Proper exposure on URSA Mini 4.6K

2 stops less light. T4

Frame exposed at T4 using T2.1 Auto correction

Screen Shot 2016-06-07 at 11.47.20 AM.png (984.51 KiB) Viewed 15148 times

About 1 stop less light. T2.8

Frame exposed at T2.8 using T2.1 Auto correction

Screen Shot 2016-06-07 at 11.50.18 AM.png (960.02 KiB) Viewed 15148 times

And the open aperture upon which the Auto colour correction was applied. T2.1

Frame exposed at T2.1 Resolve Auto correction

Screen Shot 2016-06-07 at 11.54.32 AM.png (941.41 KiB) Viewed 15148 times

How do you guys know that the 2 methods don't produce similar results? I've tried both that are mentioned and 9 times out of 10 they're very similar. Only shots that may be an exception are very bright backgrounds.

Let's look at overexposure on skin for a minute. Yes it's less of a problem than underexposure but the UM4.6 and every other manufacturer is playing highlights vs shadows vs midtones on the sensor, it's a balance between how much range on the curve we give each at base ISO. The toe and shoulder on that curve are designed to roll off highlights etc but the center of the curve is setup to maximise information in skin ie. Midtones.

You are mixing up film and tone mapped delivery curves (i.e. rec709 with a soft clip, or anything else meant for delivery) on one side, with linear and log signal (meant for post). There is no shoulder with any linear signal, or any real log curve (except Cineon and its derivatives, which are supposed to encode film scans and do replicate the shoulder of the negative): there is no shoulder over log-exposure on Arri's LogC, or Sony's S-log, or any other log curve with a true log upper part.

The consequence is that with most digital cameras shooting raw or log you will get the cleanest signal using ETTR (and yes, this includes the best and densest skin). Whether this is the best approach for your colorist, who will then need to match exposures shot-to-shot, is another story.

Thanks so much for bringing this up as it's a great way for us to get deeper on this! I'm advocating two things
That DP's use a meter or false color to hit target exposures they arrives at through lighting tests
They don't subscribe to the oversaturating the sensor and backing off of clipping in zebras philosophy as a means of exposure.

Lets do RAW first
The ETTR approach as you outline is to not use the last two stops of dynamic range and to place skin two stops higher on the sensor (linear, and I was oversimplifying earlier but I'll get into that in a sec) so as to avoid the noise floor. This is effectively just redistributing dynamic range, and to use Alexa as example moving the ISO down to 200 for +5 and -9 stops (roughly) and is a effective way to avoid noise. When we stop to think about how this will actually work in practice though things change. We shoot two stops over, avoid noise, bring it in and lower it two stops. Fine, great, clean image. But as we apply contrast and expand the data through the range we see that that redistribution is a problem in the brighter parts of skin. Yes shadows are clean but we really would have liked to have more range on the face because as the sunlight wraps around it looks a bit too pale and not dense enough and that's because we have two less stops of range to play with for contrast above midtones, while balancing out the sky we need to squeeze skin into a smaller range. We've only captured 5 stops above middle grey and that's the tradeoff.

Next we see sensor-linear, to Log-C ARRIRAW- LOG, to a REC-709 or camera or normalization LUT, we have a sensor conversion designed to maximise data in a RAW container with LOG and another conversion to look like film or the way a human eye sees the world with the inverse curve or the camera LUT. Here we see a problem because the whole thing is setup to accept a image where the middle of the curve matches skin. No one I know has ever watched a feature in LOG-C (although some directors might like to:)) so we need to talk about ETTR in this context of watching the thing in the real world on a display and when we come in to this very fine tuned workflow with a image that is two stops over we are landing skin in a place that is incorrect. The only way to get it back to where the workflow is expecting it at 18% grey is to window out skin in every shot or to shoot a grey card for every shot. Well.... Wouldn't it be great if camera manufacturers had thought this through and made it easier on everyone where they matched all these complex input and output LUTS with settings that allowed for less noise? Yup. Just shoot at 200 ISO. The entire reasoning behind ISO on every camera not the Varicam (which does dual ISO on the hardware level) is exactly this. The base ISO as we all know is what the camera is reading on the sensor level and the lower ISO's are just gaining up and down.
Case in point when dragon was released it was supposed to be a base ISO 2000 camera, after noise issues the conventional wisdom became "shoot at ISO 320 with STHL OLPF on Dragon" entirely as a way to avoid noise.

NON RAW FORMATS:
Anything uncompressed means stretching and changing values as we gain up and down and that's very bad for the image, banding color changes etc Also we see the same issues as before with landing middle grey as expected for the conversion luts etc, and more of a issue as we use log to compress highlights to fit smaller containers. This is really a non starter at all.

Again I'm not saying there is any issue with bumping up a stop to get a cleaner image, thats been happening forever, I'm arguing against using zebras as a exposure tool and backing off from clipping as I feel that landing midtones in a way where we don't know how to get them back to midtones leads to real issues in the color workflow and that we should use ISO or adding a stop on the meter for this.

Would love your feedback and we can definitely get into raw log and bit depths for the real fun;))))

Hi Guys,

I’m a little confused about exposure on the URSA MINI 4.6K. I’m thinking, if this camera isn’t overrated? Does it really has native ISO 800?
I did some tests using an incident light meter set to ISO 800 and the appropriate frame rate and shutter settings. I was very disappointed as I saw the results in DaVinci Resolve 12… Proper exposed face by light meter is about 370 (range from waveform). RAW footage seem to be ca. 1.5 - 2 stops under.

