Over exposure with Cine1?

[Michael Maier]

I've been using Cine1 lately after have used Cine4 for a while. I think Cine4 is great but for low light only. For normally lit situations I find Cine1 gives a better output. But I was so far lighting with the monitor and light meter using an ASA of 200-160 instead of using zebras. Then by recommendation I started to expose using zebras instead of the light meter because I was told by an editor I was underexposing my footage, even though it looked fine for me. Then I set my zebra to 70% and started to expose using the zebras making sure I have the 70% zebras on my talent and avoiding the clipping zebras. It turns out that it looks overexposed to me. For me my footage before looked much better, specially skins. It held a nicer tone. Now it looks overexposed and quite plastic looking if you know what I mean. It almost glows.
What gives? Nothing changed in my picture profile, just my exposure method.

[Dominik Seibold]

Try to expose only with zebra2 for getting maximal exposed images without (avoidable) clipping. Also keep in mind that zebra2 doesn't indicate real clipping because it indicates only values above 100IRE, but the ex1 records up to ~107. The values above 100 are recoverable in post (in FCP for example with ProcAmp, RGB-Balance, ColorCorrection...).
"not avoidable clipping" means that your main-subject is way darker than something other in the scene. Avoidind clipping in that situation means to totally underexpose your main-subject. In those situations don't care about clipping of those very bright picture-areas (the ex1 with detail-off shows brilliant highlight-handling).
If you think the result looks overexposed (but of course technically it isn't because nothing clipped (avoidable) and cine1s contrast in light areas isn't too low) just pull down the gamma in post.
If you follow those rules you can still get underexposed images in some situations. Then you should switch to a std-gamma with manual knee for getting "ultra-compressed" images.

[Leonard Levy]

I suspect your problem has to do with the nature of Cine 1 rather than anything to do with clipping. and its the reason I have problems with all the cine gammas. They all have an unusual amount of compression ( I consider it an extreme built in knee actually) on the highlights but the compression starts way earlier than with most knees - even in the 70's as I recall. The result is that your cine gamma looks good only when your general video exposure level is lower than usual. Your editor is calling this "underexposed" .
I don't use the cine gammas for that reason, so I don't know if your editor should just get used to the lower IRE readings or if he should be adjusting them in post by raising the levels or maybe doing a sort of reverse gamma correction.

Faces are where the problem is worst because when you put faces where they normally belong ( say 70% for caucasian flesh tone) the compression just looks like hell. Its not technically over exposed at all, but it looks like it. So you either need to expose at lower levels and figure out how to deal in post, or use a normal Std gamma.
I haven't played much with the cine gammas and other people here have extensively so I'm sure they can give you better advice.
Lenny Levy

[Dominik Seibold]

Originally Posted by Leonard Levy

I suspect your problem has to do with the nature of Cine 1 rather than anything to do with clipping. and its the reason I have problems with all the cine gammas. They all have an unusual amount of compression ( I consider it an extreme built in knee actually) on the highlights but the compression starts way earlier than with most knees - even in the 70's as I recall.

cine-gammas don't have knee. Knee means (like the word "knee" implies) there's a hard change in contrast after a certain threshold-point (or in mathematical terms: the second derivation of the transfer-function contains a dirac-delta). The cine-gammas have a gradually decreasing contrast which is per definition not concerned with knee.

Originally Posted by Leonard Levy

Faces are where the problem is worst because when you put faces where they normally belong ( say 70% for caucasian flesh tone) the compression just looks like hell.

Do you think this looks like hell?
http://www.dominik.ws/cine1.png

[Leonard Levy]

Well, I'm not a video engineer so i don't know what the official definition of a "knee is". I do know that the best engineer in San Francisco has long referred to cine gammas as effectively acting like weird knee circuits that start very early. So whether thats an improper terminology or not I wouldn't want to argue.

My point is that it functions like a knee and that's what matters to me. The exposure curve starts compressing rapidly as you get near the highlights. Its different for each of the cine gammas but they all do it. If you look at a chip chart you will see what looks like a "knee" on a characteristic video or film curve. Its more gradual though, seems to start earlier and the compression gets pretty intense well before 100%. That can look good or can look bad depending on how you use it. The term comes from film not video incidentally.

By the way when I compared the HVX200 cine D gamma on a chip chart it had less of that "knee type" compression. it was a straighter flatter curve with a more typical knee at the top. I haven't worked with either that much.

I'm also not saying you can't get a good picture if you set your exposure a little lower than on a standard gamma. I don't use them myself and maybe I'll learn to like them someday.

I will stand by what I said before though, if you set your exposures the way you do for normal gamma's at around 70% its very easy for the extreme compression of a cine gamma to look like hell on a face. That sounds liek what Michael described.

