Zebra's??

What the heck is Zebra's used for when recording in raw file format

Im lacking zebra knowledge, so any tips on appropriate wording of the question, i'm all ears...

When shooting RAW, the basic idea is to expose to the right of the histogram … that is, get as much light to the sensor as you can without clipping (except, perhaps, for the very brightest highlights, as in chrome.)

This will generally give you an image that looks pretty much washed out on the view screen … until, of course, you grade it in DaVinci Resolve and realize what a fantastic camera you bought.

The zebra indicators don’t actually get recorded on the RAW file; they just indicate what parts of the image are getting clipped. If the zebra level percentage is 100, no stripes will show up until the image has actually been clipped, and those areas are lost beyond the range of the sensor.

I tend to set zebras to 95%, which gives just a little wiggle room. That way, if the chrome highlights are just beginning to break into zebra stripes, the rest of the image is probably pretty safe.

I hope that helps; I prefer shooting this way over trying to get a finished look “in-camera;” it makes shooting incredibly easy, as you can concentrate on composition and focus, instead of exposure. Much easier to get the finished look back home in Resolve.

ETTR is still the preferred method, even when shooting RAW?

ETTR is still the preferred method, even when shooting RAW?

"expose to the right" (ETTR), I have no idea what that means.

"expose to the right" (ETTR), I have no idea what that means.

True, no one has given that context here because we're talking about zebras.

If you look at the histogram it basically means the rightmost levels (brightest parts) should go as far to the right side without clipping.

On a histogram right is bright pixels and left is dark pixels The histogram is the distribution of pixels brightness.

If you take the exposure to the rightmost (brightest) edge you should just start to see zebras (if they're set to 100).

A caution for thought:
If one wishes to work in HDR (High Dynamic Range) rather than typical SDR (Standard Dynamic Range) one needs to be cautious of clipping the highlights... all of the highlights... at least the ones desired to be included in the image. So now this requires a "dance" with the noise levels in the dark image areas. The more dynamic range the camera is capable of the better. As to the use of Zebras for such, yes they can be used... but typically the brightest highlight areas will be small, even smaller than the zebra stripes. This is one case where false color can help significantly more. ie: decrease exposure (moving to the left on the histogram) to eliminate all red areas (red basically indicates clipped in BM cameras.) I'll typically have the UMP monitor on false color and the external monitor sometimes with a LUT to approximate the look and sometimes with focus assist and I always record in raw.

What the heck is Zebra's used for when recording in raw file format

Im lacking zebra knowledge, so any tips on appropriate wording of the question, i'm all ears...

Quite simple, Zebras are an exposure tool to use with th camera. They can be set (90-100 percent) to show what part of Rh image is overexposed, or about to be clipped by the sensor (depending on the setting). You do not want to clip the image being recorded by the sensor (it will be a bright washed out looking area, no image), so the Zebra in camera shows sensor clipping level when set to 90-100 percent, for example.

Next, on some cameras you can set the zebra to show correct exposure for skin tones, by setting the level around 40-60 percent, depending on the camera. Then when a face shows Zebras, you adjust the iris to just when the zebra starts to show in face highlights. This will a,low consistent exposure of skin tones in a face. Of course the level set, will depend on the complexion of the person be shot. A little experimentation here will show the best setting.

This was the main exposure tool on Boradcast ENG cameras before Histograms (which is more of a digital still camera tool), False Color, and other digital tools. Broadcast cameras used Waveform monitors to set exposures and Vectorscopes to set color detail. ENG cameras in the field, used Zebras, one for clipping and a second zebra pattern setting (diagonals going opposite direction) for setting subject face exposure.

A Waveform Monitor and False Color are your best exposure tools. The False color setup is like a 3D light meter, showing the main Waveform values as different colors, with a Red normally showing what is being clipped.
Cheers

Just to clarify, ETTR (Expose To The Right) is a technique where you set zebras at or near 100% and set your exposure to just below that point so that nothing important in the image is blown out or clipped. In theory, you are lifting the dark areas of the image above the camera's noise floor (the darkest point below which you start to get noisy images) resulting in cleaner footage. You then correct the images in post.