Am I doing something wrong?

Thanks for Your answers!
Fryderyk

To answer your question, if you measure the light with an incident meter and use those values in the camera, the footage should be correctly exposed, not underexposed... `

Are you using ND filters, and correctly compensating the values you get from the meter?

Aharon, there is no problem with matching midgrey. LogC is well defined mathematically. No matter how you expose, it can be linearized with the inverse log-to-linear transform. Once in the linear domain, you can do an appropriate exposure correction (which is a simple multiplication in case of linear signal). Then, should you wish, you can transform back to LogC, and, voila, you will have skin (or midgrey, or whatever reference you like) wherever you think it belongs on the LogC curve (yes, minus highlights latitude).

There are some practical considerations with ETTR due to different gains, sensitivities and clipping levels on the R, G and B channels, and the non-linearity near saturation with some sensors which, depending on processing, may or may not translate to the recorded data. But doing ETTR with a safe margin will not result in loss of important highlights, including highlights on skin. Now, I am not saying it is the best way to shoot (I don't use ETTR, I prefer re-rating cameras slower than nominal speed for shot-to-shot consistency), but this does not cancel the fact that ETTR used right will deliver the cleanest image and skin per shot.

I've starred my replies. Again no disrespect intended this is simply a discussion on technique and I'm happy to learn more....
Aharon, there is no problem with matching midgrey. LogC is well defined mathematically. No matter how you expose, it can be linearized with the inverse log-to-linear transform.

*The transform from log to linear is not the issue. The issue is that midgrey is about two stops overexposed. (For example)

Once in the linear domain, you can do an appropriate exposure correction (which is a simple multiplication in case of linear signal).

*When using ETTR or backing off of 95% there is no known appropriate exposure correction as zebras for 95% highlights do not necessarily correspond to midtones.


" Then, should you wish, you can transform back to LogC, and, voila, you will have skin (or midgrey, or whatever reference you like) wherever you think it belongs on the LogC curve (yes, minus highlights latitude)."

* At this point I would use a REC-709 or camera transform LUT to add appropriate contrast. I would not go back to log-c and even if I did I would then go right back to adding contrast which however it is added is working on the s curve principle. These luts use a S curved shape and are matched closely to the camera response and are set up to deliver maximum information in middle grey while smoothly rolling off highlights etc. To bring middle grey to the optimum place as explained earlier I would need to shoot a grey card for every shot to achieve what you advocate. I also don't see how missing two stops of highlight info on skin would not effect it negatively, this is further compounded by the need to leave more information for brighter areas in the background.

" but this does not cancel the fact that ETTR used right will deliver the cleanest image and skin per shot"

 * As I've said before pushing exposure is a valid way to reduce noise. You haven't addressed how ISO is set up to deliver the same results in a less complicated way.

Please explain how losing dynamic range on skin is not a issue, please also explain how overexposure and varying exposure in skin will not create a issue in real world grading applications using rec-709 luts. Please explain the benefits of shooting with ETTR and backing off of highlights in zebras vs simply shooting with a lower ISO

Thanks a bunch!

Here is a look at an interior exposed in raw using false colour to get the palm of the hand well into the pink. Note the considerable clipping of the fence outside on the right as shown by the parade and waveform.

Rick,

Thanks for doing these tests. Even though I like using a meter (and zebras when in a run and gun situation), FC is growing on me and I'm learning how to more effectively interpret ratios so I don't have to leave my camera. IMO, the FC exposure looks best. Now that we have a camera that clips nicely, I don't mind the UM4.6k's smooth roll-off as long as the chipping is not too extreme. Besides, I see clipping everywhere on feature films. It's kinda funny how it's thought of as a major deal on this forum. On another note, your "magenta issue" seems only slightly noticeable at f8 and your Lang Night effect is clever and I will certainly give it a try next time I need a day-for-night. Thanks again.

Thanks, Scott.

It takes awhile to see the world through this false colour spectrum, but once you do, it becomes easier as you know. The most important colours to remember are blue, green, pink, yellow for those items you want some detail.

Purple and red are too dark and too bright, but can be in your frame as long as you don't expect those areas to have detail. But when purple finally showed up in my Lang Night tests, I actually thought it was dark pink.

Light grey above pink is easy enough to identify, but the medium to darker grey can be hard to pickup on in a busy shot, but in terms of exposure, all the grey shades are safe.

Blue is a warning you are about to crush detail in the shadows. Green is your 18% grey card. Skin is pretty in pink. Yellow is a warning you are close to clipping.

‚"False Color Switches the false color feature on and off. False color overlays different colors onto your image that represent exposure values for different elements in your image. For example, pink represents optimum exposure for skin tones. By monitoring the pink false color when recording people, you can maintain consistent exposure for skin tones. Similarly, when elements in your image change from yellow to red, that means they are now over exposed." False Color Chart:
See page 57 of the current URSA/Mini camera manual.

Sent from my iPad using Tapatalk

You guys saw the ETTR vs. light meter grabs I postet on page 1?

Thoughts?

