ETTR vs. subtle/natural lighting

R

RedChalice

Active member
I've been a little confused when it comes to lighting a scene. Reading through threads on dvxuser, I always come across things like exposing to the right on the cameras exposure meter. I've always gotten more cinematic results lighting to what looks good to my eye instead though. More subtle lighting that to the eye looks good, but is underexposed on the exposure meter.

I know you can manipulate whatever you want in post overexposing a little, but that seems much more difficult than just lighting the scene till it looks good/natural on camera. I came across this guys tutorials on youtube and I really like how his images come out: https://www.youtube.com/watch?v=I30zVA9mv2E

The overall image is definitely underexposed according to his exposure meter in his camera, but the results looks great regardless. This is another one of his tutorials: https://www.youtube.com/watch?v=-K_9RfHWKM8

I know the image will be noisier shooting underexposed but with cameras nowadays, I'm wondering if that matters anymore. Especially with denoising software.

Even certain movies that have dimly lit scenes, I refuse to believe that there was a lot of light initially and it was then brought down in post. It seems like a backwards workflow to me. Overlighting seems old school and stage-like/soap opera'ish.
 
WELLLLLLLLLLLLLLLLLLLLLLLLLL, we are opening a can of worms here. Proper exposure and ability to light (or not light) are not really that related. As cameraman you decided what will be seen in the frame and it's up to your sensibility and your eyes to choose what to see and how much light to pour to see it. Pro cameramen also know that their work barely ever gets screened in optimal environment (for instance that clip of a boy improperly encoded might turn to total darkness) and so they "protect" themselves with certain tricks. For instance if you choose not to light the face you still want to have something bright in the frame (maybe a light in the back ground) to give the eyes a senses of full lighting scale. So the viewer knows the cameraman chose NOT to light the subject because you still see something properly exposed and the subject simply stands in the shadows. Of course the pro cameraman also knows that this can only work to a point. If a whole dialogue-filled scene is shot like that with multiple cut aways the viewer will grow frustrated that nothing beside the light in the background is visible. So it's a constant struggle on part of the cameraman to see how far one can take darkness and still get away with it.
 
Patryk_Rebisz said:
WELLLLLLLLLLLLLLLLLLLLLLLLLL, we are opening a can of worms here. Proper exposure and ability to light (or not light) are not really that related. As cameraman you decided what will be seen in the frame and it's up to your sensibility and your eyes to choose what to see and how much light to pour to see it. Pro cameramen also know that their work barely ever gets screened in optimal environment (for instance that clip of a boy improperly encoded might turn to total darkness) and so they "protect" themselves with certain tricks. For instance if you choose not to light the face you still want to have something bright in the frame (maybe a light in the back ground) to give the eyes a senses of full lighting scale. So the viewer knows the cameraman chose NOT to light the subject because you still see something properly exposed and the subject simply stands in the shadows. Of course the pro cameraman also knows that this can only work to a point. If a whole dialogue-filled scene is shot like that with multiple cut aways the viewer will grow frustrated that nothing beside the light in the background is visible. So it's a constant struggle on part of the cameraman to see how far one can take darkness and still get away with it.
Click to expand...

I know you don't have the time Patryk, but it really would be great if you were to write a book on Cinematography the way you see it in practice.

I'd buy it.
 
RedChalice said:
I've been a little confused when it comes to lighting a scene. Reading through threads on dvxuser, I always come across things like exposing to the right on the cameras exposure meter. I've always gotten more cinematic results lighting to what looks good to my eye instead though. More subtle lighting that to the eye looks good, but is underexposed on the exposure meter.

I know you can manipulate whatever you want in post overexposing a little, but that seems much more difficult than just lighting the scene till it looks good/natural on camera. I came across this guys tutorials on youtube and I really like how his images come out: https://www.youtube.com/watch?v=I30zVA9mv2E

The overall image is definitely underexposed according to his exposure meter in his camera, but the results looks great regardless. This is another one of his tutorials: https://www.youtube.com/watch?v=-K_9RfHWKM8

I know the image will be noisier shooting underexposed but with cameras nowadays, I'm wondering if that matters anymore. Especially with denoising software.