It may depend on how even the light is, or whether you get some exposure into that compression, but I did a number of tests and didn't like it. That doesn't mean I couldn't have learned to work with it though. I'm knocking it if you're getting good results.

Lenny Levy

[Tom Roper]

Michael's problem is not with gamma. He's in an internal struggle over to go with his own intent that sets exposure for the overall scene, or his editor's intent which is to expose for the skin tones.

If he was shooting outdoor portraits, he would expose for the scene and use fill-in lighting, to maintain the dramatic contrasts while filling in the shadows.

I don't agree with using zebra2 to set the exposure from the top down.

And using zebra1 at 70% is the equivalent of spot metering. That's appropriate for a backlit subject. But if you did that all the time, your scene brightness will never be the same. The iris of the human eye doesn't adjust to skin tones, it responds to brightness. The camera should respond the same way.

Exposure should be set to recreate the brightness of the scene. If it's in a dark room, and you exposed for 70% zebras on the faces, you would be brightening a dark scene into something it is not.

I don't find anything wrong with a dark exposure that communicates the essence of what is supposed to be a dark scene.

I don't find anything wrong with setting exposure using averaged metering that communicates the brightness of the scene. If the talent thereafter has shadows, use fill in lighting for emphasis, if it's needed.

So I am with Michael on this one.

The only thing I would caution Michael about, is too much reliance/dependence on meters, light cards, histograms or whatever. Those tools are throwbacks to when we didn't have a preview. Now, you are looking at the scene with a monitor! Let your intent be guided by what you see.

[Leonard Levy]

After reading Tom, I realized that we are all making a number of assumptions about what the overall brightness of Michael's scene was and what he meant by the scene looking overexposed.

Sounds like Tom is assuming the face was not as bright as the overall scene so raising it to 70% overexposed the rest of the shot. I assumed that it was not under the rest of the scene.

So how about more info Michael, or a screen shot?

BTW - I would never use a meter to set video exposure. I do rely on zebras, but as Tom said they need to be used appropriately in the context of the shot.

Lenny

[Serena Steuart]

Each of the gamma curves has it's uses and it's just a question of understanding what they do. On the most general level you could make the assumption that the camera has only one curve and the PPs are merely customised versions. Certainly the various controls, such as gamma, black gamma and black level etc can manipulate one curve to approximate another. The cine gammas allows the user wider latitude, generally through compressing highlights. Cine1 compresses least, cine4 the most. Despite various understandings, these are very useful outdoors where the subject latitude can well exceed the "native" 8 bit range or, if you prefer, the range capability of the STD gammas (which are very much similar to each other).
A number of people have pointed to the Digital Praxis web site and it is well worth the visit to get an appreciation of the intention of the cine gamma curves. You might be interested also in these plots, but they require more explanation and I need to go cook dinner. But forget "knees" in cine gamma -- misuse of a tech term.
http://www.dvinfo.net/gallery/showim...mageuser=28517

[Leonard Levy]

Serena and others,

Can your recommend a good tutorial explaining how to make sense of the gamma curves like the one that you link to above. ( I think I understand them in a general way, but its hard to see how they can tell me how my image is going to respond to exposures.

I've based my limited experience on looking at the shape of a waveform both at chip charts (which is where I came up with my "knee analogy" and real outdoor images.

Lenny

[Dominik Seibold]

Originally Posted by Tom Roper

I don't find anything wrong with a dark exposure that communicates the essence of what is supposed to be a dark scene.

You should do that kind of look in the post. Doing that while acquisition by underexposing means to increase noise/throw away information.

Originally Posted by Leonard Levy

So how about more info Michael, or a screen shot?

Always the best idea.

Originally Posted by Leonard Levy

I do rely on zebras, but as Tom said they need to be used appropriately in the context of the shot.

What is appropriate use of zebras in your opinion? For me zebra2 just helps to get maximal exposed images on the sensors to get as much information as possible into the sxs-cards.

Originally Posted by Serena Steuart

http://www.dvinfo.net/gallery/showim...mageuser=28517

In that graph you can see that cine1s contrast (contrast = first derivation of the transfer-function) in bright areas is very similar to that of std1 - but with greater use of the sensors lattitude. Std1s short range comes from that virtual "makeup-gain" needed for knee (which was turned off in that test).
But why could you not meassure the graph of cine2/3 after the saw-max-brightness-point? Seeing the whole cine2-curve would be interesting.

Originally Posted by Leonard Levy

Can your recommend a good tutorial explaining how to make sense of the gamma curves like the one that you link to above. ( I think I understand them in a general way, but its hard to see how they can tell me how my image is going to respond to exposures.