Frank, for the pair of frames posted, the ETTR seemed the better overall. Even the right side of the frame helped make the overall impression of a better colour grade. You always use controlled light levels where possible so you can achieve your predictably satisfying results in various situations. If it ain’t broke, why fix it?

But there may be times or for other shooters with different approaches in which using the pink false colour is a timesaver and a better exposure of the talent. Whichever technique is used, always good to take a peek around the frame using the other exposure tool. From what I gather, Aharon isn’t saying to ignore zebras or light levels, but trying to keep skin where there is more latitude in the log curve. I had always thought ETTR would mean that you are feeding the sensor more light, but my image of the girl’s hand and arm illustrates that there can be times when ETTR feeds less light to the sensor than pink false colour.

F You always use controlled light levels where possible so you can achieve your predictably satisfying results in various situations.

Yes, I actually use my lights to expose
Isn't that what photography is all about - painting with light?

Cinematography, yes. Videography, not always! Always impressed with the quality of results you achieve painting with your lights.


Sent from my iPad using Tapatalk

Let's separate out the technique and effect. The technique is ETTR, noise centric, saturate the sensor, by backing off of clipping using zebras. The effect is bumping up the exposure by various amounts.
Bumping up exposure and 'printing down' to use a film term, will always result in a cleaner image with less noise.

This can be achieved by adding a stop or two on the meter or just using a lower ISO. Both will do exactly the same thing.

The issues with ETTR is that when using zebras as a exposure tool we do not know where midtones (what the meter is reading for) and skin are exposed at and the exposure for midtones will change depending on background brightness as we move through a scene.
This creates two specific problems plus a third general problem.
1. We no longer work with skin or midtones at the values intended by the camera manufacturer in the color workflow. Where skin falls in relation to shadows and highlights is very important in color and you can see this effect by running tests with rec-709 LUTS or your own curves on log material. While the mathematical transform from log to linear is lossless that is one part of a chain of correction and like I said before the camera manufacturer is the one putting out the conversion LUT and no one watches anything in Log-c etc etc
2. Skin can be under or over exposed as we see different background values and we need to correct to one look throughout, this works against the primary responsibility of the DP, namelybto deliver a coherent scene, and the correction process for different values also hurts skin.
3. We lose dynamic range by however many stops see ETTR and skin often uses a good chunk if the range above and below middle grey. When we take into account the need to leave some range for background brightness we can see where skin might end up in a flat less interesting place.

The easiest way to shoot for clean shadows is to shoot with a lower ISO. This is exactly the same thing as ETTR and avoids all the issues mentioned above. The dynamic range is redistributed downward and that would be taken into account in relation to problem 3.

I personally dislike losing range on skin so I most likely would bump down from 800 to 640 but many DPs shoot Dragon at 320.

The issues with ETTR is that when using zebras as a exposure tool we do not know where midtones (what the meter is reading for) and skin are exposed at and the exposure for midtones will change depending on background brightness as we move through a scene.

If I have anything bright in the background, I have to decide to ether let it blow out (window), make it darker, or just throw some more light at the skin. That's what my lights and flags are for. If you don't control the light, and you use false color to get skin in the ballpark, that's fine, but you gonna end up with blown/clipped highlights or noise in the shadows. The price you pay, for not lighting your scene.

But this has nothing to do with exposure per se, but with the lack of control over your ratios.
If you have proper control over your ratios, you can expose ETTR or use a meter, and after CC, you pretty much end up with the same image, as in the examples I posted.

I guess it also pretty much depends on your personal visual style.
When you look at an average frame of my typical material (when I have the saying), I grade the brightest parts (except car headlights, chrome etc) usually around 75 -80 IRE max, which makes my skintones land at about 30-35 IRE. I would not shoot it that way in camera, I expose it much brighter, ETTR and bring it down in post.

Please explain why you don't achieve the same effect in a more precise way by shooting with a lower ISO.
Thanks

Great discussion - I appreciate the solid back and forth on different exposure practices.

For those using False Color, do you go by Video or Film dynamic range on the monitoring output? Are there differences between the IRE readings with the Blackmagic EVF versus an external monitor?

I've found with my SmallHD monitor set to FC, it seems to show IRE at 100 for highlights before the zebras on the Ursa Mini display are showing 100%, so which should I go by?

Well as you've seen there are different practices:) I can advocate one approach but as you've seen with ETTR it's very much about trying different methods out. Similar to painting or writing, there are tools out there you can use the craft to create your art.
My approach is oriented towards getting the image in camera to match the final look. Others will prefer a technical capture that (to them) maximises data.
One approach is to toggle the display back to log or RAW to check false color. This is useful because you are working with what is really being recorded and can make exposure decisions that allow the freedom to work with the full dynamic range of the camera.
Another approach (and the one I advocate) is to work towards the final viewing environment. This can be accomplished by monitoring and making exposure decisions on raw or log footage viewed under a LUT that is close to where the final grade and viewing medium (theater vs broadcast vs web.) Shoot a lighting test. Grade to taste in Resolve. Use film LUTS or rec709 LUTS at different opacities, try curves, try just contrast and color. Arrive a look that you like. Right click on the clip and export as a .cube LUT. Load that onto a Oddessy or Atomos display. Adjust the brightness of the display down to match theatrical, higher gamma for web, down to adjust for eyes at night etc and this is really again to taste and should be to set values so it doesn't get messy ie. Down to %30 for night up at %100 for day. Expose with a meter while matching and tweaking lights knowing that what you see is what you get.

