ASA rating and meters

[Jesper Holmström]

During our tests we have noticed that when you set the camera to 320 ASA
and expose a graycard/grayscale using falsecolor or histogram our light meter reads 160-200 ASA.
When I look at the preview signal it looks a little to crushed in the shadows compared to the R3D file.
If I adjust contrast in camera to match what I see in the R3D file (-0.5) the Camera meters are affected and my light meter now reads 320 ASA.

How are we going to be able to use false color and zebras together with a look if they affect the reading?

Is the default setting of brightness and contrast in camera really my "flat" image I should be measuring from?

Jesper Holmstrom
Dagsljus Filmequipment
Stockholm, Sweden

[Fergus Meiklejohn]

Do you think what you are seeing could be contribute to macgregor's "Issues" thread?

[Kevin Halverson]

I believe that there is a built in .4 stops of 'headroom'. So, if you take an ISO rating of 320 and multiple it by -.4 stops, you will get ISO 243. This is closer to your findings.

320 * 2^-.4 = 320 * .758 = 243

You can either take the 'headroom' into account in your metering methodology or not as you see fit, but knowledge of its presence can be useful.

[Kevin Halverson]

Do you think what you are seeing could be contribute to macgregor's "Issues" thread?

Very possible. See my posts in that thread if your interested.

Kevin

[David Mullen ASC]

Unless you are exposing grey scales and chip charts in flat lighting, there isn't really a right or wrong exposure for most subjects, particularly one shot at night -- there is some leeway in either direction depending on the look you want. Histograms and false color information are more about giving you ideas about where to place an exposure, not so much about getting the perfect exposure.

[Kevin Halverson]

Unless you are exposing grey scales and chip charts in flat lighting, there isn't really a right or wrong exposure for most subjects, particularly one shot at night -- there is some leeway in either direction depending on the look you want. Histograms and false color information are more about giving you ideas about where to place an exposure, not so much about getting the perfect exposure.

I fully agree, I never intended to convey that there is but one 'correct' exposure for any given scene. The shooter is always allowed to place their image anywhere within the exposure range that a given system can capture, but if your unaware of the 'headroom' you might not realize that you are under by this factor. In many cases, an inexperienced shooter is better off not to know about the headroom as it will protect them (to a degree) from their own choices. However, in the case where someone is using a metering methodology, knowledge of this 'bias' could be useful information.

Kevin

[Jesper Holmström]

18% gray in 709 gamma corresponds to 41 IRE. If I follow that rule I rate the camera 160-200 ASA.
If in fact the default contrast in the camera is to high and I adjust the contrast to reflect that, I rate the camera 320 ASA.
If the preview signal is corrected to look good on a monitor then that affects histogram, zebras and false colors as well.
What I'm looking for is to find the correct "base" for the camera and a way to accurately measure the signal not the look.
After that it is of course a creative decision on where to let your exposure fall.
Jesper Holmström
Dagsljus Filmequipment
Stockholm, Sweden

[Kevin Halverson]

Hello Jesper,

Please feel free to PM me if you would like to discuss this in detail.

Kevin

[Joel Kaye]

If the preview signal is corrected to look good on a monitor then that affects histogram, zebras and false colors as well.


You may as well discuss it here Kmuse because if the above statement is accurate then that seems a little scary to me. I would think the histogram/false/zebra should be reading before any preview monitor correction since those are just XML changes anyhow and the sensor isn't changing.

[Arnaud Paris]

If you guys exchanged further on that topic thru PM; can you share your conclusions here on the forum with everybody.
I think a lot of people are indeed wondering how to set their exposure since we know that the signal from the monitor is not a direct result of the image being recorded thru the redcode raw compression.

Thanks

Arno

[Kevin Halverson]

You may as well discuss it here Kmuse because if the above statement is accurate then that seems a little scary to me. I would think the histogram/false/zebra should be reading before any preview monitor correction since those are just XML changes anyhow and the sensor isn't changing.

Hello joelnet,

I will be happy to continue the discussion here if you wish. Please clarify your exact area of interest/concern and any questions/comments that you might have.


Kevin

[Joel Kaye]

I will be happy to continue the discussion here if you wish. Please clarify your exact area of interest/concern and any questions/comments that you might have.


