the concept of contrast explained

[Guest]

hi,

there seems to be some confusion when it comes to the concepts of contrast and compression, and how it applies to the film vs video debate and film to video transfers, so i thought i'd try to explain.

first, compression doesn't mean making a digital file smaller, in this case, nor does it mean lowering the chroma sampling rate. it means squeezing something into something smaller. the same thing is used in music recording all the time, and it's easier to explain in that context, so here goes: picture an empty concert hall. somebody drops a pin, you'll hear it. somebody bangs a huge drum, you'll hear it. now try to record a concert in this hall. your won't hear the pin at all on your home stereo, and the drum kick would destroy your speakers if played at the correct volume. so, we use compression to make the soft sounds stronger and the strong sounds softer. and the same can be done with images, by lightening dark areas and darkening bright areas, to simplify a bit.

now, contrast. the problem is that contrast isn't really defined, and that it means different things in different context, but most importantly, since all steps of the image recorsing process have their own contrast, so it's not one, but several.

here's reality (from a dark coal mine to the sun):

|----------------------------------------------------------|

here's what a particular film emulsion can register:

|--------------------------|

no matter how much darker it gets, the film still only sees black, and the lighter it gets, it still only sees white. the wider this bar is, the more latitude the film has. opening or closing the aperture on your camera moves this bar towards either side.

and here's what will end up on the film:

|---------------|

this is because the emulsion isn't completetly clear, and the grain isn't completetly opaque. the wider this one is, the better projection contrast you'll get. most print stocks and reversal films perform pretty much the same here, but negative is a lot narrower, which is why it's better for telecine, as we'll see below.

so, what happens? the photographic process takes a rather big part of the contrast ratio of reality, and compresses it to the contrast available in the emulsion. all light that fits in the second bar above will be squeezed into the third. the more compression, the lower the contrast of the film, and vice versa.

now, the video camera can register this:

|-----------|

note how the latitude is a lot lower than that of film.

and here's what ends up on the tape:

|-----------|

ah, exactly the same. no compression. and what about the rest of the light levels, like shadow detail and highlight gradients? clipped. thrown away. gone.

but wait. look. video can still capture almost all of the contrast on the film emulsion. let's build a transfer machine from film to video, and we have a way of getting the full latitude and contrast of film to fit in a video signal. great. ok, a little clipping still, but not as bad as before. and what does this mean? i can light my video to avoid blown out highlight and i won't have a problem? yes, but it's harder, and latitude also affects the colors and much more. the lower the latitude, the less color saturation you'll get for example, and the lower the latitude, the lower the mixed light performance, and so on.

have fun.

what now follows is a theoretical discussion for most of us. do not read if your budget is below several thousands of dollars... ;-)

but wait again? with the modern rank machines you can even use compresion again, to avoid clipping altogether. but in that case, with all this compression, doesn't that result in a very non contrasty image, following what was said above? yes, but there are better ways of getting the right contrast and image brightness than clipping. it's called gamma and curves. the da vinci machine connected to the rank will do this for you, giving you back all the punch you would have gotten with a video camera, but with the shadow and highlight detail pretty much intact.

that's about it. questions? ;-)

/matt

[mattias]

i hate it when i get logged out for no reason. why doesn't the auto login work?

/matt

[crimson]

here's reality (from a dark coal mine to the sun):

|----------------------------------------------------------|

here's what a particular film emulsion can register:

|--------------------------|


Actually you should move the second bar to the right a bit since reality/human eye can see much darker objects (try lighting a match and compare what you see with the fastest film stock).

For the iris you should add smaller bar. Dependant to the second bar.

Also- film itself does throw away info, but much smaller when compared to video. There are some details that current photogrpahic and video technology yet cannot see the human eyes can unless using specialised euipment.





now, the video camera can register this:

|-----------|

note how the latitude is a lot lower than that of film.

and here's what ends up on the tape:

|-----------|

Is the first bar representative of every video format?

What a video camera can see (CCD) and what it actually records on tape ARE NOT EXACTLY THE SAME, even for contrast. It must be "compressed" (as in throw away things) in the format. VHS does not have the same latitude as Beta no matter what CCD you use. So you must differentiate between CCD and format. There are video cameras that are not format dependent.

Other than that I agree with u

[mattias]

> Actually you should move the second bar to the right a bit

believe me, i tried. should have used the code tag or something i guess. too late to change now, since the board think it was a guest who made the post. :-)

to all readers: picture the bars pretty much centered under each other.

> What a video camera can see (CCD) and what it actually records on
> tape ARE NOT EXACTLY THE SAME, even for contrast.

you're probably right it's not exactly the same, but close enough in this context, and as long as the camera uses clipping to throw away information that doesn't fit, we can just as well assume it was never there at all.

> VHS does not have the same latitude as Beta no matter what CCD you use.

that depends on what you mean by latitude. vhs has lower s/n ratio, which is one way of looking at it which is good for some things, but not really applicable in this case. as far as i know they both use 10 to 100 ire (or whatever the numbers are) if you want a legal signal, so the interval between completely black and completely white (what i call "contrast") is exactly the same.

/matt

[Guest]

hey,

I think you forgot to draw the reversal/negative bars. they are not the same I think.

