Cinematography Mailing List - CML
    advanced

Judging RED Exposure

 

I just got done shooting a loooong coooold weekend with the RED. We are finishing up a short. Though the entire shoot which has taken 9 total days, I have been going back and forth with the director about exposure. My question is this. What is everyone else trusting the most. The EVF image looks different from the RED LCD on-board, which is different from the 17 inch HD monitor. Then the stop lights and the waveform can contradict all the monitors. Let’s not forget "preditor vision" as we call it...the color reference. I was mainly exposing from the on-board while lighting and then checking the preditor vision and fine tuning from there. Many times the monitors looked over and the preditor vision showed me I was more average or under exposed. Just wondering what everyone else was using for an exposure workflow.

Thanks

Mike Gillis
Milwaukee
414-303-3890


Hi Mike,

I've only used RED a little bit, but after talking to a number of people and taking in a lot of advice, I like to view in RedSpace, with RAW at the touch of a button, and also use the false-color exposure "predator" mode that you mentioned. Additionally, I have zebras, but don't trust them as much. The RGB histogram is also good, kind of like my DSLR that I'm used to, but I found the RGB traffic lights to be nearly useless.

I'm not sure how to calibrate the monitor properly without color bars, and obviously you can mess with brightness, contrast, chroma and phase without knowing where to put it. I finally decided to just dial in a look that I like with those basic controls, and go from there -- usually it's means bringing up contrast and lowering chroma and brightness.

Here's another question to add to the mix... I noticed that RedSpace's colors are wacky. People have noticed a real sensitivity to Green and Magenta with the RED, but recently I had a lampshade that was deep brown to my eye, but bright purple on the LCD and monitor. I haven't seen the transcoded rushes from that yet, so I'm not sure if it will actually come out this way, but I've seen several instances where certain colors went crazy. I'm wondering if in this case it was an IR issue, since the lamp was on, or if it's a wide color-space thing.

Best,

Graham Futerfas
Director of Photography
Los Angeles, CA
www.GFuterfas.com


>> I've only used RED a little bit, but after talking to a number of people and taking in a lot of advice, I >>like to view in RedSpace, with RAW at the touch of a button, and also use the false-color exposure >>"predator" mode that you mentioned.

I like working that way too, viewing in REDSpace and toggling into raw occasionally to check highlight exposure. I don't like "predator" mode but I do use the zebras a lot and I trust the stop lights somewhat.

Before each shot I weigh a few different things :

(1) With the zebras set at a low of 103 and a high of 109, I set the exposure so that the zebras just bite into the highlights I want to keep.

(2) I then check the histogram and see what it says. I also look at the stop light. If I open up until I clip a channel on the stop light I check to see if the histogram agrees. If I'm clipping one channel on the stop light I don't worry about it.

(3) If I clip two channels I drop into raw mode to see what's really going on. More often than not all the clipping goes away as soon as the REDSpace gamma curve is removed.

(4) I prelight with my meter set at 160. The exposure from there can be flexible, but 160 seems to be a nice sweet spot to push the highlights into the top of the useable part of the linear gamma. (At least at 160 you won't find yourself falling short of exposure.)

And I always keep in mind that no matter what I see on the monitor, the viewfinder or the LCD, it always looks better in RedCine.

If all else fails I can typically dial in an exposure by eye in REDSpace and get within a half stop of what it wants to be according to how I read the histogram/zebras/traffic lights. I can usually get to within a quarter stop using a more common HD camera (F900R, Varicam).

I haven't had any really crazy color events yet, other than some mild magenta/green issues. Were you balanced for tungsten when you shot the lampshade? If we think of brown as a mixture of red and green, and purple as red and blue... I wonder if something happened to either dramatically reduce the green or boost the blue channel. My vote is for reduced green, although intuitively it seems unlikely as green is the color around which all else in that camera revolves.

Art Adams | dp
san jose | ca | usa
415.760.5167
www.artadams.net
http://provideocoalition.com/index.php/aadams


The stoplights take into account the full sensor (including look around) where you often have backlights sitting at the edge of frame, triggering the stoplights when your frame is within range.

Florian Stadler, D.P., L.A.
www.florianstadler.com


Lighting based on the monitor on the RED is like lighting a film shoot based on the videotap. It is only a general representation of the image, quickly de-Bayered and downconverted to a lower resolution to give a reference of the image. You can safely say that if you've got it in the monitor than you certainly have it in the RAW file, but you probably have a lot more. The same is true for the Phantom or any other RAW capture camera. Best way to expose it is like film -- do some exposure testing beforehand and then trust your meter.

Mitch Gross
NYC DP
Applications Specialist
Abel Cine Tech


>> Best way to expose it is like film -- do some exposure testing beforehand and then trust your meter.

I disagree somewhat. I try to set the exposure on a shot-by-shot basis using the full range of the sensor, as dictated by what the various built-in measuring tools are telling me. You can use your meter only, and you'll probably get more consistent exposure, but depending on the dynamic range of your scene you may find a fair bit of it a little too close to the noise floor.

-----------------------
Art Adams | dp
San jose | ca | usa
415.760.5167


Mitch Gross wrote :

>Best way to expose it is like film --do some exposure testing before hand and then trust your >>meter.

Mitch,

My issue with the statement that a number of guys I work with like to open up to get a "meatier" neg, and I feel that RED content needs to be exposed with the same care that a Reversal stock is. Over exposure is a sure way to screw up with RED, especially IMHO if you are working in anything other than a DPX workflow.

