I'd like to invite those using S-Log to share their exposure methods in this thread. For many of us S-Log is a new tool and I'm curious as to how people are choosing to expose when shooting with log.

There appear to be several different approaches which are in common use.

One is to use a light meter and expose using an 18% grey card. This should give you around 7 stops over and 5 stops under.

One is to use a chip chart with mid grey and white chips and expose so grey is at 38% and white is at 68%. This method is supposed to then allow for highlights that are brighter than white. This method works well in controlled lighting situations, but can also be waste a lot of available DR when there are no highlight issues above 68%, or perhaps your highlights are only reaching 90% (remembering that S-Log records up to 109). Opening up a little would bring out more shadow detail giving more to work with in the grade.

The method I have been using is to use a WFM or the Histogram and to look at the total tonal range of the scene. Then generally I will be exposing so that my highlights are between half and a full stop under clipping (this allows for specular highlights that may not be visible on the histogram). Now looking at the monitor I will look at shadows, faces etc and ensure that they are not under exposed, I like to have faces between 40 and 60 IRE depending on the mood of the scene. Clearly if the scenes range exceeds that of the camera then I have to either under expose or allow the highlights to go over. In these cases I will aim for mid grey at around 38, but allow it to go higher or lower depending on the scene.

I'm not convinced about setting mid grey at 38% and then expecting/hoping for everything else to just fit within the cameras DR. Nor do I want faces and similar tones to be greatly underexposed as even though I might be recording uncompressed to a 10 bit codec, I don't want the key parts of my images getting grainy due to stretch/lift in post (EI S-log will be better in this respect).

So what are other doing and how does it work for you?

Alister my deepest respect for your encyclopedia of knowledge and insightful technical analysis on all topics. I came from a film background where the medium was kinder to highlights and went though some rude awakenings during the HD learning process. Having been a slave with the WFM, I've gone back to exposing with an incident meter where I feel free again to judge the emotional impact of any given exposure setting first, before getting bogged down with numbers and clipping points. I just upgraded S-Log and am planning next week's shoot by establishing a working ASA with my colorist shooting color charts & gray scales plus to minus 5 stops, grading them back to normal and pick a forgiving and flattering setting, then expose key subjects (mostly faces) with incident meter based on aesthetic decisions whether subject should be at key value or, over and under. (Still a stickler with my gaffer in maintaining the film approach that 100 foot candles is needed to expose a gray card at T2.8 at ASA 100 and 1/50 sec exposure) I balance the rest of lighting by eye, trusting that ASA 800 at T2.8 is very close to WYCWYG, then finally check highlights and shadows with a spot meter and deal with the video tech who's probably screaming by now. This method may sound antiquated and I have much to learn from your suggested approaches, but I need to avoid the chicken and egg issue with going back and forth from waveforms to set to constantly adjust lighting, and compromising the spontaneity and inspiration on why I choose to light a certain scene a certain way.

Thanks for the input Brian. Your approach is similar to others I hear of for S-Log and with all these things, if it works for you then go for it!

I'm really just curious and trying to gather as many different approaches as possible, part of the reason is that I want to prepare some training materials and I don't want to just put over my approach but offer a variety of valid approaches. There may also be times where my preferred approach does not work so it's good to know what other methods can be used.

My WFM is built into my Transvideo 6" monitor which I normally have mounted on the camera, so I can see instantly any lighting or exposure adjustments. Very often I'm shooting projects with only available light so I'm not adjusting lighting, primarily just adjusting exposure and maybe a couple of reflectors. I would not call myself a slave to the WFM, but it is how I'm measuring my exposure and then looking at the scene on the monitor for problem areas.

@alister
I'm upgrading my TVLogic monitor to show wave forms now that I have S-Log. Having only used histograms and zebras before I'm wondering if you can give any tips or instructions on using the WFM. Especially as the histogram and zebras are off when using S-Log.

In the mean time I've needed a field expedient method of exposure. It seems with S-Log instead of 70% zebras on faces it should be about 40% and the F3 only goes down to 50%. At this time I've reset zebras for 50, expose so faces have a hint of zebra, and then adding a ND.3 filter to my matte box. I guess I'm using the zebras as a spot meter. Workable but clunky.