I don't know for sure but I assume the technique originates from using cameras like most Sony cams and the Panasonic GH4, etc, which become really noisy if underexposed.

The downside of ETTR is that you have to do a lot of post work because it's possible,that every shot is exposed differently. This particularly can affect skin tones. Most cameras have a sweet spot for making skin look good and by ETTR you aren't taking that into account as the exposure is based on the brightest part of the image which is going to randomise skin exposure to an extent.

I personally use ETTR occasionally when I'm shooting scenery but never when I'm shooting people as I want to make sure their skin exposure is consistent every time. Plus if I handed stuff shot ETTR to an editor or colourist they'd probably drive over and beat me to death with my own camera because of all the extra work I'd created for them.


Sent from my iPad using Tapatalk

Is exposure suppose to be set according to the output of the soft box lights?

Zebras are a form of light meter.

Shooting RAW doesn’t mean you get to skip the laws of photography. You still need to expose your shots don’t you ? You don’t want them too bright or too dark ?

Zebra is one of many tools we use as cinematographers to make an exposure CHOICE.

ETTR doesn’t really have anything to do with Zebra. ETTR is simply an exposure technique that maximises dynamic range and minimises noise.

In theory ETTR sounds great and HOWEVER I don’t believe it’s a good technique for motion imaging. ETTR creates a lot of shot to shot variation (you’re exposing differently for each setup) and that can be maddening to grade and often create shots that can’t be matched.

What’s the point of noise reduced High dynamic range shots if they take you out of the story on every cut ?

Many people use Zebra at a high number say of 95% or 100% as a way to use the ETTR technique because this high number indicates clipping or near clipping.

For several years BMD didn’t have false colour and Zebra was the only exposure tool available built into the camera.

False colour now allows us to see what’s clipping and makes it easier to visualise the exposure SPREAD.

Zebra is still useful as it’s the least offensive tool to use in the middle of take for the camera operator.

JB

i may be a little more clear with understanding zebra'safter playing around with my bmpcc's settings.

When i turned the iris on my lens adapter, i would turn it so its all the way open, letting light in. Then i would slightly back it or (close it)until i saw zebra's appear, is that what it means by 'expose to the right?

zebra's were set at 100 percent.

my iso was set at 400

and white balance at 3200 for both outdoor and indoor.

when indoors, i saw the zebra's appear in the window of my kitchen (slightly cloudy outside), then i would back it off until the zebras were almost gone but when in resolve, the edited clip was a little too dark inside the kitchen.

Many set their zebras at 95% so you may see zebras appear before someone else setting zebras at 100%. I set my zebras around 90% subject to the light. I believe zebras use an averaging of the three colour channels so at 100% you have a good chance you already clipped a colour before the zebras appear.

In the EXT-INT DAY you described, often you don’t need to worry about losing some information in the exterior if the focus of our attention is the interior. You normally wouldn’t go full tilt on ETTR.

Remember if you are shooting ProRes, you likely should not ETTR; you’ll get better results exposing for the scene, not the ‘right.’ If you do continue to ETTR for all codecs, you’re making too much work and the colour-corrected highlights may be too thin when in post.


Sent from my iPhone using Tapatalk

Hey guys,

Any input on this FB post saying that the zebras and false colors are only accurate in Video mode?

https://www.facebook.com/groups/8969003 ... 870055187/

That's a simplification. The internal Zebras and False Color work pre-LUT, which is the most useful way in my opinion.

trying to understand zebras....

when shot in raw, I adjusted my manual aperture on my sigma art lens (non booster) adapter until zebras appeared on a bright computer screen in my scene, then i turned down aperture until zebras dissappeared. then i turned it down a little more until that white graph on bottom left hand side no longer clipped. Is that what ETTR (Expose To The Right) means?

and do i maintain more editable detail with less aperture opening?