Aharon, you might like the URSA 4.6K firmware 4.0 release that I'm hoping will be delivered in June. It includes supports for 17x17x17 LUTs.


Sent from my iPad using Tapatalk

[Dmitry Kitsov BMD]

Respectfully disagree that 95% zebras should be a hard and fast rule. I was shooting with 90% zebras and saw no clipping, but when grading the video, a blue channel appears to clip using BMDFilm. I could pull it down though so I guess this wasn't a hard clip. I've set my URSA Mini 4.6K to 85% zebras to keep me a decent margin below full ETTR.


Sent from my iPad using Tapatalk

Rick, if you are able to “pull it down” in grade, then it was not clipped, but rather you are observing the waveform of the interpretation of the image data by the color transform. Be mindful, that waveforms and other image analysis tools in Resolve are centered around the output/delivery values, rather than the incoming image data values. If the blue channel was not oversaturated to the point of the flat shelf on the top of the blue channel waveform, then it was not clipped.

Thanks, Dmitry!


Sent from my iPad using Tapatalk

Very nice thread indeed, guys!

This can be accomplished by monitoring and making exposure decisions on raw or log footage viewed under a LUT that is close to where the final grade and viewing medium

Creatively, yes. I think that the best way is to actually "calibrate" your monitor depending on your final output with a LUT that's close to the one used for delivery and be able to judge the images based on what you see. It's easier for you as a DP because you can practically "bake in" the look, easier for the client (if you've got people on set) and much easier to respect the ratios from scene to scene (especially if you apply a false color over the LUT).

But, technically, I think it's a question about how the camera "holds" the LOG for the entire dynamic range.

So, Aharon, because you also mentioned the Dragon, I believe experience will show how the URSA Mini 4.6k actually behaves in both scenarios because there are some (and different) best practices from camera to camera.

@ Octavian, yes I think a discussion on craft and how to get the image the best place with different techniques is useful because there are different ways to do the same thing and these different approaches. We can all learn from this back and forth and I've definetly taken a lot from the discussion here!

@ Rick we've seen the BM color science come a very very long way from those early yellow overly contrasty rec-709 LUTS. I think the current color science (from what I've seen here... haven't really worked with any Ursa 4.6 material in post yet) is going to be excellent and looks very usable right out of camera. This is a real credit to BM and kind of something that's gone a little under the radar honestly with all the magenta noise (some of which I've made as well:)))) Anyways again just from what I've seen here I would prefer to work with BM LUTS rather than the complicated Sony S-gamma3 mess with all the variants and a still very hard and technical feeling to the LUTs.
Anyways having the ability to load whatever you want in camera is great because for instance you can combine some BM color science with a LUT in Resolve at 30-40 percent and say a film emulation LUT at 30 percent for a really nice custom look. Some conversion needed to go from 33x33 to 17x17 and not as precise etc but hey it's a good idea.

we've seen the BM color science come a very very long way from those early yellow overly contrasty rec-709 LUTS. I think the current color science (from what I've seen here... haven't really worked with any Ursa 4.6 material in post yet) is going to be excellent and looks very usable right out of camera.

So far in my testing of the 4.6k raw and log images, the BMD 4.6K log to REC709v3 LUT may be an improvement on the past, but it still results in very contrasty images and extremely saturated highlights. I assume this is by design as the REC709 LUT is there to replicate typical broadcast video which increases saturation as luminance increases all the way to 100%. Film (and some of the newer digital cinema cameras) gently rolls off saturation once luminance goes above 40-50%. Hook released a set of power grades a couple years back that included a saturation node that did exactly this using the LUT v SAT curve to gently roll off saturation to nearly zero as luminance passed from 50% to 100%. If you're going for a look more similar to film, it really pays to do your own grading from the log image or to create your own LUT.

In general, the 4.6K grades really easily from log. Add some contrast, a little saturation and you have a great starting point for a grade.

Short video interior day scene, no audio, exposing using false colour and an Apple iOS App, Pocket Light Meter. Not an exhaustive or very scientific test and I graded all clips with identical or comparable settings in DaVinci Resolve Studio 12.5. Camera set to either ISO 800 or 1600. I like what 1600 gave me when I knew light was low. You be the judge.



password UM46K

I did it, Frank. First I set zebras to 95% instead of 85%, and then: I've turned off zebras. Relying on false colour now for a proper exposure. I frame with the monitor or BM Viewfinder normally, hit button three to use false colour to adjust the exposure settings, then turn it off to shoot and monitor the shot. Goodbye zebras and goodbye peaking on the monitor. I use peaking on the BM Viewfinder though.


Sent from my iPad using Tapatalk

I know that this topic is dedicated to Ursa Mini 4.6K but can somebody tell me how i can make a proper exposure on my Ursa mini 4K?

In a nutshell: shooting raw, rely on the zebras set at 95% or even 85%; shooting ProRes, rely on a scene average for general exposure.

The corollary here is if you are primarily shooting people as in a narrative film, try to keep your exposure for skin consistent from shot to shot as it will speed up your grading in post.