I think the v1.3 firmware will solve many of these issues, but in the meantime any tips on the WFM would be very welcome.
Duke

TV Logic false color can be adjusted down so that it looks to me like it could be useful for sLog. I wish it was easier to turn on & off though

Don't forget the F3 does have a spot meter. You can turn on the level reading in the display menu and then it will read the brightness level for whatever is in the centre marker box. I often use this to spot check parts of my scene.

Setting the zebra to 50 means that you will start to see the pattern when the luma level reaches 45IRE. There is a 10% window, so zebras will be present +/- 5 of the nominal set point.
I think (my view) that 40% for faces is a little low with S-Log. A common Log practice is to set mid grey at 38 IRE or 40% near as damn it which then puts white at around 68. Part of the reason behind this is because video cameras respond better to underexposure than overexposure. By putting white at 68 you have plenty of headroom to cope with whiter than white, or highlights, specular's, sky etc. 38 is also the level normally used by still photographers and film users for faces, but conventional video gamma tends to use very much higher levels. If you are not having to deal with highlight extremes, like shooting on a day like today in the UK where there is a heavy overcast and flat light, then you can afford to record at higher levels than these which will give you more shadow information and less noise after grading.

That's why I choose to allow my skin tones to ride between 40 and 60 depending on the overall scene, and in particular highlights.

A waveform monitor is similar to the histogram. The histogram gives you the amount of a certain luma level vertically and the luma level horizontally. A waveform monitor shows the image brightness on the vertical scale. the horizontal scale is in effect a slice through the video frame, os objects left of frame will be to the left of the waveform. You should have a vertical scale for the voltage level going from below zero up to 110% or 7.7v (100% = 7v). The thing I like about a WFM is that it's easy to correlate different parts of the image to the waveform which makes it easier to workout what level relates to what part of the picture. I find it hard to see exactly how far I am from under/over exposure with the histogram, but very easy with a WFM. You can also see how much any parts of the dynamic range are being compressed as you will actually see the levels stretching and contracting as you open and close the iris.

Subscribed. I can't say that I've got enough LOG experience to share yet but I'm working on it. :)

Don't forget the F3 does have a spot meter. You can turn on the level reading in the display menu and then it will read the brightness level for whatever is in the centre marker box. I often use this to spot check parts of my scene..

Am following this thread with interest, and have a question: how do you engage spot metering on the F3? (I don't own one - yet)?

Thanks.

Its a setting in the display menu. You can turn on/off individual info objects on the monitor (including the spot meter).

Thanks Chris. Can you set a programmable button for this?

Any of the 8 assign buttons can be programmed for it. By default, Asiign Button 2 is set for it from the factory.

Thanks Doug. If I buy the camera my first purchase (even before I get the camera) will be your training video. I have your Canon XF300 one and found it very helpful.

The spot meter is really useful and very accurate. I'll often pan around the scene quickly using it to check any possible problem areas. It's particularly useful with S-Log where the Zebras don't really go down far enough if you are using any of the mid grey at 38% methods.

I think the key thing to remember with S-Log is that it grades both up and down very well. You can bring shadows up, but also take highlights or even mids down. Grading down will reduce noise, stretching and lifting will increase it and I think this has to be considered during your exposure choices. You also have to remember that if you have LUT's turned on all the built in exposure tools are offset by the LUT so are no longer accurate. I'm lucky enough to have Transvideo monitors with WFM's so where possible I'll use these, I can also apply LUT's on the monitors. The Convergent Design Gemini and Cinedeck can also apply LUT's so with these you can leave the LUT's off on the F3 and then the exposure tools are accurate.

Would be great to get some input from a colourist on this topic.

Hi Allister,

My method is mostly same as you.
Altho in the times before HD, I used to use lightmeters, but since I got an HD-SDI monitor with waveform in 2006, my lightmeter always stays in the closet.

I am a colorist too, so I take extra care at shadows to lower noise while preserving detail.

Back in the days of F350, I notice there's noise in IRE 20% down, especially prominent in night and lower shutter speeds. There's even more noise if using tungsten lighting - with corrected kelvin.
So based on this knowledge, I take caution how to manage exposure in the blacks, how many noise is OK or not OK.
I will then adjust ambient lighting to acomodate if there is much noise in the blacks. Of course, I will look at waveform and zoom in to certain part just to see the IRE value. But now in F3 it is so much easier.
Also using F3, the black noise is soo much lower..I still fiddling with my F3 - so Im yet to decide the acceptable black noise %.