Hey guys,

Any input on this FB post saying that the zebras and false colors are only accurate in Video mode?

https://www.facebook.com/groups/8969003 ... 870055187/

I haven't seen that post, but based on the performance of the BMPCC 4K, I think you'd have to say it's the opposite: shooting a relatively bright scene in log, with a small amount of clipping in highlights as indicated by zebras and false color (red), produces very different on-screen results than extended video and video.

With no change of exposure (same scene, f-stop and shutter speed as above), 'extended video' shows substantially more clipping than log and 'video' shows an image which is almost entirely clipped -- solid red.

But in fact, there's only slightly more clipping in the 'video' and 'extended video' shots than log, after the shot levels are adjusted in Resolve. The only accurate zebras and false color were seen with log. With video and extended video, the display was way off, to the point of being useless. What appeared to be near 100% clipping in 'video" was barely clipped at all, far less than 1% of the shot.

If anyone regularly shoots in non-log modes, maybe there's some explanation?

Hey guys,

Any input on this FB post saying that the zebras and false colors are only accurate in Video mode?

https://www.facebook.com/groups/8969003 ... 870055187/

Actually you have that wrong.

The test shows that exposure tools are accurate PRE LUT.

They are accurate for the unprocessed RAW image, and when you apply a LUT if you use external exposure tools to monitor that image, you’ll be getting misleading exposure measurements.

The camera is always recording in RAW or FILM, why make exposure measurements through the complications of a LUT ?

This by the way, is how the camera has always worked. It’s expected behaviour.

JB

when indoors, i saw the zebra's appear in the window of my kitchen (slightly cloudy outside), then i would back it off until the zebras were almost gone but when in resolve, the edited clip was a little too dark inside the kitchen.

So you are doing exactly what ETTR practice tells you to do.

BUT

You’ve found one of the major downsides of this exposure technique.

It means you have a perfectly exposed image that’s too dark on the information that’s important.

Imagine you had two cameras rolling on that setup. Your second camera is a tighter close up on the subjects face. Imagine that the window IS NOT in the shot.

That would mean your exposure on the wide shot and on the close up would be different because your ETTR technique would have you brighten the close up shot, probably by a couple of stops.

So then in the grade you’re then faced with a nearly impossible job, of trying to match two shots with wildly different exposures. The wide shot has what’s important in your frame (the subject) a bit dark but hey, you get a perfectly exposed window to look at instead !

The art of being a Cinematographer is deciding what’s IMPORTANT in your exposure.

Zebra (and false colour) are just tools that tell you where the exposure spread is sitting.

You have to make a CHOICE about what to expose for.

I use Zebras as a way of telling me what’s about to clip and be lost forever. It’s a highlights are not recoverable alarm system.

Now you have to step up and MAKE the exposure choice. You have to evaluate what you’re shooting. Zebras will tell you what’s over exposed. You then DECIDE if it’s important to try and keep that highlight detail, or, do you let it go.

In your scenario, I would have let it go. I’m more interested in the subject, not the window, and I want my other shots to match the wide shot in exposure as well.

So either I have my gaffer and grip do something about how much light is coming though the window, or I just let it clip[ and burn.

It’s a choice.

Zebras just give you an absolute value for an exposure point. But you have to manage a whole spread of exposure.

JB

The test shows that exposure tools are accurate PRE LUT.

They are accurate for the unprocessed RAW image, and when you apply a LUT if you use external exposure tools to monitor that image, you’ll be getting misleading exposure measurements.

The camera is always recording in RAW or FILM, why make exposure measurements through the complications of a LUT ?

This by the way, is how the camera has always worked. It’s expected behaviour.

JB

All right on JB.

One current wrinkle I've run into when recording in BRAW and setting Extended Video (or Video) as the viewing metadata, the zebras and false color on the camera LCD (and the BMD EVF) are calculated after the viewing metadata, not before as with viewer LUTs. I was told this would at some point be remedied to have them calculated before the viewing metadata, but it doesn't appear to have been done yet.