Even certain movies that have dimly lit scenes, I refuse to believe that there was a lot of light initially and it was then brought down in post. It seems like a backwards workflow to me. Overlighting seems old school and stage-like/soap opera'ish.
Click to expand...

Perhaps you maybe missed the point - you don't KEEP the over exposed image that way, you darken it back down to how you like to see it - thus pushing the noise down right along with it.

When I do it I light the scene the way I want it and THEN I raise the exposure either via the lens or ND filter and hit record.
It can also be done by raising ALL the lighting on set, and that may be a more "purist" way of doing it, but obviously that can be fairly difficult when it comes to raising everything equally.
 
RedChalice said:
I've been a little confused when it comes to lighting a scene. Reading through threads on dvxuser, I always come across things like exposing to the right on the cameras exposure meter. I've always gotten more cinematic results lighting to what looks good to my eye instead though. More subtle lighting that to the eye looks good, but is underexposed on the exposure meter.

I know you can manipulate whatever you want in post overexposing a little, but that seems much more difficult than just lighting the scene till it looks good/natural on camera. I came across this guys tutorials on youtube and I really like how his images come out: https://www.youtube.com/watch?v=I30zVA9mv2E

The overall image is definitely underexposed according to his exposure meter in his camera, but the results looks great regardless. This is another one of his tutorials: https://www.youtube.com/watch?v=-K_9RfHWKM8

I know the image will be noisier shooting underexposed but with cameras nowadays, I'm wondering if that matters anymore. Especially with denoising software.

Even certain movies that have dimly lit scenes, I refuse to believe that there was a lot of light initially and it was then brought down in post. It seems like a backwards workflow to me. Overlighting seems old school and stage-like/soap opera'ish.
Click to expand...

The aesthetic of the 'old school' movies was to have everything in the shot 'well lit'. That's not 'over lit', just a different aesthetic. There was of course 'noir' film based on two elements... the stories were often 'dark' in subject and plot. They were also lit with high contrast in many cases, or during some set of scenes.

For example, from the "Maltese Falcon"(1941)

The-Maltese-Falcon-690.jpg
 
Again: ETTR does NOT mean you keep the brighter image.

NO. You lower the exposure back down in post so it's right where you originally wanted it. And obviously this pushes the noise down lower right along with it.
There are other reasons to Expose a stop brighter (ETTR) but this is one of the big reasons.
 
I think one has to adjust exposure strategies based on scene conditions and aesthetic intent. ETTR works well in conditions where you can control the scene contrast ratios and keep it within a 5 to 7 stop range. You can get a very fat exposure to the right under those conditions without risking your highlights. Makes a nice clean well exposed canvas for grading. Outdoor daylight without some sort of contrast controls you will expose to the left as much as necessary to protect your highlights from clipping, usually one or two stops under exposed.
 
RedChalice said:
I've been a little confused when it comes to lighting a scene. Reading through threads on dvxuser, I always come across things like exposing to the right on the cameras exposure meter. I've always gotten more cinematic results lighting to what looks good to my eye instead though. More subtle lighting that to the eye looks good, but is underexposed on the exposure meter.
Click to expand...

One of the issues with 'light meters' is in the case of reflectance meters is that the area the reading is taken, yields an exposure such that the result is 'middle grey', or 18% relfectance, (or there abouts... )
In any case, that reading may require an adjustment to the exposure for what the actual reflectance is... say a white table cloth, or a 'fair skinned' person... so, one has to 'open up one stop' for the 'fair skin' or even open up 2 stops for the table cloth. And it goes the other way for darker complexion or darker subject elements, close down one stop... or two...

In the case of an incident meter... the reading yields an exposure such that any 18% (or there abouts...) element yields the standard 'density'... in the olden days, it was a density of .5 above film base + fog... in the modern case it is related to IRE % values, and dependent on the type of data encoding, such as 'log' or 'rec 709', etc.

For the Blackmagic Pocket Cinema Camera, and I presume all the Blackmagic cameras... for the BMDFilm encoding a 18% grey card should yield a IRE value of 38.4%... obviously that is so precise few people can actually measure that accurately... but say the value should be about 37-39%... and that is also the value for other cameras working in a 'log' mode.