Paul

[Guest]

Some questions:

but wait. look. video can still capture almost all of the contrast on the film emulsion

this is what I take for granted somehow, but do not understand at all, video can capture all the contrast of film, but it can only record limited contrast when actually filming, say, a bus station which shadows one part of the passengers and the others not. It's the same ccd/format which captures the film and the bus station, so how can it record contrast/latitude differently then???

the lower the latitude, the less color saturation you'll get for example

I always thought reversal was better in saturated colours, while it's latitude
is more limited then negative....but probably I'm wrong here then.


What I also do not understand is the concept of projection contrast. Often the contrast for projection is lowered becuase it works better for the audience, but then, why bother recording it at first; what's the point of using negative (recording) film while the projection film has the same latitude as reversal..so why nor recording on reversal directly then...?????

Could yu also explain the difference between contrats and gamma, other then "gamma applies more to the midtones the gamma"???

Thanks Pofessor!

Paul

[Guest]

sorry about all the spelling mistakes..I really start believing I'm dyslectic.

Paul

[mattias]

I think you forgot to draw the reversal/negative bars. they are not the same I think.

true, reversal has less latitude (shorter first bar) and higher emulsion contrast (longer second bar).

/matt

[mattias]

but wait. look. video can still capture almost all of the contrast on the film emulsion

this is what I take for granted somehow, but do not understand at all, video can capture all the contrast of film, but it can only record limited contrast when actually filming, say, a bus station which shadows one part of the passengers and the others not. It's the same ccd/format which captures the film and the bus station, so how can it record contrast/latitude differently then???

because the film compresses it's latitude into a much narrower range. highlights won't be as bright on the film as in real life, and shadows not as dark, but they will be visible and not clipped like in video.

I always thought reversal was better in saturated colours, while it's latitude
is more limited then negative....but probably I'm wrong here then.

that's true in many cases, but not as you push the latitude limits. reversal film loses color fast when over or under expose. (even though some underexposure can increase the saturation, but that's another issue).

what's the point of using negative (recording) film while the projection film has the same latitude as reversal..so why nor recording on reversal directly then...?????

many reasons. the main reason is that you get an extra control step with nagative, since you have more dynamic range to play with when you make the print. the second main reason is that contrast builds up with each printing step, so if you start with a high contrast original, you get a *really* high contrast after a few generations.

Could yu also explain the difference between contrats and gamma?

i'll try. the gamma is the slope of the transfer function the film emulsion performs, converting light to grain. there's no true linear relationship between light level and image density, except for in a small portion of the midtones. changing the gamma changes the slope of this curve, so this linear section shifts, pushing the midtones towards darker or lighter, without changing the maximum and minimum densities, thus keeping the same dynamic range, more or less. this is what push processing is based on for example. did that make any sense? ;-)

/matt

[MovieStuff]

Well!

This is sure to get a spanking but I'll take the risk.

I agree with everything you've written as it regards negative transferred to video. However, as far as reversal is concerned, I have found that the contrast actually increases measurably when transferring reversal to video with a video camera as opposed to shooting the same scene "live" with the very same video camera.

My experience with this stems from years and years of transferring to video slides that were taken at the same time as I was shooting video for various events and documentaries. Inevitably, the original BetaSP footage would have lower contrast than the accompanying transferred slide of the very same subject, shot at the same time, and transferred with the very same video camera right off the emulsion.

While I agree that reversal (when viewed as film) can handle contrast better than video, I have never found that reversal could "compress" the contrast *enough* to overcome the natural increase in contrast one gets when doing positive to positive duplication; be it from film to film or film to video. Certainly reversal to reversal duplication (on regular stock) is iffy and duplication to video is even worse UNLESS the camera is "painted" to offset the natural increase in contrast. While this certainly works, it is not because of any attributes of the reversal but, rather, in spite of its increased contrast. But with the video camera operating normally, with no monkey business in its set up, the build up in contrast I've seen is certainly more than what the original video looked like of the very same subject shot "live" directly to tape without the reversal intermediate.

Now, negative is a different thing altogether as it is a two part process and the contrast can be managed so much better during the transfer to video. But reversal transfers always made the whites blow out and the blacks plunge, compared to the same subject shot originally on BetaSP.

Roger

[mattias]

I have found that the contrast actually increases measurably when transferring reversal to video with a video camera as opposed to shooting the same scene "live" with the very same video camera.

that's probably true. you'll still get som clipping, but my experience is that for very contrasty scenes, such as backlit exteriors, it's not nearly as severe as with a video camera. if you have a "properly lit" scene though, with nothing too hot in it, the video camera will probably perform better.

actually, here's a question: most people feel that film has to be lit, while video doesn't, which would imply that it's easier to shoot video and really hard to shoot film. i actually think it's the other way around, if you're after some kind of quality at all. sure, it's nice to have a wysiwyg viewfinder, but while most point and shoot video material looks like shit, or like news footage at best, a super-8 camera in automatic mode almost always produces great images. what do you think?

/matt

[Guest]

this is what push processing is based on for example. did that make any sense? ;-)

/matt

more or less, I think I have to get more familiar with the used terms first, but I've come to a closer understanding.
Paul

[supa_ate_sixteen]

Where can I get a wysiwyg viewfinder ? Perhaps on Ebay ?

[mattias]

i'm not sure i get your joke. in case you're actually asking a question, a what you see is what you get viewfinder is a video monitor. :-)

/matt