Crafting Emulsion for Short

[Guest cypher]

Hey guys-

 

I am about to embark on shooting a short fiction video. It consists of some exterior scenes and one main interior scene.

 

About a month ago I talked to Rob Draper asc, and he told me that I have to figure out the characteristic curve of my video camera (or emulision as he calls it) ; then find the ASA rating; then do tests to find out how far I can move in either direction on the curve with

out blacking the shadows or blowing out the highlights.

 

I have done minimal research on how to find ASA rating but there seems to be much controversy; and i left the forums more confused then when i went in.

 

Now Rob has worked with film and video cameras all his life, and his work is spectacular. I realize that his parlance is that of the film idiom, but i don't care.

I would like to understand my images the way he does. (until i hear a better argument for understanding how light is interpolated by my ccds (emulsion as Rob would put it)

 

So my problem,

I want to figure out the characteristic curve ( the -4 / normal /+2 scale) that Rob showed me, and I want to find an ASA rating(which if i'm not mistaken will give me some of the info that i need to define that c.curve).

I realize that this is one man's way of explaining this stuff, so if no one can relate i understand.

 

My question would then be this: Rob also talked about building your curve in the menus in the camera ("crafting your emulsion" as he calls it);

So i'm thinking if i can't comprehend how to find my asa and my curve soon, then I might as well just shoot some tests after tweeking some of the menu variables.

 

So which varaibles to you recomend tweeking in the menus?

I have a gydv5000u for my main shooting camera and a vx2000 for b-roll.

 

Thanks,

 

From : Nonplussed in New England

[Guest cypher]

(duplicate removed)

[Phil Rhodes]

Hi,

 

Personally I think both you and Mr. Draper have rather the wrong idea about this, although I am hardly in the position to question an ASC member. The issue with trying to characterise it like a film stock, with ISO ratings and curve estimates, is that you have practicall no room to "move" in any direction. There's no such thing as bracketing exposures in video; the dynamic range is sufficiently limited that it's either right or it's not, and if you want it to look the way we all expect drama to look, it has to be right. Working with film allows you to place your image at different points on the non-linear response that the film has to light; there is simply no opportunity to do this in video - you pick what level of hilight clipping you're happy with, and shoot it. Now the response of CCDs to light isn't precisely linear, but that's generally a fairly minor concern. So, trying to directly apply film metrics to video is not something I generally encourage people to try and do. When you're looking at a monitor, you've got four hundred thousand spot and colour meters looking at every of the frame. One of the principal and oft-overlooked advantages of video is that you can see what you're getting instantly - use that advantage.

 

Yes, you can play with in-camera processing, and it's certainly somethnig you should look at. The exact functionality varies from camera to camea, but it's possible to bend the colorimitery of the camera around enough that the output actually is very nonlinear. In concert with optical filtering, which should on no account be overlooked, this can be a powerful way to make video look less flat and nasty. It's possible to emulate most of the special processing tricks film labs can do, as well as some they can't do very well, like increase saturation.

 

All that said, I think that shooting stuff on video is even more a capture-to-tweak-later scenario than the average film shoot. You end up becoming so obsessed about holding the hillights and making the absolute best use of your limited dynamic range and things like compressed recording formats that little issues like shot to shot continuity become less important. This is generally fine and OK so long as you know you will have the opportunity to grade later, and the availability of suitable tools in low-cost software like Final Cut makes this even easier. Video needs postproduction grading to look anything, even more than film. With film you have a kind of built-in look - the grain, the crushed blacks, the odd response to things like fluorescent lights give you a stylishness right out of the box (which I tend to think things like the low-contrast vision 2 stocks are actually removing, gradually, but that's opinion.) Video doesn't come with any kind of inbuilt pretty filter; it tends to give you a blow by blow account of what you put in front of it, which is almost never right. So: get the scene on the CCDs to within the range of the camera, as evidenced by zebra stripes, careful picture and waveform monitoring, and plan to make it right on your desktop computer.

 

Phil

[Greg Gross]

Phil,

Thank you and well said sir. I've been doing exactly what you talked

about(won't go into detailed description). I have'nt been disappointed

yet! Presently working on a film I shot of a man performing Ti Chi on

a beach,its on my hard drive and I have just started editing. I've been

trying to figure out how to say what you just said,for a long time. I shot

the film while I was on location during some free time. I got the courage

to lay my meter down and I used the "eye of the photographer"and the

camera's brains(manual,1/60) to shoot the film(dv).

 

Greg

[Felipe Perez-Burchard]

hey,

 

Someone will correct me if I'm wrong, but here's how I've found the ASA of a video camera... you need a handheld meter, a grey card and a waveform monitor (or adjustable zebra bars on the camera).

 

point the camera at the grey card... light should be hitting it full on, but make sure it is not at an angle that will produce glare. Zoom in to fill the frame with the grey card. Adjust the exposure in the camera so that the waveform reads 55 IRE or the zebra bars are gone (correct exposure, should be 55 IRE becasue it's middle grey); take note of the f/stop.

Now take your handheld meter and read the incident light with the meter dome on "flat" mode; remember incident readings are telling you the exposure for middle grey (which is what you are reading), now adjust the ASA on the meter until the f/stop matches what's on your camera, this will be your video camera's film speed.

To double check, now take an reflected (spot) meter reading with your handheld meter with the New ASA set. The exposure should be the same, after all the zebra bars or camera meter is reading reflected light.

 

Now that you have this information, to figure out the curve, just shoot the same setup (with a person's face some black and some white information, and the grey card if you want) at normal, and then overexpose by 1/2 stops and then underexpose. It would be useful if you have the waveform here becasue it will clearly tell you when something has "clipped" on either end, but you can also just look at it and you'll find out where things go white with no information and how many stops it takes to put a face there for example.