I guess my concern is if you changed the ASA to something high just to get a good look at a scene on your monitor that false color might indicate the wrong exposure since the real exposure is always based somewhere around 320.

Or to rephrase - if you raise the ASA in camera does false color indicate highlights are blowing out (at assumed Redcine ASA) when in fact the camera is capturing those highlights (actual ASA of around 320)?

Might 'cause you to drop exposure when you didn't need to.

[Kevin Halverson]

The last testing that I did was at the end of September and this behavior is certainly under the control of the firmware, so the question is somewhat 'build' dependent. I will check with an early camera owner and confirm if this is the current behavior. Give me a day or so and I will get back with a confirmation.

[Emmanuel Cambier]

I just have a hard time believing that the false color and zebras would be affected by changes applied to the monitoring (contrast, brightness…).
This wouldn't make sens, since those settings are not baked in but your exposure is.
Anyone could confirm this behavior?

Emmanuel

[Billy Summers]

exposure is baked in, but not ASA...

[Alexis Vanier]

(Thinking out loud)

This means that the in camera "exposure lookup" is based on the signal to be sent for output? So, if you were to load a LUT in camera, that would affect how the zebras and false colors react since the metering is done after processing.

Then, the only way to get a very precise exposure rating would be to load the footage into say redalert (or redcine) and check against log gamma (to see your absolute position to the total camera dynamic range), then taking into account the size of your usable dynamic range depending on the iso you plan on applying in post (since there's no in-camera actual signal gain)... taking your final format.

(Did I get this right?)

We're really back to the film educated guess workflow... I like it.

[David Mullen ASC]

You'd think that the clip point is the clip point no matter what ASA setting you use, i.e. a false color reading showing clipped areas must always be accurate.

[Emmanuel Cambier]

You'd think that the clip point is the clip point no matter what ASA setting you use, i.e. a false color reading showing clipped areas must always be accurate.

Exactly
This is precisely the question that needs answering, are false color and zebras affected by changes made to monitoring settings?
And clearly it shouldn't be so.:help:

Emmanuel

[Kip Hewitt]

The last testing that I did was at the end of September and this behavior is certainly under the control of the firmware, so the question is somewhat 'build' dependent. I will check with an early camera owner and confirm if this is the current behavior. Give me a day or so and I will get back with a confirmation.

Hey Khmuse,

It sounds like you need a camera to test with. I would be happy to donate mine to your cause for a couple of days if you like. It looks like we are both in OC so it shouldn't be any hassle.

PM me and I will send you my info.

[Jesper Holmström]

The monitor output corrections does affect zebras, false color and histogram, and maybe it should, especially for ASA settings, as long as there is a quick way to revert to a flat technical image for measuring.

Jesper Holmström
Dagsljus Filmequipment
Stockholm, Sweden

[Miguel "Macgregor" De Olaso]

ISO, ISO, it´s called ISO, not ASA anymore...

BTW, when i was working with a lightmeter with the RED i was getting confusing readings. As i once posted, my readings set the camera at 160ISO at the Salt Flats in Utah, but again we were working in RAW, so you might be better preserving highlights by rating it at 320. More tests should be performed anyway.

I threw the lightmeter at the end of the day. It was useless.

[Greg M]

ISO, ISO, it´s called ISO, not ASA anymore...

BTW, when i was working with a lightmeter with the RED i was getting confusing readings. As i once posted, my readings set the camera at 160ISO at the Salt Flats in Utah, but again we were working in RAW, so you might be better preserving highlights by rating it at 320. More tests should be performed anyway.

I threw the lightmeter at the end of the day. It was useless.

I'll take it...send it to me.

[Jendra]

Exactly
This is precisely the question that needs answering, are false color and zebras affected by changes made to monitoring settings?
And clearly it shouldn't be so.:help:

Emmanuel

I agree, it shouldn't be, but it is. False color (and the other exposure tools) are not an absolute for the camera. Being that we are working in RAW, I would prefer tools that are an absolute of what the camera can and can't handle. That's a little subjective on the shadow side (based on your personal tolerance for noise) but as far as clipping, when it's gone, it's gone!

