonic monitor on the field to set up a look for each challenging lighting situation. it looks great indoors under controlled lighting, but i need to find a good outdoor look that i can have as a go to with the push of one or two buttons and get results that i can live with.

i have a hvx170, and that camera looks great almost anywhere, but has that little sensor that makes everything super sharp. i just wish the af100 would have the colors of the 170. maybe it can, but i need to find the setting that does that.

I have an Alexa, i know this af100 is no alexa (it costs 16 times less) and i don't expect it to have 13+stops of latitude, i just need to be able to trust it that it won't bite me in the a_s.
 
Last edited:
lady-orig.jpg lady-color.jpg
Rough adjustment in Gimp of the screen grab to a stronger S-curve.

Most video cameras do not put a strong-enough curve no matter the setting, so you might have to do it in post. To prepare for doing S-curves in post, you need a lot of highlight headroom. In other words, when shooting, underexpose. Otherwise, the S-curve will make it overexposed.

graph.jpg

Another way to think of the S-curve is:
1. Low-contrast shadows
2. High-contrast mids
3. Low-contrast highlights

This is what chemical film does. The crystals start by responding slowly to light (low contrast lows) then speed up (higher contrast mids) then put on the brakes as they hit their limit (lower contrast highs). If you want the film look, you should apply an S curve.

"But that's not realistic," you might say. "I want normal-contrast lows, normal-contrast mids, and normal-contrast highs. No need to chase this past film look. Time to move on, to a more accurate representation of real life." But what you don't realize is that is impossible, even undesirable.

To see why, think about the very beginning and end of the image-making process.
1. At the beginning, going into the camera is reality. Reality often has a wide range of brightness --- like 11 stops.
2. At the end is your TV screen. Your screen has a lower range of brightness --- like 4 or 5 stops.

As you can see you need to cram 11 stops into 5, and at this point you may say that the solution is that manufacturers make screens with a wider range of brightness.
To do that, they would either have to make the screens have darker darks or brighter highlights, or both. They already have been making them with darker darks. That's how they reach those amazing million-to-one contrast ratios. They aren't brighter so much as the screen, when black, is reeaally black (go ahead, look at your TV screen when it's off. Is it perfectly black or dark gray?). But this only goes so far, and what they really need to do to make an 11- or 12-f-stop screen is to jack up the brightness potential. But that's not really what you want. If a TV was as bright as real life, then watching TV would be just like looking out your window. And a scene of something outside on a sunny day would be just like looking out your window on a sunny day. This may cause sunburn. In any case, it would be uncomfortable for the viewer, especially for most shows where they cut from day to night, inside to outside.

I wax philosophic here, but you might say that this lower-contrast TV is better than reality-contrast TV. We really don't want reality, at least not superficial reality. If the ultimate goal was "reality," then we would not go to movies. Because all I have to do is walk around, and I have high-definition, 360-degree, 3-D picture, with surround sound, and perfect acting. But art is not about replicating reality but maybe distilling it. You don't see many movies with scenes of the character going to the bathroom, even though characters no doubt need to go to the bathroom at least once or twice a day. We cut out those parts --- as well as a whole lot of other things --- because they are not important to the story. The story is about discovering a certain thing. For example, courage in the face of a city-wide terrorist attack. We want to see what that looks like. This film is a study of that one subject. So we pare away all the extraneous things (characters going to the bathroom, characters getting their tongues tied like we all do, the waiter getting the order wrong) because those are just distractions in this particular movie. In another movie, those things might be what the movie is about, like in a comedy.

So back to our problem, how do you cram 11 stops into 5? Well, there are at least 3 ways:
1. Cut off the top 6 stops. This is what we call highlight clipping. And it's ugly.
2. Cut off the bottom 6 stops. This is called crushed blacks, and for some reason to me it is less ugly, but the picture is veeery dark.
3. Apply an S-curve. This preserves the mid-tones while sort of washing out the lows and highs. Actually, the mids are often even higher-contrast than reality, but there is a reason to this: people's faces fall into the mids. What are we mostly paying attention to, both in real life and movies? People's faces. So we bring them out.

In summary, you might say the S-curve is unrealistic, or an artificial preference from watching film for years and years. But I say that it may, serendipitously, be the ideal solution, even when we have other options.
 
Last edited:
No, you just have to properly expose. Part of my testing with the camera was how to treat exposure and skin tone with this cam to avoid the yellow. Part of it is making sure my white balance is smack on as well ( but even here it's slightly off ) Then for me the key part is applying my unsharp mask to my lower detailed raw which also clips values above 100 IRE.

Then applying a little desaturation to the highlights helps, or maybe a 3 way color corrector to adjust the color of the highlights with a bit of desaturation, and then a nice S curve helps the transition quite a bit.
 
combatentropy said:
View attachment 29167

Rough adjustment in Gimp of the screen grabs to a stronger S-curve.

Most video cameras do not put a strong-enough curve no matter the setting, so you might have to do it in post. To prepare for doing S-curves in post, you need a lot of highlight headroom. In other words, when shooting, underexpose. Otherwise, the S-curve will make it overexposed.