Sent from my iPad using Tapatalk

I've read this entire thread and I'm still no closer to understanding what is the best practice for exposure on the 4.6k... Is it ETTR with Zebras at 75,85,95 or 100. Is it exposing skin tones correctly using a light meters and FC? Which one? From the test pictures I really can't tell much of a difference, although maybe a little better on skins with FC? Argh!

If in doubt, get a real rgb scope (external monitors usually have that) and use that to expose

i was just on a shoot yesterday with my 4.6K and now i'm leaning towards "get as much light as possible to your image" after seeing noise @ 800.

yes, vertical blue lines at 800 ISO.

in my mind, i'm exposing the shot as i want it to look like in post, but apparently i'm wrong.
and apparently you can't film scenes with dark backgrounds (even with properly lit subjects) because the vertical blue lines were there at every ISO.

is this normal for the 4.6K or is this unique and bad?

i was just on a shoot yesterday with my 4.6K and now i'm leaning towards "get as much light as possible to your image" after seeing noise @ 800.

yes, vertical blue lines at 800 ISO.

in my mind, i'm exposing the shot as i want it to look like in post, but apparently i'm wrong.
and apparently you can't film scenes with dark backgrounds (even with properly lit subjects) because the vertical blue lines were there at every ISO.

is this normal for the 4.6K or is this unique and bad?

I have the same problem. For most occasions I seem to be able to grade it out, but its still there if you know what to look for. Its quite annoying.

With that said, other scenarios, I love the look of the footage.

Adam, my testing revealed that sometimes ETTR is feeding the sensor less light than using false colour. No one had mentioned this happening so I was surprised, but it's true. We all know that you can't expose to the right for specular highlights or your scene will be way too dark. But there can be many things in a scene that are very bright but also may be non-essential to your shot. For example, your subject is interior, but there's that window and the brilliant outdoor sky. ETTR will force you to stop down or add ND (assuming you are not using gel over the window) and your subject will be underexposed. Using false colour with a consistent pink on skintones, will give you a great exposure of your subject and a blown out sky. I'm not going to worry about the sky. Pushing skintones is not a good idea. Pulling sky might work, but if it doesn't, who cares? So your false colour exposure is maybe two stops more light than ETTR. FC rules!


Sent from my iPad using Tapatalk

A So your false colour exposure is maybe two stops more light than ETTR. FC rules!

There seems to be some confusion between exposing and ratios, and the use of ETTR and FC..

If you ETTR, of course you don't stop/ND down, till a super hot window, car headlights or a lamp shining into your camera don't show zebra anymore. You let those things burn out - with 14 stops, you can't have it all anyway (unless you want to bring in 10000k of HDMI to counter light this).

If you ETTR that way, i.e. saturate the sensor as much as possible in the given situation, you have all fine tuning options in post.

If you use FC to nail your skin tones, everything else has to fall in place. That means you ether have to adjust your ratios (bring more lights for your shadows, or flag a tablecloth or white shirt), or you pay with noise in the shadows/blown out shirts.

Exposure (iris/ND) in camera, always means that you only control the overall brightness. that is like having only the offset wheel in post, and nothing else.

There is no lift-gamma-gain in the camera. You have to do lift-gamma-gain with your lights/flags on set, to get the proper ratios. Relying only on FC skintone doesn't do the trick, since it only controls the overall brightness, ruled by your skin tone.

Thank you Frank!

Frank, I agree with what you are saying. But I've been in lighting situations where using false colour meant I open the iris more than when I used ETTR on the same scene. I wasn't blowing out or trying to protect car headlamps. We're both trying to feed the sensor as much light as we can. Your stuff always looks superb, but you know I don't have all the tools or talent that you have. So I'm relying more on false colour at this point. I don't always have mostly pink skin while using false colour as I can let some of it go to light grey and/or medium gray. Thanks for your comments. I do watch for and try to adjust for red and purple too when I can.


Sent from my iPad using Tapatalk

I'd say expose to flesh tone 18% for the majority of your material and only ETTR if it doesn't compromise your dynamic range. ETTR can reduce your DR by potentially under exposing your shadows at the expense of highlight retention (depending on scene DR). Conversely ETTR on a low con image will allow for less noise in shadows when brought down in post but you also lose consistency in your material if the latitude slides up and down from shot to shot. Work out what's important in the image, is it better to hold the highlight information or the shadow information, or can the camera hold the entire scene DR at 18% grey. Do I really need to see highlight information in the white window outside with flat white sky or is it more important to see the face inside that's 10 stops the difference? Conversely do I care about the side of the lady's face that has a hot side light that's 2 stops over my 15 stops DR, if so then LIGHT her face or bounce some light on the fill side or reduce the side light brightness so that the DR is full but not exceeded. If noise at 800asa is usable I shoot at 800 across the board inside and out (with NDs obviously, although I sometimes use NDs inside to control DOF). That way the DR, noise and look is consistent on the whole film. If you want to reduce the noise later then run a slight NR across the whole thing.


Glenn Hanns
DOP
Sydney, OZ

So when I have a dark spooky scene or a bright cheerful beach scene, I should still use the same false color levels for skin tone? The skin tones would be consistent, but it seems this method wouldn't serve the intended final look for each scene. Or would it?

You could. Dark can be created in post. But seriously the guidance on using false colour for consistent skin tones may be more applicable in your scene (depending upon lighting intent) or very similar scenes, but not your entire film.