Unless something has changed in this regard, when I shot in October, the false color settings, waveform etc. all change based on the ISO. Therefore if you were rating at 500 ISO and you were clipping (red in false color) and then turned your ISO down to 320, you wouldn't be clipping anymore.

So false color doesn't tell you when the sensor clips, just when you are hitting 100% at your working ISO. This is why it is so important to not shoot at 100 ISO because then you are just overexposing the camera and you can be reaching sensor clip (aka: full well) without even being told you are clipping at all!

The false color (and waveform, zebras etc) are all based on the viewing channel (which has a REC 709 LUT on it to begin with*) so if you load a look into THAT, than yes, your exposure controls will read based on that look.

The only way we found the true sensor clip was through testing. I want to corroborate my finding with future tests, but so far what we determined, when false color reads red with the camera set at 200 ISO, that is pretty close to actual "sensor clip." Which is not to say that I shoot at 200, and I don't consider this the means to determine the camera's native ISO. I still shoot primarily at 320 and let my brightest whites go red in false color mode. But when I am "iffy" on whether something is in the danger zone of being unrecoverable, I switch it over to 200 to check.


* Since the viewfinder output is REC 709, we found that when we put a waveform monitor on that signal, it correlated pretty well with what the camera was indicating in terms of exposure. I didn't have time on set to transcribe this, but does anyone have a legend for false color and what the different values are for each color?

[Kevin Halverson]

Hey Khmuse,

It sounds like you need a camera to test with. I would be happy to donate mine to your cause for a couple of days if you like. It looks like we are both in OC so it shouldn't be any hassle.

PM me and I will send you my info.

Thanks for the very generous offer! I just sent you a PM with my contact details. I have been using Mark's (#41) for testing, but as he is in Studio City and I would certainly prefer to avoid the long drive.

Thanks again and I look forward to hearing back from you.

Kevin

[Adrian T.]

From another thread:


Very important! You can set the Red to ASA 100 which is dangerous because the built in zebras, false color, histograms, etc will CHANGE based on your camera rating... you'll be effectively clipping but zebras, false color, etc may look fine.

In that thread (http://www.reduser.net/forum/showthread.php?p=118585#post118585) we started to discuss this and I think we came to the conclusion that this is not right. The indicators should always be based on the RAW data and not on the RGB preview.

Otherwise we get clipping without knowing it or we get serious underexposure because we only thought that we were clipping.

RED people, are you listening?

[Emmanuel Cambier]

Thanks Jendra

But is it just the ISO settings that change the zebra and false color or also the changes to the monitoring like brightness and contrast and so on?
Because if it's only the ISO then we can just leave it to 320 and have the tools work as supposed.

Emmanuel

[Adrian T.]

But is it just the ISO settings that change the zebra and false color or also the changes to the monitoring like brightness and contrast and so on?
Because if it's only the ISO then we can just leave it to 320 and have the tools work as supposed.

I don't think so. The indicators would still be based on the RGB preview instead of the RAW data. That's not the same, even at ISO 320.

[Arnaud Paris]

On our Red#88 with firmware v9 the indicators do react to the changes we make to the ASA settings.
Just our two cents...
But I'm very interested in knowing how other red DPs are setting their exposure.
We're actually leaving for a short film shoot in two days and after testing our camera the DP decided to base his lighting onto a 250 ASA value using his lightmeter.
The DP also checked some of the footage of his testing onto our Scratch histogram. He is impressed but also surprised at how underexposed the images look still based on the 250 ASA value.

Arno
www.locared.com

[Kevin Halverson]

On our Red#88 with firmware v9 the indicators do react to the changes we make to the ASA settings.
Just our two cents...

Arno
www.locared.com

I haven't gotten another camera to retest yet, but this is exactly what I was concerned might be the case. As soon as I have one, I will go through and do a full test with each build and see if this is a consistent behavior across the various versions. My last testing revealed that with an 'ISO' setting of 241 (250 indicated), the false color and actual clipping points were coincident.

[Kevin Halverson]

Great news! Ninety-Nine has generously offered a loan of his camera so that we can conduct a full test on this behavior. We will be posting the results here in the next few days.

[David Mullen ASC]

If that is true, I agree that the histogram and false color data should be an indication of what is being captured in RAW independent of the ASA rating selected, if the entire point of RED is to use the RAW format to its fullest.

