ive been trying to figure out what the differences are in film and video, you should all know the production qualities. im going to first assume you know how to do properly lighting and staging and camera work, and that you know how to use your camera as well as DOF and the motion issues (frame rate, motion blur, shutter speed, etc).

secondly, im going to assume you know that the contrast lattitude of digital video on average is about 4 stops versus around 9 for film.

so.....after much experimenting and deep thought and study of film frames and digital frames i have found the closest thing to an answer.

CURVES.

the article posted on this site has a very vague and in my opinion incorrect description of how to use curves (in after effects) when describing the creation of a 'film' look, they might as well have just increased the contrast.

i have determined that you can give digital video something equivalent to a 16 or super16 quality look in post, at least in terms of latitude....digital is its own animal, not film for sure, but not without hope.

by increasing (or extending) the APPEARANCE of more values in the highlights and more values in the low mids and less in the darks you can fool the human eye into seeing more lattitude then is there.
it seems silly and its difficult to explain and must be tailored to each scene and a desired look. but the key is to use curves and leave the mid tones alone, slope the highlights downward (towards grey) subtly and the darks towards black.

also, this only works well if you dont have large areas of blow-out or deep shadows, a neutral image is preffered as it gives you creative freedom with your images look.

i am working on a guide with specific examples for a film look, once its done ill post a link to the guide.

of course color grading and color correction are also critical to the process as well....but thats just because modern film uses it all over the place....so im assuming you want a contemporary look as well.

Hmmm, so your basically saying to DECRASE the contrast? It doesn't look right.

If you have a well composed shot, you shouldn't have to deal with any dynamic range problems.

I did what you described (basically decreased the contrast) and this is what I got, along with "my method" and the original. It's a framegrab from some video I shot yesterday doing tests for that extra extra wide cinema look.

http://img.villagephotos.com/p/2003-7/252285/Image475.jpg

Right, decreasing the contrast brings out more detail, but it also gives you a ugly grey overcast. I'd rather loose some detail and get nice colors. :D

yah, so thats not quite it.....it'll be clear what i mean when i post some pics (of the curve), the key is only lowering the contrast at the very peak of the brights and subtle slopes....your method is similar to what i get and quite good as well....as far as i can tell at the small size....if your method works you dont need to listen to me anyway, heh.

....heres my sample using my method on the middle picture (untouched).

http://www.letterstoyou.com/sample.jpg

of course what i said about highlights doesnt really apply here because the picture is so small that the detail isnt really there to tweak, but you can see how all the other levels are adjusted as well.

Joe,

You're a genius and don't know it. What you've described is a relatively recent method of High Dynamic Range (HDR) imaging used in computer graphics and, recently, some computer games. Essentially, they're taking the true range of available light and moving it into the range of the viewed medium.

For example, a scene with bright sun and deep shadow in a range from 1 to 100 to be displayed on a medium with a range of 1 to 10. Calculate what percentage each brightness value is and fit it into the smaller range. This way very bright and very dark areas look the same though you have adjusted their range.

This is the same thing someone in a darkroom does with a film negative. The phrase is "expose for the shadow, develop for the highlights". This way the photographer is sure to capture all the image information on the negative. He can then pull detail out of areas by dodging and burning.

Film though is an analog medium. You are correct about curves. Everything on film is recorded in an analog curve of light. Digital is not a smooth curve due to the digital recording method. While this gives digital its "sharper looking" appearance, the reality is it loses information between color and light levels that film captures. An edge between a solid black and solid white area may be sharper than film because the digital camera can't record all the light between the two ranges. But that is also what causes stairstepping.

<< This is the same thing someone in a darkroom does with a film negative. The phrase is "expose for the shadow, develop for the highlights". >>

Yup, and the trick for good looking video is "expose for the highlights" (similar to the practice one would follow for shooting reversal film). I routinely underexpose exteriors by anywhere from 1 to 2.5 stops if they feature bright sunlight and contrast. You can always expand the range back in post. The important thing is to capture detail in the highlights; what your camera tells you is the correct exposure usually results in washed-out faces and blotchy skies.

Adjusting the gamma curve has long been a core element in filmlook emulation.

Charles- I have seen photographers using a light meter, and something similar used in film. What do you use?

From a layman's (newbie) perspective - if Charles underexposed exterior shots (making sure highlighted areas have detail), then could I apply that to the example below?

If I am shooting a stage play (which I frequently do) where faces are sometimes washed out by bright lights, I should shoot manual mode, slightly UNDER-exposed so that I can see the detail in their faces. Then in post, tweak the clip so that the appropriate color/contrast is shown. And theoretically I'll have a finished video with a good, sharp image with good detail in their faces.

