How to expose for the sensor

D

David Evans

Well-known member
Hi!

My question is simple, though the answer might not be. I've been used to 8bit cameras and always struggled to expose correctly, especially with harsh highlights. Now, I know these new high DR cameras are not an excuse to poor lighting but they do help with a few highlight issues. But that's not really my problem. Highlight recovery seems to be getting easier with these Blackmagic cameras with 13+ stops of DR. I, myself, am currently saving to buy a Ursa mini 4.6k and will do so as soon as all magenta issues and other QC are taken care of.

My question has more to do with getting the right exposure for the sensor. I mean, highlight or shadow recovery aside, how do you expose for each camera's sensor? Do you "saturate" the sensor like a lot of people do? This would mean exposing a few stops over and get it all back down on post to get clean shots. The instant con I can think of is that this would ruin any kind of ambience we could create for actors to interact with. Like shooting a night scene with lots and lots of light (turning day in night effect).

But some DPs also say that saturating the sensor is the wrong way to go, and everything should be exposed within half a stop of what the directors is aiming for, because those are the luminance and color values that the makers of the camera were expecting to "see" when they made the camera in the first place. This would mean a lot of light measuring, probably a lot of false color reading to get the skin tones where we want them to.

So, my question is: how are you exposing your shots and do you change techniques if using different cameras? Do you ETTR, do you use false color, do you overexpose to get as much light as possible while setting zebras at 95%?

Thanks for sharing your experience!
 
!'m with the expose correctly for middle gray crowd. Start with that and adjust +/- EI depending on scene contrast ratios. What middle gray is will vary depending on your camera profile. Typically you will have a practical exposure latitude of +/- 2 stops around middle gray. I still use an incident light meter and have profiled all my cameras to it for decades. I shoot an 18% gray card at known illumination value, bracket stops at least +/-5 on each side of ISO reference. That determines by scope values what meter ISO should be for middle gray at any given combination of camera profile and gain settings. IMO knowing where middle gray is against an absolute light value reference is the first step towards learning your camera in depth.
One of the great joys of shooting a Bayer raw format camera for me is that this whole process becomes much simpler. My Bolex for instance is accurately referenced to the meter at 200 ISO for nominal middle gray. Other ISO's are just shifting the reference middle gray up or down the total DR range around 200 reference. The camera has a solid usable 12 stops of DR, with 4 stops to white clip above middle gray at 200 ISO and 8+ stops to background noise floor below. Expose to the right (lower ISO, +EI) to improve shadow detail and reduce noise, Expose to the left (higher ISO, -EI) for more highlight protection in high contrast situations.
 
I use the grey card too, but then I allow myself to adjust my exposure by one or even two stops if I know the shot requires more headroom on the highlights (or shadows) than in the shadows (or highlights). And of course I check the histogram and zebras to make sure I'm not going to get into trouble.

This works great if you are shooting RAW or log, but it's not such a good idea if that's not the case.

I might use false color if my camera had that.
 
David Evans said:
So, my question is: how are you exposing your shots and do you change techniques if using different cameras? Do you ETTR, do you use false color, do you overexpose to get as much light as possible while setting zebras at 95%?

Thanks for sharing your experience!
Click to expand...

It took me a while to figure out how to 'expose'. I first tried to calibrate my DVX100B with my meter, and only had the db gain control. I then got a Lumix GH-1, and eventually I developed this method for myself. I have used it on my 'upgrade' camera, the Blackmagic Pocket Cinema Camera (BMPCC).

Figure out what a reasonable representation for a 18% grey card of the captured image. I use the Waveform display in Premiere to show the values of a grey card placed in a scene.

For the GH-1, and most Rec 709 based capture devices, with 8 bit representation... use 45% IRE for the resulting 18% grey card value. I used 50% for a while but that may be to high for some...

For the BMPCC, using blacmagic RAW and log representation, use 38% (technically 38.4%...) IRE.

Then verify how many stops above and below one can 'retain', using a step wedge target. There's X-Rite and Spyder color charts. I use the Spyder Color Checkr.

In any case, with the waveform display one can see how the high values are represented as well as the low values.

Depending on where the 18% grey falls one may have 2-3 stops above where values are well represented.

In the case of the GH-1 and Rec 709 and 8 bits... I only realistically had about 1 1/2 - 2 stops above middle grey. That's just the way it is... no marketing malarky... just a limit of Rec 709 and 8 bit representation.

The BMPCC has about 3 above and so one can see that deeper bit depth does yield the ability to accommodate a wider scene contrast range.

But ultimately one has to display the image, and these days displays have perhaps 6-7 stops worth of dynamic range, and so whatever is captured, needs to be compressed into that range.

Most people will not be presenting in a theater with fancy projector... most people will be viewing media on their computer, mobile device or home TV display, and a such, limits the contrast range to 'very limited' relative to the original scene.
 
