Exposing For DV - Aggressive Is Better

The traditional thinking behind exposing for DV is to expose conservatively to avoid clipping important image details. I used to believe that... until I did the following test. What I'm finding is that you can get better-looking results if you expose aggressively for the subject. I ran a quick test varying the iris to get different exposures.

What I mean by aggressive exposure: Expose to whatever looks best subjectively, even if it means significant clipping is some area of the shot (i.e. the subject's shirt in this case). Clipping is ok if the image looks better.

See the pictures and download the original footage at :

http://www.glennchan.info/video/exposure/exposure.htm

Ideally, you should view the material on a broadcast monitor. Computer monitors (especially LCD's) will not display things right.

I'd be interested to hear what your thoughts on this are, since "what looks better" is hard to objectify.

Glenn Chan

> I'd be interested to hear what your thoughts on this are, since "what >looks better" is hard to objectify.

Thanks for sharing your thoughts with us. A very quick response...

I think image no. 5 is more film-like looking. The minimal presence of noise in the shot doesn't bother me, esp. since in a moving image the noise is probably going to mimic film grain.

However, if you were planning on going for more of a bleach bypass type look in post, I think image 2 would be a more appropriate starting point.

Jessica Gallant
Los Angeles based Director of Photography
West Coast Systems Administrator, Cinematography Mailing List
https://cinematography.net/

Jessica Gallant wrote :

>I think image no. 5 is more film-like looking.

Maybe it's just my monitor, but I found the face on the graded image 5 looking very "plastic" (pushed too far). The shirt looks good on that one though.

Cheers

Martin Heffels
Filmmaker/DP/Editor/
Maastricht, the Netherlands

Speaking of DV exposure, has anybody any experience with underexposing 1-2 stops and then doing corrections in post? Is it better than to expose correctly and then do a film-look?

Thanks a lot!

Vidu Gunaratna
Video/Camera/Edit
Prague, Czech Republic

> Maybe it's just my monitor, but I found the face on the graded image 5 >looking very "plastic" (pushed too far). The shirt looks good on that one >though.

We may be experiencing differences of opinion based on our monitors. However, I liked image 5 better in part because of the skin on the student - his complexion is a bit blemished, and in image 5 it was smoothed out slightly.

(If you're shooting on miniDV, the production may not be able to afford a make up artist and instead try to rely on the actors themselves to do make up. Then again, I've shot miniDV features where the make up and hair dept. outnumbered the entire grip, electric and camera crews. <sigh> )

Jessica Gallant
Los Angeles based Director of Photography
West Coast Systems Administrator, Cinematography Mailing List

Glenn Chan writes :

> http://www.glennchan.info/video/exposure/exposure.htm
>Ideally, you should view the material on a broadcast monitor.

I'm unsure where to start. There are a number of variables you request for us to evaluate simultaneously. Ones that intrinsically make your test methodology too soft to produce meaningful results ... even with you request for subjectivity.

1) the particular browser used to open this page. Each browser has it's own image rendering technology. Try opening the same page with Firefox, Opera, Netscape and IE. You'll see.

2) our monitor's brand and model, and then calibration

IF every one of us lined up with our web-wired broadcast monitors to accommodate your test, we'd still be testing for the difference between browsers ... unless you meant for us to download your clips and bring them over to our editing stations.

IF what your seeking is validation that in run and gun situations you often times have to decide where to place your exposure, given that some things are going to suffer ... that makes perfect sense. That's been the purpose of exposure meters, spot meters, grey cards and d log e curves for about 70 years.

I admire your enthusiasm, but sometimes technology for technology's sake can lead one down very complex processes. Complex processes that have been well addressed, and rendered much simpler, for a bunch of decades.

"DV" doesn't change the relationship of light to recording medium. Those issues have remained the same since Louis Jacques Maude Daguerre hung out a window in down town Paris.

Shooting a MacBeth card (there's another brand more popular with cinematographers I can't remember), while varying exposure in small increments, from "Black Cat Coal Bins" to "Polar Bear Snow Storms" .... would produce, in a shorter period of time, all you need to know about the relationship your camera has to light levels ... without any extraneous, error producing, variables.

Simpler does not mean less accurate.

As someone posted very recently, each camera has it's own signature. They now pick a camera like they used to pick film stocks.

Cliff Hancuff
Washington, DC

I'd ask the subject to move and change to a t-shirt with no white in it.

Sam Wells
film/../NJ

Cliff :

You raise a valid point about browsers rendering the pages differently. I alt-tabbed between IE and Firefox and found that the image position is different because the line spacing in each program is different. However, I lined the images up as best and I could and didn't discern any difference between IE and Firefox. If there is a difference, I consider it to be reliable/insignificant because I can't see it.

If you really wanted, you can print screen and compare the difference in Photoshop using the difference composite mode. But because I can't see the difference, I wouldn't waste my time doing so.

2 - Downloading the clips and putting them into your editing program (and viewing the results on a broadcast monitor) would be ideal. One major thing is that your broadcast monitor will display the colour space correctly.

Another major advantage is that LCD’s can be really, really crappy monitors in terms of colour accuracy. Some only deliver 6-bit colour bit depth, and are overly bright. Old CRTs can also give a very inaccurate picture. Consumer TVs are overly bright, saturated, and may have flesh tone "correction".

3 - What is better?