You should play with the curves-tool in Photoshop.

[Michael Maier]

I think what I'm experiencing when I said "plastic looking skin" is more what Levy is saying. The skin looks like fake skin if you will. I think it's exactly the compression Levy talks about. It looks very artificial. I have been reading an article by Adam Wilt where he addresses how Cine gamma affects skin tonal range.
Also, bright objects in the scene like a white book cover glows, even though it didn't show any clipping zebras, so it was under 100 IRE. It is not a problem of overall brightness of the scene. I only use the zebras to set the skin Key light. Then I use a light meter to set the contrast ratio I'm shooting for and also to set the light on the background. So 70% zebras are only used to set the skin exposure and I make sure I don't have the clipping zebras anywhere in the frame.
Maybe I should just knock my zebras down to 60% so it won't be as bad. I think for the cine gammas skins look better when slightly underexposed.
About the standard gammas I don't like them. I think they look too video-ish for my tastes.

Unfortunately I can't post any samples. Client won't allow it.

[Serena Steuart]

Originally Posted by Dominik Seibold

But why could you not measure the graph of cine2/3 after the saw-max-brightness-point? Seeing the whole cine2-curve would be interesting.

OK. I was most interested in the Cine 1 and Cine 4 curves, so concentrated on those. Those curves were extrapolated beyond the max brightness of the SAW pattern generator, so take those regions as indicative rather than quantitative (apologies for not having time to add that when posted). However those extrapolations are generally correct, determined by the relative exposure for 100% white. I'm going to repeat using a transmission step wedge to overcome the SAW limitation.
Adding a little gamma adjustment to CINE4 (as little as +10) will roll the top of the curve to asymptote 95% white, so no matter what you cannot clip (but the angel glowing highlights are quite unsatisfactory). That is a characteristic of CINE2.
Black gamma adjusts the curve below 50% white (for CINE4), +ve bulging it out and -ve dragging it in.

The interpretation of how the image is affected by a gamma curve can be seen by taking an increment of subject brightness (horizontal axis), extending vertical lines from each end of the segment to intercept the curve, and then carrying those points horizontally across to the image brightness axis. You can see how CINE4 gives a much greater proportion to darker subject tones than to bright tones (compresses highlights relative to low lights).

[Leonard Levy]

Adam Wilt's comments about skin tones in Cine Gamma came after a conversation with me. He went back and looked at what I had seen and adjusted his comments and his use of the those gammas when you have flesh tone in the shot.
I don't want to put words in his mouth but I got the idea that you should generally underexpose with Cine gammas in a follow up conversation with Adam. He still liked them though. Underexposure has its downsides though.

Michael, even with the standard gammas I've come to set my skin tones lower in HD . That's pretty common usage these days. So I now have one of my zebras set for 65% and after a shoot the other day I lowered it to 60% but I don't know where I'll keep it. That seems to work pretty well in HD and I suspect I would always ere on the lower side in Cine 1. I don't remember the differences in the cine modes though so use your own judgement.

I don't feel I'm getting particularely videoish footage with Std gamma and a pre-set knee that kicks in around 93, but at least with "lit" footage I have a lot of lighting experience and light to my shot.
On exteriors the cine gammas seem more useful to me, but I still see the compression issue and I need to experiment more to figure out when I can use them. It is great on an exterior to be able to hold a wider range of brightness.

Re how I use zebras - I check to see if and where I'm going over 100 for general exposure and if I have faces the 65% is a double check that I'm not getting fooled by my viewfinder or monitor. It's an art not a science and the zebras or the histogram are more safety checks against teh viewfinder than ironclad standards.

I don't know about those gamma curves - I still think that looking at a chip chart on a waveform really tells you an enormous amount immediately. I'd like to see someone post those comparisons. If you look at some of the cines in that way, you'll see a nice response going up to around 70 or even earlier in the 60's ( they vary quite a bit) , but then its starts to compress more and more and it gets hard to even 100 with some.

That will allow you to capture a great deal of information but there are affects to the compression. if you are used to using the 100% as a way to judge exposure, you might never get there at all even though some things will look clearly overexposed, they've done it in the 90's.

Lenny

[Michael Maier]

Interesting post Levy. What standard gamma do you use with the 93 knee? Would you mind sharing your knee, black, gamma etc settings?
I would love to test them under my shooting conditions. Maybe they will solve the problem. Otherwsie I will just underexpose for cine1, at least skin. I don't find the rest to be as much of a problem as skin.

Thanks.

[Michael Maier]

It might really be a limitation of Cine gamma after all as technically it is not really overexposed. Nowhere in the frame has reached 100IRE according to FCP's scopes. But it sure looks overexposed for the eye, specially the skin highlights.