Sent from my iPhone using Tapatalk

You could. Dark can be created in post. But seriously the guidance on using false colour for consistent skin tones may be more applicable in your scene (depending upon lighting intent) or very similar scenes, but not your entire film.
Sent from my iPhone using Tapatalk

Thanks Rick,

Yes, I believe using false color levels to keep a particular scene consistent for skin tones, is the best way to use false color. With regards to creating dark in post, I found that if you expose too light (normally lit) that can be difficult to adjust succesfully. I have attempted to create night in post (day for night) and if the scene is lit too bright (with no clipping at all in RAW) it just doesn't look as believable in the same way if I shoot say one or even two stops under. But false color is definitely the way to go for keeping skin tones consistent.

Would be good to do some testing of that. I wonder if a thinner exposure in DAY creates a better NIGHT in post because in real nighttime, you don’t see the normal daytime rendition that is richer in colour and details.


Sent from my iPhone using Tapatalk

Would be good to do some testing of that. I wonder if a thinner exposure in DAY creates a better NIGHT in post because in real nighttime, you don’t see the normal daytime rendition that is richer in colour and details.


Sent from my iPhone using Tapatalk

I think its a combination of less color detail and less shadow fill when you under expose a bit which helps to sell the look when you take it down even further in post to create a night time look. I've definitely fought with a too bright scene and learned my lesson. Shoot brighter than what you want the final scene to be, but not so much as to capture it as a properly lit daylight scene.

Then, should you wish, you can transform back to LogC, and, voila, you will have skin (or midgrey, or whatever reference you like) wherever you think it belongs on the LogC curve (yes, minus highlights latitude)..

Aharon and Mihail,

Can you explain why you're losing skin latitude or dynamic range in the case of "overexposed" ETTR? If you haven't clipped anything, and the sensor data representing that skin tone is linear, then wouldn't you have the exact same amount of data to work with on that skin tone independent of putting it just above middle gray or putting it two stops higher? No one disputes that it's an extra step for the colorist to compensate in post via a log offset or linear gain. I'm just confused as to why you think the ETTR approach compromises skin tones Aharon.

Thanks.

Hi Jacob

I am new to filming (1 year self taught hobbits), but have many years experience in print and digital reproduction and retouching.

The problem with ETTR in general is that you are always starting at the highlight first and even though it is linear not all things are created equal, that is the sensor registers more points lower down the line than in the shows.

You are therefore shifting everything down, no matter what the environment light condition is just to get yourself in a safe place. It’s not just one slide for the colourist to get the skin and everything else to line up. Every step you take in adjusting pixels in an image is degrading the quality. The slightest variation in the noise from one point on the linear curve to another another varies.

The fact is, if your lighting is correct and in relation to your other shots, you have measured it, set the right aperture, iso etc then your shots will match and all the colourist has to do is style it.

All of the above from aperture to light conditions have a profound effect on your image. ETTR as a systematic approach is just another way of saying I don’t know what my scene exposure is so I will just shuts it until I reach my safe point.

It’s ok to ETTR, capturing an image is capturing an image, but it’s not the right thing to do or something to be thought as a final solution.

Ask Roger Deakins if he ETTR on every shot.

I have lots of trouble with my exposure but persist in learning everything about light to hopefully get there in the end.

Chris

Note that Zebras are generated differently in different hardware. Some zebras indicate only Green channel clipping, some zebras indicate only average Luma clipping, and some may indicate any separate clipped channel.

Although ETTR was popular when the BMCC was introduced in order to “feed the sensor,” with the 4.6K sensor, I now use false colour for nearly all exposures and keeping light skin tones at or close to pink with shadows close to green. Sometimes I will intentionally shift one stop to the right if shadows might be a problem, but it’s working very well for me and is very easy to manage in camera and in post. I’d worry about going to extremes with ETTR both in terms of possible clipping of a single channel and that in a way, that’s a method that takes care of highlights at the expense of mid-tones and for me, good skin is paramount.

I’ve mentioned before that the assumption always seems to be the ETTR will feed your camera sensor the most light, but that is not always true. I’ve had many occasions where getting my skin to show pink or medium grey was actually feeding more light than ETTR. Of course then I might clip a cloud or more likely other strong light sources, but I’m personally okay with that if the highlight is not important content or dominating the scene. Just something to keep in mind.

It’s been a long time since I’ve had a problem where the shot was bad. With the range of the 4.6K sensor, even light skin at green is acceptable if it helps retain highlights though. Remember, you really can save 4-5 stops of exposure if you need to, but I prefer to keep skin within a stop of the correct exposure if possible.

I’m no expert so I know others may approach this differently in ways that they find are best for them.


Sent from my iPhone using Tapatalk

Agree with you Rick.

IMO,exposing for the skin tones with grey card is what i do on narrative film all the time.
FC is used to keep check on overall scene once i am confident that my skin tones are good and i adjust lighting for the rest of the scene where required if i needed to preserve shadow or highlights.

However if i am shooting scene where there is no people and i am only interested at best shot with all data, i would ETTR.

But before all of this takes place i actually trust my own eyes to light scene not relying on any light meters
or FC or ratios.

Continuity is obviously very important and if it is not done right on the day it is hell in the post.

So all these techniques are great at what they offer,FC is amazing tool but i would always refer to light meter during the shoot.