I am also a fan of using 200-250 as the base ISO, I like monitoring in REDSpace but I am intrigued peoples use of a standard LCD monitor, have any of you had any issue with trying to monitor a "raw" feed on a REC709 display?

I prefer to monitor with the Speedgrade calibration via a Cine-Tal, with the built in calibration between the app and the hardware I feel that offers me the best of both. I have also found that the color syncing within Speedgrade maintains the correction on the display as long as the app is still open, allowing me a chance to jump into REDalert and check my test corrections on the Cine-tal then switching back and forth from DVI to HDSDI to check the camera feed.

My one warning is to not use a cheap DVI cable with this setup, I have seen a couple of issues that were directly related to a poorly shielded cable .

Gary Adcock
Studio37
HD & Film Consultation
Chicago, USA


class="style18"
>>my issue with the statement that a number of guys I work with like to open up to get a "meatier" >>neg, and I feel that RED content needs to be exposed with the same care that a Reversal stock is. >>Over exposure is a sure way to screw up with RED, especially IMHO if you are working in >>anything other than a DPX workflow.

I agree, although I would still say that this falls under the heading of "do some test exposures first." I remember the first time I shot reversal and wanted to put a highlight on a clock so I snooted an inkie and pointed right at that clock. Got the footage back to see a perfect white circle on the wall where the clock had once been. Learn how to use your meter in relation to your stock or sensor or whatever you are capturing an image with and you will then know where you stand in your exposure.

Seriously, how many of us have prelit a scene by eye before there was ever a camera around? On some really big setups the design of the lighting has to be in place long before something with a sensor other than your light meter is around. Get to know how it reacts and you're good to go. Think what it must have been like decades ago with no videotaps and stocks that couldn't handle the over/under we take for granted now in film. Yet they still got gorgeous images with a Spectra and a Weston.

Mitch Gross
NYC DP
Applications Specialist
Abel Cine Tech


Thanks for all the chatter about judging exposure. I guess I will know how I did in a few months when we get to looking at everything Redcine.

I have shot with RED for a few projects now and the main complaint from me is the EVF. It is great to look at a large sharp color image, but it comes with a few negatives. First I found that when hot sections of the frame were covered up by actors and then revealed again, the image would milk out. After some testing and experience it seems to only be happening in the EVF...not on the on-board or in any of the final footage. It can be distracting when operating. I also had trouble with the flicker image. It was worst when shooting a scene with candles in the foreground. The whole image was flickering with the candles. We also determined it was only in the EVF. It was set at the wrong refresh rate.

One other helpful tip. When powering up after a battery change we had a white screen on the back status screen and it wouldn't start up. This happened to us only twice. Both times the camera would not start up after repeated attempts. We were told to hold the power button in at start up for 10 seconds and then release it. I don't know why, but this worked. Good to know if you are in the field and have this issue

Other than those minor things the camera performed great. We had it under very cold conditions one night and had no issues. I will try to link you to some scenes as we get them finished. It is a short about a female serial killer who targets classical musicians called "The Violinist".

I must also say that we had great technical support from North American Camera in Milwaukee. Bob has 2 REDs and Milwaukee is very busy shooting with RED.

Mike Gillis
Milwaukee
Disclaimer: I don't work for North American Camera, but I do owe him some favors.


"I've only used RED a little bit, but after talking to a number of people and taking in a lot of advice, I like to view in RedSpace, with RAW at the touch of a button, and also use the false-color exposure "predator" mode that you mentioned. Additionally, I have zebras, but don't trust them as much."

The has been my standard operating procedure.

However, I recently found out through cryptic posts on REDUSER that when you have selected "view=RAW", false color (or predator mode) is not accurate. Apparently it is not calibrated for RAW view. That said, it is supposed to be only slightly inaccurate, but the scary quote I got from someone at RED was "be careful" when using false color and RAW view!

The new "color bar" feature apparently constantly monitors the RAW signal. The histogram and RGB stop lights ARE accurate when view=RAW.

Hope that helps.

Dylan Macleod, CSC
Cinematographer
Toronto, Canada
www.dylanmacleod.com


>when you have selected "view=RAW", false color (or predator mode) is not accurate.

I'm not sure how that works out; prior to build 17 I found it to be the ONLY reliable metering method (that, and the RGB parade in raw mode). When the false color hits red (appropriately enough), you are starting to clip.

In the non-raw modes there's a purple overexposure indicator (which at ISO 320 gives about half a stop of warning before real damage is done), but it only works when the color temp is set close to 5000K. Move below 4100K or above 6300K, and you'll never see purple no matter how overexposed thing are.

>>The new "color bar" feature apparently constantly monitors the RAW signal. The histogram and >>RGB stop lights ARE accurate when view=RAW.

So far (in build 17) the bar graph seems pretty good, but I still don't trust the traffic lights. Too often in build 16, I've had a light come on as I open the iris, then go off as I open it more, then come on again. Haven't seen that behavior yet in b.17, but I've only had it loaded one day.

There's also a raw histogram, which is shown superimposed on the current view LUT's monochrome histogram. The raw histogram seems to be truly raw: no gamma curve applied! It pretty much sits lumped up on the left side of the scale, until you get a hotspot, when it starts to spike on the right side.
As to REDspace "looking like ass", Art Adams says (and I tend to agree) that ALL the viewing modes look horrible, compared to what comes out of REDCine or RED ALERT.
Dane, if you have more info on the genesis of REDspace, please share.

Adam Wilt / filmmaker, Meets The Eye / writer,
provideocoalition.com / Mt View CA USA




Copyright © CML. All rights reserved.