 

I'm not familiar with the specific settings of your camera, but you can probably adjust the black level and the highlights and this obviously affects the results. I would adjust the settings until I am achieving what I want, and then do the exposure test.

 

Dont forgett to repeat the ASA test with the different gain settings if you plan to use those.

 

hope this helps.

 

-felipe.

[David Mullen ASC]

First of all, you have to understand the video normally does not plot exposure on a curve but a straight line -- i.e. an each increase in exposure produces an equal increase in charge (in film terms, density.)

 

Film, on the other hand, has a lower-contrast (flatter) response in extreme dark and bright areas (the "toe" and the "shoulder"), which is why the exposure plots out on a curve with a central straight-line portion for midtones (just like video) but a less steep climb in the bright and dark areas at the beginning and end of the curve, creating a rising "s" curve.

 

Video can be forced to have a curve by lowering the gamma (contrast) of the shadow region (called "black gamma" or "black stretch" on some cameras) and in the bright areas reaching the clip point (point where detail is lost in overexposing), called the "knee". However, there is no free lunch in anything and you do get an increase in noise when compressing detail in the shadows and bright areas.

 

To truly "plot the curve" you'd need something like an 11-step chip chart and a waveform monitor. While that would be highly informative, it should be enough to know how much extra detail using black gamma and knee compression will give you and how much noise results and what's an acceptable amount. This could be done by simply overexposing a grey scale with no knee compression so that the upper (lighter) bands are clipped (detailess) and then adding knee compression and seeing if you gain a whole stop worth of detail or not. Same with the black areas.

 

After you find an acceptable amount of black gamma and knee compression, it would be simple to figure out how bright or dark an 18% gray card would be before it lost detail and thus determine a range that you can correlate with a spot meter. However, since with video you can SEE when detail is lost at the high or low end just by looking at the monitor, I'm not sure how often you will pulling out a meter except in pre-lighting a set or scouting a location. It's a lot easier to do this stuff by eye with video. For example, if a daylight window is clipping, you can put ND.6 gel on it and then adjust the interior light level until you like the balance by eye, looking at the monitor.

 

You can also use the zebras in your viewfinder as a guide.

 

As for determining ASA range, understanding that unflitered internally, the camera is basically tungsten-balanced, you can do this with a gray scale or card and the zebras or a waveform monitor. For example, the FotoKem gray scale has an 18% grey card (Zone 5) on one side, but on the other side, it has a gray field one Zone lighter (Zone 6) to correspond to caucasian skintone brightness. It turns out that Zone 6 is very close to 70 IRE. So if you set your viewfinder's zebras to appear at 70 IRE and 100 IRE and adjust the f-stop on the camera, at 0 db, until zebras appear in the light gray band and the white band, you can set the same shutter speed as the video camera is using to your meter, and find an ASA value that gets you the same f-stop you set. It's close enough to be practical. I wouldn't use a meter to set the final exposure on the lens though, just for setting light levels.

 

Not that for a 60i NTSC video camera, shutter "off" or the default is often 1/60th of a second. Noting the gain setting and the shutter speed is essential in determining the ASA value.

 

I find the contrast of video cameas in the shadow region to be fine without playing with the black gamma. In fact, on my last HD feature, "Dot", I turned off the Black Gamma on the Sony F900 and I set the Master Video Black level to -1, just to clean up the noise in the shadows and keep some contrast to the image. I did use Auto Knee compression (DCC) now and then, but turned it off whenever I noticed something weird it was doing. For example, I shot some scenes looking into windows whited-out by putting a white UltraBounce outside and turning off the DCC helped make sure that the white clipped rather than show detail in the white fabric outside.

[Michael Nash]

Cypher, please don't post the same question in more than one forum. That just makes it hard for people to respond in a single thread.

 

Regarding Phil's comments on the general discussion thread; the non-linearity of the video sensor is PRECICELY the issue that Mr. Draper is addressing when he talks about "crafting an emulsion." Finding the characteristic curve of the camera is the first thing I do when I pick up a new video camera.

 

Look at it this way: Without testing a camera you can look at the monitor and see what you get; but if you know the response curve you can look at a scene and anticipate what you'll get, and take measures to make the shot come out how you want. And taken a step further, if you can modify the characteristic curve of the camera you can then control what you get, putting you several steps ahead of the game.

 

So the first thing I do is find the sensitivity (effective ASA) of the camera, so I at least know what kind of f-stop I'll get in a variety of lighting situations. Then next thing I'll do is find the characteristic curve, or how far into the shadows and highlights the camera will read. This is not for the purposes of "bracketing" the expsoures, but instead to simply know what shadow levels and highlight levels the camera will be able to see, relative to my exposure.

 

I know from experience that -4 stops at about 10 IRE lopks about "normal," or simlilar to what your eye sees. If the camera can't see that far down into the shadows, that simply tells you the contrast of the camera will render shadows slightly darker than what you see by eye. This is important because you can't always be a slave to the image on the monitor -- sometimes you just have to know what you'll get and trust your own experience.

 

For highlights, I know that most video cameras will only give me about 1-1/2 to 2 stops of overexposure (with the knee circuit on) before it clips to complete white.

 

The basic adjustments for the characteristic curves are BLACK LEVEL (PEDESTAL OR MASTER BLACK), which controls the level of the darkest black; BLACK STRETCH (BLACK GAMMA), which adjusts the brightness of the shadow region; MASTER GAMMA, which controls the brightness of the midtones relative to exposure, and also affects contrast; and the KNEE to control the highlights.

 

Some cameras may not allow control over all these things, and higher-end cameras like the F-900 will allow even more control within each of those areas.