You can still "flood sensor" by using light meter or FC making sure skin tones are in right place and by lighting / defusing the rests in way you still ETTR but more of a controlled ETTR rather then guessing based on camera sensor.

What some members are rightfully saying with this 4.6 sensor it is safe to do so ETTR only as there is enough DR to "fix" skin tones in post, but that doesn't make job easier necessarily and it is dangerous to assume.

This is just my personal opinion and i appreciate that there is lot more experienced guys here so feel free to enlighten me if i am wrong, i take well constructive criticism and always keen to learn from others rather then my own mistakes.
Cheers.

Thanks, Darko. Your approach is commendable and infused with experience that trumps following any exposure rule rigorously.

One aspect that I like about using false colour as the basis for evaluating a ‘best’ exposure is that the light that is being measured is the light that falls on the sensor. When you use a handheld meter, there can be variations how you measure the light depending upon how and where you take your various readings. In some scenes it may not matter, but it could. False colour eliminates that possible variation.

To be fair though to the use of an exposure meter, the false colours are broadly brushed swaths of colour that paint an area of similar light. That may not be precise enough information for some folks. For me, it’s always well within the ballpark though and at times it is spot on for the shoot.


Sent from my iPhone using Tapatalk

We lose dynamic range by however many stops see ETTR and skin often uses a good chunk if the range above and below middle grey. When we take into account the need to leave some range for background brightness we can see where skin might end up in a flat less interesting place.

I'm still hoping Aharon and Mihail will chime back in on this. Is there a way in the forums to "tag" someone (e.g. like using the @ sign in other environments)? Or do you just have to rely on the fact that the individual is quoted to hopefully notify them?

They both refer to something negatively affecting highlight information (before clipping) and I'm not yet convinced as to what that something could be. Exposure "fix" via a gain change in linear or an offset in true log is not hard in post, and since the camera's not writing to a curve with a shoulder, it seems to me that exposing to the right would often improve noise performance without sacrificing anything major, *if* the utmost image quality is the primary objective. I think a lot of times it's not. An exposure process familiar to more traditional methodology for DP and crew, and one that is easy to monitor consistently, could easily be more important. What's important to a single operator who spends too much time optimizing his camera and reading internet threads (me) is often very different from what's important to a full-time DP producing long form narrative content.

As far as I see it, there can't be any curve inherent to a sensor. Photons hit each photo site and generate an electrical charge according to whatever its quantum efficiency is, right? These are all 'sensor-referred' values that treat highlights and shadows equally. That charge gets post-sensor amplified based on ISO across the whole luma range and A/D'ed into a digital value right? Now that digital value could be assigned to a curve in some cases where different portions of the luma range are allocated more or less bits for efficiency, but I've seen nothing related to the Ursa 4.6k that's proved the presence of any sort of highlight compression. (Incidentally, is this shoulder for a pleasant highlight rolloff always expected to come from the normalizing LUT then?)

I have heard the argument that increasing exposure (e.g. opening aperture) increases the amount of light gathered from a scene and therefore increases the dynamic range requirement to capture that scene. Under that circumstance I could see why some would say not to ETTR. Alister Chapman advocates this way of thinking and I respect him highly, but this doesn't seem quite accurate to me. Maybe I just don't understand it.

I think Rick's point is very important–that ETTR has a degree of subjectivity because you're deciding what highlights you're choosing to let blow out, and that could easily mean underexposing skin. Where I haven't yet been convinced of the danger of overexposure (before clipping in any channel of course) I do recognize noise in the shadows as a very relevant concern. Maybe we have something of a parallel to the audio world which so often precedes us. I used to hear so much more "get the signal as hot as you can without clipping" (audio ETTR), and now, with higher bit depth recordings you're much more likely to see your mixer give himself more headroom and expect the signal to be boosted. They just don't have to worry as much about the SNR with low signals like they used to. It seems to me that future video capture could take this route where a sensor and its A/D converter are of sufficient quality that we just capture a scene to some (hopefully efficient) format, expose coarsely to leave lots of headroom (like our audio friends) where everything but the brightest speculars are recorded and we pick our exposure in post.

I can see situations where ETTR could be a better option and where false color or metering for skin could be a better option. I suppose this thread's legacy is that it's helped show some of the pros and cons of each method. Just that I'm still not understanding the specific con of whatever detrimental thing is happening to ETTR highlights pulled down in post...

Jacob, seems like you’re ready to setup a skin test (shooting raw ISO 800 using the 4.6K sensor) perhaps looking for any differences in facial colour and tonal qualities when exposing for middle grey or false colour pink (or green as you wish as that’s middle grey) or full ETTR or something between pink and ETTR. The scene itself should not be dominated by extremely bright portions, just an average scene where the skin would normally be a mid-tone. Hope you have access to an interesting face for this test.

Under those average lighting conditions, ETTR results should be very good as you won’t be forcing your skin exposure too high and you should be feeding more light to the shadows.

If the lighting was such that your highlights were also your skin tones, that’s the point where there may be a difference in quality of your skin in post. I’m thinking if too much of the face makes up your scene’s highlights and you ETTR, the exposure-adjusted face in post might suffer.

At the other extreme, if your skin tones were your shadows in a scene with high dynamic range, using ETTR may not be the best approach either as the skin may suffer from a lack of light if you pushed it too far in post. Nothing wrong with much of a face being dark if it suits the scene.