Perhaps I am looking at things the wrong way. I think it may be more important to understand the differences aggressive and conservative exposure makes (which is summarized in the table if you scroll down). Each method has its own benefits.

If you agree with that, I don't think conservative exposure should be the "rule". In my opinion, I think aggressive exposure gives better results most of the time.

4 - Run and gun VS controlled situations :

In controlled situations, you can finesse exposure like I did the my examples. You could pick which image you wanted: #1 through #6. I suggest using a (calibrated) field monitor on location so you can evaluate your image as how it looks like. You can see for yourself how the clipped areas look.

In my opinion, a waveform monitor is a more useful tool than spot/exposure meters and a grey card. If shooting film, this obviously doesn't apply. Also, a waveform monitor does not tell the whole story and would lead you to around #5 (no clipping... but it doesn't look as good as #2).

In run and gun situations, you'll probably be a little off in exposure. So if you wanted #2, you might get #1 or #3. Suppose you think #1 looks horrible, while #3 and #4 look comparable to #2 (which is your favourite).

You might expose around #3 then, so you'll only end up with #2-#4 instead of #1-3. You also have to figure out how your viewfinder/LCD correlates to exposure level. It may be that the shirt has lots of zebras if you want to get exposure #2 (on 70% zebra, the shirt will probably be full of zebras).

5 - It would've been nice if I shot a test chart.

For example, I could figure out what was going on with the colour shifting.

However, you do have to watch out that the test chart results carry over to real world situations (where you probably won't be shooting test charts).

Glenn Chan

Cliff Hancuff writes :

>Each browser has it's own image rendering technology.

Glenn Chan writes :

>Cliff: You raise a valid point about browsers rendering the >pages >differently.

Not pages, images.

< sigh >

At this rate, new construction in Paris would have blurred Daguerre's image beyond recognition.

Glenn Chan writes :

>Consumer TVs are overly bright, saturated, and may have flesh tone >"correction".

I'll let someone else try to introduce this gent to the Avia calibration DVD.

"Glen, what Avia DVD? I didn't mention it! I just suggested someone braver than I might try."

This could well turn out to be the most entertaining thread CML's seen in a long while.

Anyone else want to take a shot?

Cliff Hancuff
Washington, DC

(Responding to an off list reply but a sense of duty compels me to share it)

We have Glenn Chan's set up to shoot.

A person in a room with some shelves.

You and your crew set up calibrated viewing environments with D6500 lighting (after testing various alternatives with Colour meters at the local Garden Supply store). Then, you guys can work on the monitor calibration, including sending off someone with test footage to IBC so it can be seen on Martin Euredjian's latest DCM23 (someone *please* confirm the black to black response times so that these folks can enter that data in their Palm Pilots); then you'll all be ready to proceed the to the double-blind tests with various web browsers....

Meanwhile, I'll ask the person if he can change to a different T-shirt; move him or the camera a foot or so, get the shot and go down the street and have a beer.

Glenn : Your methodology is good but verging on over thinking this. If the camera is properly white balanced and you've set your viewfinder contrast right, you have enough to proceed. I don't like white clipping, I could be happy if nothing exceeded even 90 IRE. Or less.

The argument to have any peak white is if you don't, you risk someone else pumping this signal up later.

OK, enough about you, let's talk about me. For a quarter of a century I have variously shot video using colour monitors ranging from bargain but serviceable Videotek conversions of Sony portable TV sets to a custom made Shibasoku with a hand lacquered aperture grille. In fact , prior to their being shipped to the US these Shibasokus were required to spend two years in a Zen monastery contemplating the very essence of red, green and blue. For this reason they require special Pluge bars
for set up which in addition to reference white and reference black also generate reference Void.

To this day I carry my own tatami mat on all video projects and only evaluate colour balance when seated in the lotus position.

Sam Wells

Sam Wells wrote:

"In fact, prior to their being shipped to the US these Shibasokus were required to spend two years in a Zen monastery contemplating the very essence of red, green and blue. For this reason they require special Pluge bars for set up which in addition to reference white and reference black also generate reference Void."

I remember those monitors... the chanting was sometimes a bit
distracting, but they looked great!

Jeff Kreines

My own DV experience is mainly with Sony cameras. My general approach with them is to:

1) keep lighting and reflectance ratios reasonably low, use grad and pola filters where necessary and possible, etc.

2) use 100% zebras and carefully consider how much burnout I can tolerate in a given shot

3) click my exposure DOWN to the next *detent* (which is not
necessarily the next lowest full stop)

4) in post, bring up the gamma as needed to put skin tones where they need to be.

I also have a bunch of "ideal" reference images stored on the memory stick to calibrate my eye to the VF, LCD or external monitor in a given environment. Very useful. With cameras other than Sony’s you can store such reference shots on a tape.

Dan Drasin
Producer/DP
Marin County, CA

I just want to add to this thread that Glenn, I meant no dismissal of what you were trying to learn, you're approach seems disciplined (read the many threads on proper monitoring, colour issues on the HD or Video lists here).

It's just that you can set up a Video Village with thousands worth of gear too, but sometimes that's an elaborate confirmation of the evident.

Also it might not be wise to generalize based on one specific DV camera.

When you get into things like DCC or knee point adjustment, well -- you see why some cameras give you more wiggle room in tough situations than others.

Sam Wells
film/../NJ

Copyright © CML. All rights reserved.