I also put skintones in 35-60%, depends on mood, and depends on actor's race as different race has huge difference in skin color.

While for highlights, I also take extra care to preserve as much detail as possible.
I use waveform and zebra. I put zebra at 100 and play with the iris to see how much detail is clipped.
Now with F3 it is very convenient to see IRE % at certain parts just by slightly zoom in and pan to that part. In F350, I dont use -3dB as it lowers whole signal including the top whites - clipping it 3dB. So I will use ND filters.

As for F3, I'm still studying it's workings, but in short study, I keep highlight detail much more than F350, because F3 is much less noise. I record F3 using 10bit recorder - so in grading I can lift the blacks as much as 2 stops - not more, if needed. I keep this in mind while shooting the picture - how much gain will you do in grading but still get acceptable noise.

I also play around using the shutter speed. Most DP I know dont like playing with shutter - they use mostly 1/50. But I like to play from No Shutter to even 1/1000. I always use high shutter speed for high-action or war scenes - in day times of course. In nights where lighting is not optimal, I use no shutter (1/25). Even at times, when lighting is very poor, and the scene is only actor talking with little movements, I use slowshutter 2 frames - it doesnt strobe when movement is minimal - and I can get 1 stop brighter and cleaner shot! - for still establishes at night I have used 2 seconds slow shutter!! - I also use OFFshutter to clean small rain drops - when there's very light rain but director dont want the rain to be captured.

I never use effect filters as most can be done better in grading/post - but I use ultra-polar a lot to get cleaner atmosphere and vivid outdoors - also as a slight ND.

Regards,
Hornady.

The "how much noise will my grade add" question will be a lot easier to deal with with the new EI S-Log firmware which adds gain to the LUT outputs while leaving the S-Log out intact. This allows you to get a feel for how much grain lifting the image by one or 2 stops in post will make.

One thing I've discovered is that if you record S-Log to ProRes you will loose potentially loose half a stop of DR. The problem is that S-Log can go up to 109IRE while ProRes clips at 104. So if your recording direct to an external ProRes recorder you need to keep your highlights at least half a stop below the F3's clip point. If using the Convergent Design Gemini then when you transcode from the DPX files to ProRes you have to do a slight level reduction.

NOTE Since writing this I discovered that ProRes is NOT clipping on encode, the files go up to 109%, but some Quicktime apps will clip the files to 104 on playback.

For you guys that have shot with this cam before, would I be correct in saying that I would be getting the full DR of the camera while capturing to .DPX with my decklink quad? It doesn't sound like that much of a clipping issue going from 104-109 IRE, but when shooting in harsh contrasty additions, Ill deal with the extra size of DPX for that extra bit of latitude in post.

104 to 109 clipping isn't much but it is almost 5% of the total range, which if we assume to be 12 to 13 stops is over half a stop.

DPX should be OK. Most recorders and codec will allow at least 108%. The 104 clip is a ProRes specific issue as according to Apple they want to ensure that users of ProRes stay within the limits normally specified for broadcast TV, which is a nominal 100% maximum with overshoots to 104 allowed.

Although I haven't seen it yet AJA said they were going to give a free firmware upgrade that included the ability to record to DNxHD. I'm looking forward to getting that extra bit of DR back.

OK, I wrote about the clipping issue without digging deep enough. If you read the original post about clipping I claimed that ProRes was clipping my files at 104%. Well it's NOT. The ProRes files are just fine, BUT some Quick Time applications are clipping the files at playback. In FCP the files are OK. Premiere appears to be reducing the level of the files a little and Quick Time player is clipping the files at approx 104. So this isn't as big an issue as I thought, but you do need to keep an eye out as to what is happening with highlights and super whites depending on what software you are using. I was wondering why I hadn't seen this before. In part it because I am no longer using FCP.

OK, I wrote about the clipping issue without digging deep enough. If you read the original post about clipping I claimed that ProRes was clipping my files at 104%. Well it's NOT. The ProRes files are just fine, BUT some Quick Time applications are clipping the files at playback.

This is the same phenomenon as Canon Quicktime files clipping, etc etc. Quicktime re-wrapped XDCAM native files do it too.