If Jendra is correct, I guess we could switch the camera temporarily to 200 ASA to check that info, but that seems inconvenient.

[Kevin Halverson]

I will do a second testing round tomorrow and confirm the behavior with as many builds as I can. I will also reconfirm the native sensitivity and map out the clipping points based upon actual illumination values at the sensor.

Kevin

[Mark Mannschreck]

I will do a second testing round tomorrow and confirm the behavior with as many builds as I can. I will also reconfirm the native sensitivity and map out the clipping points based upon actual illumination values at the sensor.

Kevin

Kevin rules!

I actually haven't been all that concerned about this issue because I'll rate (slightly) higher but never down... When I rate higher I DO want the meters to transpose because I need to know where I'm at and it's my only reference. It's the relativity of the range I'm interested in. BTW - you're gonna get the same results with the SDI outs with a waveform and/or monitor with a waveform. I don't think there should be an ISO 100 available though... That's where dangers may lie.

[Alexander Christ]

I don't think so. The indicators would still be based on the RGB preview instead of the RAW data. That's not the same, even at ISO 320.

I second that!

My post from the other thread: A DSLR works in the same way: if you shoot RAW and use the histogram to check exposure, you can't fully trust on what you see, because the histogram is the translation of a JPEG-transformation, which differs from the information of the RAW-image.

[Jendra]

I will do a second testing round tomorrow and confirm the behavior with as many builds as I can. I will also reconfirm the native sensitivity and map out the clipping points based upon actual illumination values at the sensor.

Kevin

I'm all for more tests that benefit the community, but I had a long talk with Stuart from RED yesterday, and confirmed that nothing has changed as far as the exposure indicators reading the REC 709 signal, not the RAW. The way all these work: false color, zebra, histogram, waveform etc. are all based on video signals, not RAW data. There is currently no way to meter RAW data, but they are thinking about this issue...

[Jendra]

does anyone have a legend for false color and what the different values are for each color?

I found it. It's in the operations manual, available on www.red.com/support.

[David Mullen ASC]

exposure indicators reading the REC 709 signal, not the RAW. The way all these work: false color, zebra, histogram, waveform etc. are all based on video signals, not RAW data. There is currently no way to meter RAW data, but they are thinking about this issue...

Then I guess you'd have to do what Jendra did, devise a test. For example, shoot some chip charts in RAW and figure out the real clip point, then play around with the ASA ratings until the false color & histograms give you the same indication for clipping and use that as your guide. Or learn to use a spot meter after you've figured out how many stops over your key exposure can a bright area be and not clip.

[Sanjin Jukic]

I'm all for more tests that benefit the community, but I had a long talk with Stuart from RED yesterday, and confirmed that nothing has changed as far as the exposure indicators reading the REC 709 signal, not the RAW. The way all these work: false color, zebra, histogram, waveform etc. are all based on video signals, not RAW data. There is currently no way to meter RAW data, but they are thinking about this issue...

Thanks Jendra for the info.

So the classis shooting control tethered is still not possible with RED.

Shooting tethered is something the most people here are interested in.

They said (RED) that all is in RAW and later you could do what ever you want(!?).

You MUST bring the files immediately to computer to see what you got.

Otherwise you do not know.

Don't trust to your preview monitor.

That's the most frustrating for a lot of users.

Fortunately people have laptops or desktops in studio or in field and they could immediately try out to "squeeze" the best from the cam.

Make notes and continue.

TESTS, TESTS and more TESTS before you make decision to have a real ACTION.

I see shooting with RED only like this.

You test and test, you learn and learn, then you go to shoot.

Maybe one day when you learn a lot then you could master RED.

[Kevin Halverson]

I'm all for more tests that benefit the community, but I had a long talk with Stuart from RED yesterday, and confirmed that nothing has changed as far as the exposure indicators reading the REC 709 signal, not the RAW. The way all these work: false color, zebra, histogram, waveform etc. are all based on video signals, not RAW data. There is currently no way to meter RAW data, but they are thinking about this issue...

This concurs with my earlier testing (end of September). If, all the monitoring paths are via the same decoded signal (including zebras, histogram, etc) then what would be useful is to know an in camera setup that provides the best correlation with the actual raw recorded signal. My intention is to quantify the actual illumination level on the surface of the sensor necessary to just hit the clipping point (this will be a repeat of my earlier test). I will then work back to find in camera settings that corresponds as closely as possible to the actual raw data set.