Did I explain that right?

Thanks in advance. This is very interesting for me!

Keith:

For film work I do use meters: a Spectra incident meter and a Minolta spot meter.

However, for video work I usually light to the monitor and a waveform monitor if available. The exception would be if I was pre-lighting a set without the camera present, in which case I would use a meter to make sure the light level is correct.

Mark:

That sounds like a good plan. A while back I shot a play that a close friend had written and produced, probably the first time in 10 years or so I'd done that sort of shooting. I had my XL1 with the manual lens, and we had another camera with a GL1. I had hoped to work with the tech crew to balance out the light levels for the video, but it wasn't going to happen, and there were several scenes that were very high contrast (spotlights and pools of light). During the performance I was constantly riding the manual iris to net the best exposure, which as you indicated was slightly under-exposed depending on the scene. In those high contrast scenes I went tight on the characters within the hot areas and rode the stop down to an appropriate level. The other shooter went with auto iris.

The result was that his camera was constantly too hot, and the actors in the spotlights or hot pools were blown out, sometimes beyond recognition. On my camera, they rendered nicely. During the editing phase, the producer constantly urged me to try to bring back some detail into the blown-out faces from the GL1, and of course I had to tell him that it wasn't possible.

Sometimes I ended up with a slightly dark image overall from the XL1, with the actors out of the spotlight coming in a bit muddy, but using the color correction in FCP3 I was always able to punch up the midtones to a satisfactory level.

Hope this answers the question!

By the way, the underexposure technique for exteriors was well used in "28 Days Later" (one of the articles I read described this process exactly).

joe, please do post some specific techniques to use in After Effects when you get a chance. I like the results you are getting. I find that the tutorial on this site results in footage that is overly saturated. It's definitley an improvment, but I'd like to try out your method and I'm a bit confused on which curves to manipulate in which way. When you get a chance, I'm sure we'd all appreciate you writing out a mini-tutorial with some screenshots of those color curves. thanks!

Thanks for that info, Charles. I shoot all of the stage plays for the City Arts and the Playwright's Forum I belong to and a good percentage of the finished video looks good (GL1), but when actors move around on the stage (ie: separate to either end) and I have to have a very wide shot, then that's when their faces are blown out.

If I zoom in tight, I can get good detail in the brightest conditions, but then the other 95% of the stage is unseen - which is not an optimal condition shooting theatre!

I'm going to try the manual way on AUG 9, as I'm shooting a doc that culminates in three short plays in the evening (three playwrights, three directors, one music composer and twelve actors write, rehearse and perform three short plays, beginning at 10:00 a.m. with live performances at 8:00 p.m.).

One question though - I don't suppose a 2x or 4x Nuetral Density filter would do the same thing would it? Or would it just under-expose everything?

Thanks again.

In another thread somewhere I came across this tutorial which seems to cover similar ground

http://www.dvinfo.net/articles/filmlook/broadway2.php

and shows curves etc -- it's all about replicating the density curve of film isn't it?

Scot

that is the tutorial listed on this site that i find does not properly explain or (necesarily) understand the use of curves to as you said properly emulate the curve density of film

since this is the big example ive made a few examples myself of different looks using different, curves, since the look you want will affect how you adjust the curves and then the color grading all are unique. i should have my guide finished soon and then ill put it up, with specific examples on how to understand the original picture, how to make that emulate the look of film properly and then how to distort the curves further to create your own look that is still film-like. the main thing with curves is that the curves have to slope subtly to be believable.

the top left is the original and the top right is the post they made (too saturated for my taste but otherwise good) and then 4 different looks i created below using curves and then color balance.

http://www.letterstoyou.com/curves00.jpg

Mark, to answer your question (sorry Joe, we seem to have two different threads going at once in here), ND filters won't help in this situation as you guessed. The problem is too much contrast within typical stage lighting which can really only be addressed by altering the lighting itself (or the gamma curve of the camera--to make this more relative to the subject of the thread!!--which can't be done with the GL2).

By sloping the colour curves, that's almost the same as adjusting the contrast from what I see.

Why not just adjust the contrast then?

Joe, slightly off topic but I would like to add that there is a very good, lengthy film look article on http://www.dv.com

<<<-- Originally posted by Michael Chen : By sloping the colour curves, that's almost the same as adjusting the contrast from what I see.