For a Rec 709 capture, the IRE % for 'middle grey', 18% grey, seems to be between 45-50%...

I tend to meter for what I expect to see on the output... So for my BMPCC I tend to expect to see 18% grey at 38% IRE, and then use Resolve to 'adjust' from there for the presentation image. Give the dynamic range of the BMPCC, I can 'expect' to have about 3-4 stops over and account for highlights. The problem with under is one does get into sensor noise issues.

Even then I'll accept a 'noisy' image, just as I would have accepted a 'grainy' negative in the olden days for a 'no light' situation...

In that case I 'expose' for what I can get... ISO 1600 @ f/2.0 @ 180 deg. 24 fps.

23159506820_7bdba634e6_z.jpg
 
Last edited:
RedChalice said:
I've been a little confused when it comes to lighting a scene. Reading through threads on dvxuser, I always come across things like exposing to the right on the cameras exposure meter. I've always gotten more cinematic results lighting to what looks good to my eye instead though. More subtle lighting that to the eye looks good, but is underexposed on the exposure meter.
Click to expand...

There are basically two schools of thought here. One is to capture the look you want in camera, the other is to properly expose in camera and build the look you want in post.

The first is seemingly a left over from the film days, when the level of color grading one could do in post was limited at best, and the dynamic range of capture was huge, and the noise was in the highlights (negative film) and not the shadows. This caused DOPs to expose for the shadows to make sure they had the shadow detail they wanted (basically, ETTL).

The second acknowledges the "new reality" of silicon sensors that work more or less backwards from negative film, have a more limited dynamic range, have noise in the shadows, and have available considerably more power in post.

I'm not saying that you can't try to capture the look you want in camera. But I am saying that an experienced DOP standing next to you capturing the same scene by getting a good exposure and then grading in post will generally produce an arguably better image. But here's the question: is it enough better of an image that it makes you want to do the extra work? You have to decide that for yourself.
 
Keep in mind the camera is an important factor in this decision. You have to stay within the limits of your sensor - high end for rolloff, low end for shadows. Various sensors and recording formats allow differing amounts of post leeway. I've been bitten before, say, holding some info in a sky that didn't actually end up being "useful" data - correcting to see more clouds made a grainy mess. You really have to know how far you can go in flattening an image or leaving grading leeway.

But if I'm shooting 8-bit - I try to get it very close on set, and stay very clear of the high end of the histogram if anything will need delicate highlight detail.
 
Michael Carter said:
But if I'm shooting 8-bit - I try to get it very close on set, and stay very clear of the high end of the histogram if anything will need delicate highlight detail.
Click to expand...

This is pretty much what one had to do with 'reversal' film aka 'slide film'... one could under expose quite a bit, and still get a 'good' image, because highlight detail was retained, at the 'aesthetic' expense of shadow... so... just don't have large areas of shadow... problem solved...
 
If you're shooting Log or RAW, and don't have problems with blown highlights, you can overexpose by a stop or two in camera, then bring it down in post. This will result in a cleaner image with less noise. It works the same with 8-bit, 10-bit, and 14-bit. What you need is that the image is recorded in Log or RAW form, because for those it's easy to bring down exposure in post.

As DPStewart said, you don't keep the brighter image. You bring it down in post, and it ends up just as bright as if you didn't overexpose in the first place, but cleaner.
 
Bruce Watson said:
There are basically two schools of thought here. One is to capture the look you want in camera, the other is to properly expose in camera and build the look you want in post.

The first is seemingly a left over from the film days, when the level of color grading one could do in post was limited at best, and the dynamic range of capture was huge, and the noise was in the highlights (negative film) and not the shadows. This caused DOPs to expose for the shadows to make sure they had the shadow detail they wanted (basically, ETTL).

The second acknowledges the "new reality" of silicon sensors that work more or less backwards from negative film, have a more limited dynamic range, have noise in the shadows, and have available considerably more power in post.

I'm not saying that you can't try to capture the look you want in camera. But I am saying that an experienced DOP standing next to you capturing the same scene by getting a good exposure and then grading in post will generally produce an arguably better image. But here's the question: is it enough better of an image that it makes you want to do the extra work? You have to decide that for yourself.
Click to expand...