In the header of each Quicktime file is a set of metadata that instructs the decoder how to render the file, if the decoding program chooses to heed it. Display resolution, contrast, saturation level, etc etc.

Anyway, with that info, you can see how all sorts of wacky things that can happen from program to program.

In any case, i've found that any highlights "clipped" in this manner are recoverable in almost any color correction program, and most NLEs. When working in Davinci with Quicktime files, I routinely check the material for this and often I can lower the gain wheel and watch highlights come back on the scopes as I dial it down.

I was wondering why, as a big user of ProRes, I had not seen this before. I've stopped using FCP and am flipping between Avid and Premiere and other apps as well as using Quicktime Player a lot more and this is where I'm seeing the issue. As you say if you bring the levels down, all is the data is there.

That will teach me to not investigate properly. Should have gone back to FCP and checked sooner, but I'm trying to force myself to learn the other NLE's rather than falling back on "old faithful" all the time.

Don't be too hard on yourself Alister. Your attention to detail has saved hundred's of us from needing to bother at all. Thanks for all that work and willingness to share the results.

I was wondering why, as a big user of ProRes, I had not seen this before.

It's not even a ProRes issue, per se. More of an issue with the interactions of the device/software that generated the Quicktimes, and the program reading them.

First time I saw it was with 5D files.

Thank you guys,
your findings are truly insights that even Apple, Sony, etc. didn't see, let alone me ordinary user.
Very valuable information.

Now, is there info on service menu in F3?
In F350 I can display gray ramp and see the gamma curve in waveform monitor. Can we do this in F3?

Hornady,
I do not believe a Test Saw option is on the F3.
On the 800 it is in the Paint Menu

OK, I wrote about the clipping issue without digging deep enough. If you read the original post about clipping I claimed that ProRes was clipping my files at 104%. Well it's NOT. The ProRes files are just fine, BUT some Quick Time applications are clipping the files at playback. In FCP the files are OK. Premiere appears to be reducing the level of the files a little and Quick Time player is clipping the files at approx 104. So this isn't as big an issue as I thought, but you do need to keep an eye out as to what is happening with highlights and super whites depending on what software you are using. I was wondering why I hadn't seen this before. In part it because I am no longer using FCP.

Avid Liquid would do the same thing. It clipped any imported video at 100% but if you captured video via firewire (or SDI) the limit was 109%. I ran into that problem when I started capturing HDV video onto hard drive recorders and bringing it in as files. All the highlights were harsh and crunchy.

Luckily that hasn't been the case with Media Composer.

Hornady,
I do not believe a Test Saw option is on the F3.
On the 800 it is in the Paint Menu

There is, I've used it. It's in the service menu.

There is, I've used it. It's in the service menu.

Hi Nate,
Please give us some insights how to enter service menu in F3? :)

Thanks!

There is, I've used it. It's in the service menu.

Nate
Thanks. Sony is a bit hesitant to provide me with the special handshake. Understandable.

I just checked some footage and I don't see anything different between XDCAM EX and Pro Res in Premiere Pro CS5. (recorded to a PIX240 at 0 or 6db).

XDCAM on left and Pro Res on right.

Neither is reaching 109 IRE on the scopes. I expect if you looked at the clips in FCP you would see them reaching 109. They should be going up to 109. As Nate pointed out it's a wider issue with some types of quicktime files than just just ProRes. The reason I'm seeing it now is because I've switched away from FCP and I'm noticing different levels and clipping in different platforms which should not be the case.

I was running some tests over the weekend with the new 1.2 firmware and EI S-Log, and I noticed that the SDI output to a Cinedeck and Sony's pro 24" and even a Panasonic 1760 that the IRE was cutting out at 106 on every waveform. Seems like a F3 problem. No LUT was applied and the dual-link was going into the Cinedeck and the SDI out into a monitor (using EI 800).

Another thing to check for in the next couple of days. I'll get back to you Steve as this might have consequences for anyone shooting 422 S-log using the monitor out.

Alister,

FWIW, I am definitely seeing massive ProRes differences between FCP X, FCP X, PPPro and MC 6. I had noticed prior to 1.2 a clipping around 109-110 at ISO 1600 so I am not surprised that you are seeing clipping at values equal to or approaching that original 109 reading.

Ned Soltz