Why not just adjust the contrast then? -->>>

because it is not the same as contrast, contrast increases the difference between the brights and the darks. thats not what im doing. as the brights reach the high end you are reducing the contrast and values in the mid range you usually leave alone (depending on the scene and the mood wanted) and only the darks do you really increase contrast on, and its only the lower part, near the lower third you actually lower the contrast a little to bring out shadow detail.

Nice examples Joe - out of the four you posted the one in the top left looked the nicest I thought - it had the presence of the redone one form the tutorial but seemed more natural than that one.

I'm very much looking foreward to your tutorial - I'm going to be shooting a feature later in the year and am facing the film vs video question at the moment -- it's a period piece so definitely more suited to the film look - but we've only got a small budget -- it's 16mm vs HD really.

Scot

scott, your in luck, ive made another break through!
...and now i can complete my tutorial and explain (i think) in clearly enough that it would be useful to others.

now all i need is a good range of example shots (from different cameras if possible) to use for my tutorial. i have a few but if any1 could send me screen grabs from stuff theyve shot with their camera (decently setup shots please) itd be a great help to completing this tutorial with proper examples. the more variations in cameras i have the better to demonstrate this technique.

if any1 can send me shots please email me (jpruss2001@hotmail.com) or post them here.

thanks. tutorial will be up by the end of the weekend...promise =]

Joe, you might be interested in checking out a little article I did a while ago with Vegas 4.0.

http://www.freewebs.com/vegas4/vegas4.htm

I've refined it very much now, and might do a new version. :D

Hi Joe

I shot a short on 16mm while a go and because it was the first time we had used that camera (a Beaulieau R16) we also re did every shot with my JVC one-chip miniDV camcorder - so we'd at least have something if the R16 didn't work.

Anyway the R16 worked fine but I have all that video shots of well set up and lit scenes. Some of which are on the web page for the film, though most there are just digital stills I also took with that miniDV camera. I haven't updated the page for a while and haven't even posted any of the telecined 16mm shots there yet - but will soon!

Feel free to use any of these shots for experimentation - if it makes any difference over whether they're digital stills or stills grabbed from the miniDV tape I think the last two are stills that were grabbed (and I'm pretty sure I de-interlaced them), all the rest are the digital stills taken onto a flash card.

If you need more off the tape I can send them to you.

The web site is:
http://www.mango-a-gogo.com/couch/couch.htm


Alex,

great tutorial - thanks for going to all that trouble - I'm still working throught it :-)

Scot

"An edge between a solid black and solid white area may be sharper than film because the digital camera can't record all the light between the two ranges. But that is also what causes stairstepping."

Actually, the edge will also be blurred in digital cameras. While it's easy to *assign* a fully black pixel value next to a fully white value, it's usually not possible to aquire such values from CCDs or A/Ds or to display them.

I went and tried it in Vegas and man did it look good. The blacks are real smooth and more pronounced also the transitions between colors seems to be smoother. I posted some video grabs from Vegas one thing I love about vegas is that you can do split scene comparisons with the effects even play them back with before and after near real time to see how it looks with the changes you made.

Here is a link to what I did with a curve filter that I applied to all of these this is just one custom filter I did not adjust this filter for each scene it just seemed to make everything way cleaner!

The filter is applied on the left the original footage is on the right.
ftp://totalsolutions.bz/pub/curves.jpg

My article, Make Video Look Like Film has several examples of curves -- but also goes into the other aspects that are important. The "look" is really a complex soup of subtle details, not just one thing.

The article is online at:

http://www.dv.com/features/features_item.jhtml?LookupId=/xml/feature/2002/jackman1202

Ok. I am reading this thread trying to get some more insight on the 'film look' thing. I am using a Sony PDX10 for it's native 16:9 capability. One thing that comes to mind is this: whatever I will be doing to the footage in FCP or some other tool, it seems that video overexposure is very 'unfilmlike', so the first thing I will do is shift the AE point down two or three points and try to never reach 100% if possible. Since the PDX10 is quite a low noise camera (the CCDs are oversampled and processing is 14 bit) I guess it is a much better bet going dark than going bright, right? Thus I should later be able to play around with levels or curves and get a more pleasing image. Is this a good idea or am I missing something? Also, I am setting the camera to less 'sharpness' and more 'color' than default. My preliminarry results shooting with theese settings and adjusting levels in FCP seem quite satisfactory indeed.

BTW, has anybody tried Stib's effects? It is a set of plug-ins for FCP, and one of them "imitates film's response curve". It is a free download from http://www.scriptgeek.netfirms.com/

Underexposure is absolutely better than overexposure with video. You may pick up a touch of noise by having to boost the levels back up in post, but the tradeoff is easily worth it.