Sent from my iPhone using Tapatalk

We lose dynamic range by however many stops see ETTR and skin often uses a good chunk if the range above and below middle grey. When we take into account the need to leave some range for background brightness we can see where skin might end up in a flat less interesting place.

I'm still hoping Aharon and Mihail will chime back in on this. Is there a way in the forums to "tag" someone (e.g. like using the @ sign in other environments)? Or do you just have to rely on the fact that the individual is quoted to hopefully notify them?

They both refer to something negatively affecting highlight information (before clipping) and I'm not yet convinced as to what that something could be. Exposure "fix" via a gain change in linear or an offset in true log is not hard in post, and since the camera's not writing to a curve with a shoulder, it seems to me that exposing to the right would often improve noise performance without sacrificing anything major, *if* the utmost image quality is the primary objective. I think a lot of times it's not. An exposure process familiar to more traditional methodology for DP and crew, and one that is easy to monitor consistently, could easily be more important. What's important to a single operator who spends too much time optimizing his camera and reading internet threads (me) is often very different from what's important to a full-time DP producing long form narrative content.

As far as I see it, there can't be any curve inherent to a sensor. Photons hit each photo site and generate an electrical charge according to whatever its quantum efficiency is, right? These are all 'sensor-referred' values that treat highlights and shadows equally. That charge gets post-sensor amplified based on ISO across the whole luma range and A/D'ed into a digital value right? Now that digital value could be assigned to a curve in some cases where different portions of the luma range are allocated more or less bits for efficiency, but I've seen nothing related to the Ursa 4.6k that's proved the presence of any sort of highlight compression. (Incidentally, is this shoulder for a pleasant highlight rolloff always expected to come from the normalizing LUT then?)

I have heard the argument that increasing exposure (e.g. opening aperture) increases the amount of light gathered from a scene and therefore increases the dynamic range requirement to capture that scene. Under that circumstance I could see why some would say not to ETTR. Alister Chapman advocates this way of thinking and I respect him highly, but this doesn't seem quite accurate to me. Maybe I just don't understand it.

I think Rick's point is very important–that ETTR has a degree of subjectivity because you're deciding what highlights you're choosing to let blow out, and that could easily mean underexposing skin. Where I haven't yet been convinced of the danger of overexposure (before clipping in any channel of course) I do recognize noise in the shadows as a very relevant concern. Maybe we have something of a parallel to the audio world which so often precedes us. I used to hear so much more "get the signal as hot as you can without clipping" (audio ETTR), and now, with higher bit depth recordings you're much more likely to see your mixer give himself more headroom and expect the signal to be boosted. They just don't have to worry as much about the SNR with low signals like they used to. It seems to me that future video capture could take this route where a sensor and its A/D converter are of sufficient quality that we just capture a scene to some (hopefully efficient) format, expose coarsely to leave lots of headroom (like our audio friends) where everything but the brightest speculars are recorded and we pick our exposure in post.

I can see situations where ETTR could be a better option and where false color or metering for skin could be a better option. I suppose this thread's legacy is that it's helped show some of the pros and cons of each method. Just that I'm still not understanding the specific con of whatever detrimental thing is happening to ETTR highlights pulled down in post...

BMD cameras follow the Red and Arri approach to ISO, which is to say it's just metadata... they don't apply any gain at record time.

The sensor itself records linear light, and the camera encodes it as log, which does include a form of highlight compression. It's how the camera provides 15 stops of dynamic range in a 12-bit payload without banding problems.

ETTR is in general not a good idea for narratives. I don't even use it in stills, yo be honest. It's just extra work, IMO.

In narrative, it's a very good idea to keep your skin tone exposures consistent throughout a scene. It saves work and it makes the scene a lot easier to grade.

The assertion that you're getting higher image quality by using ETTR is largely a myth, as long as you aren't letting the shadows fall into the noise floor.

Sent from my SM-G930V using Tapatalk

This is a great thread. I’m also not a fan of ETTR ever. I prefer a waveform over a histogram. False color is also a great tool.

Just going to add my view on this...it’s very simple: for quick documentary work I expose for skintones using a waveform and I do that with a mix of ISO, aperture, and ND. I shoot reality eith my DVX200 and that camera is quite fast with achieving the right exposure with that trinity of settings. For narrative work (read: projects where I have time to light properly for drama) I use a light meter to get the lighting where I need it for the settings I am already using in-camera.

Let’s say I am at ISO 800, 180° shutter, T2.8. I need my skintones to look creamy and natural so I get the light to work for me, making sure there is detail in the shadows and allowing for very dark dark areas or very bright bright areas if necessary or appropriate.

But, I never ETTR. I don’t ever use a histogram. Not as a videographer, not as a cinematographer, and not as a photographer. In fact, it was years ago working solely as a photographer where I learned ETTR-ing was not the proper solution for anything. Maybe it’s a last resort sort of thing, but it can’t be true all the time.

There are better methods. Zebras at 95% is one. I like zebras at 75% for skintones in Rec.709 and I leave the zebras on the face. Just a touch. If I close the iris a smidge they go away, so I make sure they’re just there and then I know the face is at the proper IRE for that scenario and I can turn them off if they bother me.

TL;DR:
So, quick reality/ENG work: zebras at 75% for skin. Narrative work: light meter, waveform, false color. ETTR never.