I always expose for the highlights I want to retain, and then add fill light as necessary depending on the camera and it's dynamic range.

People b*tch about only have 10 or 12 stops of DR with some cameras, but I remember back in the bad old film days where you were lucky to have 5-6 stops of DR when shooting with reversal film. Everything had to be bang-on in order to get a usable image.
 
Thanks for all the great tips. Theoretically, if using false color, would it still be necessary to use gray cards to correctly expose?
 
DIfferent cameras and camera settings allocate different quantities of bits to each brightness stop.

You want to expose so that your critical image information is in that 'zone'

For 'thin' 8 bit camera that can be very proscriptive, for 10bit or raw cameras the distribution of bits of brightness per stop is much more linear.

The more evenly the data is spread per stop the more you can ETTR treating your camera as a bit collection device fighting only S/N ratio, where S/N is higher further to the right.

So how you expose is dependant on the camera, its set up and also some artistic choices like if you want to protect your highlights or not.

Also you will need to consider how much you are preppared to work the image in post.

S
 
For me the gray card tests are to profile the camera so I know how it exposes under different settings and scene conditions. I don't typically use in the field. But I do use an incident light meter to check key and ambient shadow middle gray exposure ratios of a scene. Then I can select the right exposure by eye anywhere in the scene. Having the broader DR of modern cameras makes this easier if one knows where the middle should be in the first place.
Many years ago I shot event publicity photos for an art museum, covering major openings. Lighting changed in the galleries depending on how a show was hung. I could walk through the gallery before the event with my Sekonic L-398 and profile the room lighting. I shot mostly with a very discreet quiet old Leica 3G that had no internal metering. Shooting to the lighting profile I did not need to use the meter at all during the event , just compose shots and set aperture within one or two stops range around center by eye. Never used flash. It wasn't allowed in the galleries.
 
I generally just use my lightmeter for exposure, as it's fast, it's consistent, and unlike zebras or other rubbish on-camera exposure tools - it gives you your actual lighting ratios.

Personally I don't think exposing your image has changed all that much with digital versus film (beyond the fact that your detail is skewed towards your shadows rather than your highlights). So most of the time I stick to the same principals of exposure I always have:

- if you want it bright, expose your subject one stop over.
- when you want it dark on a subject, expose two stops under (then just crush your blacks further in post if you need to see less)
- in heavy backlight, split the difference in stop between the lit side and shadow side of your subject

With the wide dynamic range cameras available these days, it's really only when you're pushing the limits of your sensor's latitude that you should need worry about adjusting your exposure to 'protect' your highlights.

Use fill light, nets, and grad NDs to balance the exposure in your frame to put it where you want it.
 
Grugs advice makes sense to me. I listen to the wanderingdp podcast and he's always talking about changing iso based on certain factors. It's all very confusing Starving the censor and over saturating it makes some sense to me but I still feel like I don't totally get it. What I'm saying is I like this thread. Thx.
 
Grug said:
Personally I don't think exposing your image has changed all that much with digital versus film
Click to expand...

Of course it has. A digital sensor is just collecting 0s and 1s, with a poorer S/N ratio as the signal is weak (ie less exposed). Those 01s are then mapped in camera or post to make a viewable image. The mission is therefore to collect the best signal and the least noise.

Some cameras (with knee/slope/log) however dont collect good signal when brightly exposed so ETTR is not alway the method to collect best signal.

The best method of collecting good signal is dependant on camera and setup of the camera.

S
 
morgan_moore said:
Of course it has. A digital sensor is just collecting 0s and 1s, with a poorer S/N ratio as the signal is weak (ie less exposed). Those 01s are then mapped in camera or post to make a viewable image. The mission is therefore to collect the best signal and the least noise.

Some cameras (with knee/slope/log) however dont collect good signal when brightly exposed so ETTR is not alway the method to collect best signal.

The best method of collecting good signal is dependant on camera and setup of the camera.

S
Click to expand...

Thanks for the info! If talking about BMD cameras, ETTR is always the way? Let's assume the Mini 4.6k for that matter.
 
My experience of BM cameras is that you need to ETTR fully on a low ISO but be aware that any clipping will not be recoverable at all.. and clip point in film mode can, I think be quite low.. like 90 out of 100 on the WFM.

Your image would then arrive in your post suite looking light and washy and be darkened appropriate to the scene.

To present the director (and help your lighting) you would on set toggle between the untouched image and a suitable LUT.

you should perform your own tests with your own monitor, camera and post pipeline.

BM cameras dont really have a 'knee' that would encourage you to keep data out of the close to clip range.

S
 
Additionally before the anti ETTR nuts chime in.

I would suggest not going for pure ETTR on each shot because you will make a headache in post.

One must learn to expose consistently across scenes/sequences where possible for a smooth post journey.

This means that some shots may not be fully exposed.

S
 
You must log in or register to reply here.
Back
Top