Hi Charles. Can you explain this "Expose for highlights" thing? Specifically, I would have thought that whether you expose for highlights or shadow aren't you going to lose details either way (Assuming your contrast range is too high). Is it more a psychological thing, in that we can accept crushed blacks easier than blownout whites so it's best to expose so the highlights are not blown? Or is there some other magic at work there ;)

Cheers
Aaron

I am not the authority; but in OIL PAINTING (which has tons of information on the 'use of light') a highlight can be as bright as the SUN or as simple as a wood panel on a roof. The highlights are the lightest point in the painting. IF you take an all black canvas and splash a little grey dot (does not have to be white), that grey dot will visually look brighter than it really is.

I would think 'Exposing for Highlights' would mean adjusting to the point where you lightest light is just visible? What am I saying here? I feel like I get it, but can explain it.

Maybe we do need Charles. CHARLES!

Boy, there are so many ways to explain it, using reference scales from all different facets of image capture: IRE from video, stops from film, zones from still photography...I'll just try to keep it simple.

If you take a typical high contrast image with a digital video camera such as a sidelit sun situation, what I am suggesting is that you dial the exposure down to the point where the sun side is as hot as possible without losing detail and color, i.e. the dreaded white blowout that we are all familiar with. This is probably a good 1.5 stops down from what the AE of the camera calls "proper" exposure. At that point, the shadow side may well be looking pretty dark, but unless it is absolutely black, i.e. no detail at all which is fairly unlikely, you will still have enough information on tape to work with. Chances are that you'll still have a full 100% of video somewhere in the frame, so the contrast will still exist in the image, and if in post you deem that you need more snap, it's a matter of lifting up the midtones a little bit.

Now, my preference working in uncontrolled situations such as sunlight is to shoot either with the sun behind me or in front of me, that is, frontlit or backlit. Shooting backlit is very pretty but does generally require some fill for faces, which can be as simple as a bounce card. The trick again is to underexpose so that the backlight is not blown out (watch it on blond heads of hair!). Frontlit late afternoon sun can be gorgeous also, just keep that exposure down and the colors will saturate beautifully.

Aaron, in terms of shifting the range by underexposing: digital video has significantly more latitude in underexposure, meaning that one can underexpose by as much as four stops and still retain some shadow detail, whereas you have maybe 1.5 to 2 stops of overexposure before you hit pure white. Different cameras perform differently, without doubt, but this is generally true. It is also possible through the use of filtration to elevate the blacks somewhat (I like the Tiffen Ultracons for this) which keeps them from crushing also. You just have to be prepared to lower the blacks in the color correction phase, and don't fret about your raw video looking "milky".

I talked a bit about controlling contrast in this thread (http://www.dvinfo.net/conf/showthread.php?s=&threadid=6281&highlight=first+born), which references a short that I shot on the XL1 a few years ago. The short is still viewable at iFilm, the link is listed at the top of the thread, but the level of compression is a bit high at that site so it's not all that easy to see what I describe in the posts, yet it may be helpful.

For the PSA I shot yesterday that was all daylight exterior, I pulled out every trick I can think of and then some to deal with heavy contrast all day long. When that spot comes online, I'll break that one down also.

I have printed this topic out and filed it.

I think this is similar to the approach I used for my Lady X
episode. I've also written down how I did this in this thread (http://www.dvinfo.net/conf/showthread.php?s=&threadid=14250&perpage=15&pagenumber=2)

(the post about visuals is near the bottom)

in that one pic with the 6 pics of the same girl.

what was being adjusted? (more than one i know)

sharpness?
contrast?
tint?

help?

thx
-arthur

Maybe I'm being a bit thick here. But what curves? What tutorial?

I have seen posts here that make claims, but no examples of the levels curves actually being proposed, or of this eagerly awaited tutorial/article.

Have I missed something?

Julian

Forget the theory, forget the mysticism. Just try out stib's curves, free for FCP and you will be delighted. If you also use FCP's deinterlacing or Stib's deinterlacing filter (supposedly a little faster) you will get very nice 'film like' results with DV resolution.

http://www.scriptgeek.netfirms.com/

Ignacio,

Any examples with using these filters? They look good (uh, free!), but was wondering if you have some examples? Thanks...
Clay

Sorry can't upload anything right now (not at home). However, just try the plug-in on your favorite footage, preferrably slightly underexposed stuff. Make sure you set FCP to 10-bit YUV for best results. You will not be dissapointed.