Other than the most recent modern negative film stocks, the idea that film has gobs of dynamic range is a meme. It is not necessarily true. The 16mm negative film stocks I shot with through the 80's never exceeded 9-10 stops usable DR. The Ektachrome reversal stocks, 6 to 8 stops DR. Cinematographers then shot everything within 6-7 stops contrast range if they wanted to see any detail in the shadows. Reversal stocks blew highlights pretty much the same as modern digital cameras do, though without some of the uglier digital artifacts.
 
Last edited:
Keep in mind that there is a difference between a camera's light meter (reflected) and a an external light meter which you place by your subject (incident.) An camera's internal light meter will average a scene so if your scene has a lot of white then a middle exposure is going to make your scene look grey and dark (and vice versa when filming a dark scene with lots of shadow). A true light meter is used to gage the exposure on a subject by being pointed towards the light source. Either way, there is always an element of artistic preference which is a good thing! Not everything should have a technical answer or a right way of doing things.

Anyway, ETTR is a different concept alltogether. If you have unused bit depth and dynamic range to work with in your image then its best to expose to the right. This allows you to utilize the cleanest range of the sensor.
 
Razz16mm said:
Other than the most recent modern negative film stocks, the idea that film has gobs of dynamic range is a meme. It is not necessarily true. The 16mm negative film stocks I shot with through the 80's never exceeded 9-10 stops usable DR. The Ektachrome reversal stocks, 6 to 8 stops DR. Cinematographers then shot everything within 6-7 stops contrast range if they wanted to see any detail in the shadows. Reversal stocks blew highlights pretty much the same as modern digital cameras do, though without some of the uglier digital artifacts.
Click to expand...
Good point, Razz. I think what we had was a tendency to accept the aesthetics of film grain, whether it be in large quantity or size...
 
scorsesefan said:
Good point, Razz. I think what we had was a tendency to accept the aesthetics of film grain, whether it be in large quantity or size...
Click to expand...

It was not just 'grain'. Since I only did still Film film photography, my main concern was to fit a wide contrast range into the 7 or so stops of the print. Now it is true, that when one makes a print, other than viewing lighting, it will be the exact same print seen by others.

In the case of modern media... the material can be viewed on almost any type of display, and calibrated... or uncalibrated to any degree conceivable...

So it is still good to think of 7 stops on the output display side.

The benefit of wider range capture, is one can work in the NLE/DI phase and fit the contrast range captured to the display range... and if one captures with only 7-8 stops... one is necessarily limited to how much 'futzing' can be done.

With 10-12+ one can go 'over' and bring back to the display range without major issues... in many cases... and any other proviso or addendum one may want to add...
 
Razz16mm said:
Other than the most recent modern negative film stocks, the idea that film has gobs of dynamic range is a meme. It is not necessarily true. The 16mm negative film stocks I shot with through the 80's...
Click to expand...

Yep, film stocks from 30 years ago didn't have as much dynamic range as film stocks today. But they had a heck of a lot higher dynamic range than digital did 30 years ago, largely because digital was still in the laboratory 30 years ago. If you compare apples to apples, current film to current digital, film (for all practical applications that's negative film) has had a dynamic range advantage during all of the last 30 years. Only in the present day has digital caught up, and that's largely because no one is doing any R&D with film any more. Film has topped out, but I gotta say that Kodak Vision 3 is a beast that set the bar very damn high. That an Arri Alexa can more or less equal it is seriously saying something about how far the quality of digital has come in 30 short years.
 
Bruce Watson said:
...That an Arri Alexa can more or less equal it is seriously saying something about how far the quality of digital has come in 30 short years.
Click to expand...

I worked in electronic imaging 35 years ago... the equipment cost 100s of $K... and only produced a 412x480 NTSC and 'square' pixel image (we tweaked the horizontal scan timing... since we were doing image analysis and not interested in maintaining broadcast anything...)... then there was the 'image analysis' engine... another couple 100 Ks... from an analog camera source. We had just begun to evaluate a 1024 linear ccd array when I left for 'brighter pastures'... probably should have stayed in imaging... but alas... life choices which could have been better chosen are a plague of human existence...
 
You must log in or register to reply here.
Back
Top