View attachment 29166

Another way to think of the S-curve is:
1. Low-contrast shadows
2. High-contrast mids
3. Low-contrast highlights

This is what chemical film does. The crystals start by responding slowly to light (low contrast lows) then speed up (higher contrast mids) then put on the brakes as they hit their limit (lower contrast highs). If you want the film look, you should apply an S curve.

"But that's not realistic," you might say. "I want normal-contrast lows, normal-contrast mids, and normal-contrast highs. No need to chase this past film look. Time to move on, to a more accurate representation of real life." But what you don't realize is that is impossible, even undesirable.

To see why, think about the very beginning and end of the image-making process.
1. At the beginning, going into the camera is reality. Reality often has a wide range of brightness --- like 11 stops.
2. At the end is your TV screen. Your screen has a lower range of brightness --- like 4 or 5 stops.

As you can see you need to cram 11 stops into 5, and at this point you may say that the solution is that manufacturers make screens with a wider range of brightness.
To do that, they would either have to make the screens have darker darks or brighter highlights, or both. They already have been making them with darker darks. That's how they reach those amazing million-to-one contrast ratios. They aren't brighter so much as the screen, when black, is reeaally black (go ahead, look at your TV screen when it's off. Is it perfectly black or dark gray?). But this only goes so far, and what they really need to do to make an 11- or 12-f-stop screen is to jack up the brightness potential. But that's not really what you want. If a TV was as bright as real life, then watching TV would be just like looking out your window. And a scene of something outside on a sunny day would be just like looking out your window on a sunny day. This may cause sunburn. In any case, it would be uncomfortable for the viewer, especially for most shows where they cut from day to night, inside to outside.

I wax philosophic here, but you might say that this lower-contrast TV is better than reality-contrast TV. We really don't want reality, at least not superficial reality. If the ultimate goal was "reality," then we would not go to movies. Because all I have to do is walk around, and I have high-definition, 360-degree, 3-D picture, with surround sound, and perfect acting. But art is not about replicating reality but maybe distilling it. You don't see many movies with scenes of the character going to the bathroom, even though characters no doubt need to go to the bathroom at least once or twice a day. We cut out those parts --- as well as a whole lot of other things --- because they are not important to the story. The story is about discovering a certain thing. For example, courage in the face of a city-wide terrorist attack. We want to see what that looks like. This film is a study of that one subject. So we pare away all the extraneous things (characters going to the bathroom, characters getting their tongues tied like we all do, the waiter getting the order wrong) because those are just distractions in this particular movie. In another movie, those things might be what the movie is about, like a comedy.

So back to our problem, how do you cram 11 stops into 5? Well, there are at least 3 ways:
1. Cut off the top 6 stops. This is what we call highlight clipping. And it's ugly.
2. Cut off the bottom 6 stops. This is called crushed blacks, and for some reason to me it is less ugly, but the picture is veeery dark.
3. Apply an S-curve. This preserves the mid-tones while sort of washing out the lows and highs. Actually, the mids are often even higher-contrast than reality, but there is a reason to this: people's faces fall into the mids. What we mostly watching, both in real life and movies? People's faces. So we bring them out.

In summary, you might say the S-curve is unrealistic, or an artificial preference from watching film for years and years. But I say that it may, serendipitously, be the ideal solution, even with other options.
Click to expand...

I applied an S curve in the shot i posted. Basically , in camera, I always shoot a little flatter knowing very well that contrast will be punched in post, so i really have to go by lighting ratios to know what to expect in post.
 
Here is the oringal with a graded. Notice the reflection and highlights on the sink faucet, they start off blooming slightly purple and then it's fixed!

dancompare.jpg


All i know is that this test made me really really happy and excited to start my new series. I know my settings:

Detail: -5
VDetail: -5
Coring: +1
Chroma: +1
DRS: 1 (i always keep it on, have yet to see a situation in a controlled lighting environment that it hasn't help me produce a "better" shot)
Gamme: CineD
Matrix: Norm1
 
Last edited:
One thing to bear in mind here is I am looking at a HDSDI JVC DT series calibrated monitor through an AJA Kona card, when grading. The way it looks on a computer screen is fairly irrelevant. I also keep an eye on the scopes.

I was just trying to make the point that the AF100 can capture very accurate skin tones, and you can then grade them any way you want. Like the cool wintry slightly purple look in my graded image.
 
agree completely, once you learn how to light the cam, stuff looks great. And im pleased to know that i've found my settings.
 
PaPa said:
I know my settings:

Detail: -5
VDetail: -5
Coring: +1
Chroma: +1
DRS: 1 (i always keep it on, have yet to see a situation in a controlled lighting environment that it hasn't help me produce a "better" shot)
Gamme: CineD
Matrix: Norm1
Click to expand...

If you use DRS, you are no longer shooting in Cine D or any other of the available gammas, DRS will apply its own curve to highlights, while dynamically changing mid-blacks, at least that's the way it is in the P2 Varicams and HPX3000, DRS automatically applies Film-Rec gamma for highlights.
 
You must log in or register to reply here.
Back
Top