If anyone has any suggestions for other testing that would be useful, please feel free to post it here and if time allows, I will perform and quantify the results.

Kevin

[Sanjin Jukic]

If we would have TETHEARED SHOOTING with REDCINE something like photographers have with Aperture, Lightroom, CapturePro etc.

What does it mean?

Answer/quote:

"...Captured images are sent directly to the computer,
and in many cases you can even control your camera's settings
and trigger the shutter directly from your desktop.
Many studio photographers love to shoot tethered.
The photographer can focus on the subject while an editor
or art director watches the images appear on screen as they are captured..."

Link>> (http://digitalmedia.oreilly.com/2006/12/05/tethered-shooting-in-aperture.html)

We all need:

What You See Is What You Get ( WYSIWYG )

[David Mullen ASC]

All I can say is that at the speed of most narrative production, opening up shots before you actually shoot the scene on computers to check exposure levels of the RAW file is too time-consuming.

Feedback on exposure has to be immediate and live for it to be of any practical value in most shooting situations where you are doing 20, 30, 50 set-ups a day. Especially in an uncontrolled situation where you have to suddenly grab some bit of unplanned action.

[Sanjin Jukic]

All I can say is that at the speed of most narrative production, opening up shots before you actually shoot the scene on computers to check exposure levels of the RAW file is too time-consuming.

Feedback on exposure has to be immediate and live for it to be of any practical value in most shooting situations where you are doing 20, 30, 50 set-ups a day. Especially in an uncontrolled situation where you have to suddenly grab some bit of unplanned action.


Then you should have a lot of archived SET UP notes about shooting RAW with
RED to choose a right ONE for some or any sort of suddenly happen action.

[Kevin Halverson]

I fully agree with David, efficiency on set is essential and having to depend upon something/someone else to get something as essential as exposure information is going to reduce that efficiency.

Knowing the exact behavior of the sensor and a way to get on camera metering to correspond is the goal of the test that I will perform today. The ability to have a 'look' that represents something near the final post workflow product is a wonderful on set tool, but as or more important is an ability to know where your exposure is in all three color channels in a quick and efficient manner. This requires either metering of the RAW data set or an in camera setup which, as closely as possible, corresponds to it.

Kevin

[Sanjin Jukic]

I fully agree with David, efficiency on set is essential and having to depend upon something/someone else to get something as essential as exposure information is going to reduce that efficiency.

Knowing the exact behavior of the sensor and a way to get on camera metering to correspond is the goal of the test that I will perform today. The ability to have a 'look' that represents something near the final post workflow product is a wonderful on set tool, but as or more important is an ability to know where your exposure is in all three color channels in a quick and efficient manner. This requires either metering of the RAW data set or an in camera setup which, as closely as possible, corresponds to it.

Kevin

Agree.

"Learning Curve" shooting with RED could be good title for a book written from a RED expert.

Waiting for RED and also a book...

[Jendra]

All I can say is that at the speed of most narrative production, opening up shots before you actually shoot the scene on computers to check exposure levels of the RAW file is too time-consuming.

Feedback on exposure has to be immediate and live for it to be of any practical value in most shooting situations where you are doing 20, 30, 50 set-ups a day. Especially in an uncontrolled situation where you have to suddenly grab some bit of unplanned action.


Absolutely. My "best practice" is use "false color" at ISO 320. Then you have a little built in headroom. Also to be totally exact and know precisely how much headroom, I need to do more complete dynamic range tests. I would then supplement my exposure judgments by using a spot meter.

[Emmanuel Cambier]

Absolutely. My "best practice" is use "false color" at ISO 320. Then you have a little built in headroom. Also to be totally exact and know precisely how much headroom, I need to do more complete dynamic range tests. I would then supplement my exposure judgments by using a spot meter.

This could be the best solution: stick to 320 ISO for good.
The reason why one would want to change the ISO setting is getting very unclear for me, to begin with it's not changing the sensor's true sensitivity, and now it's become clear that it would render the exposure tools inaccurate to what's really recorded.

Now it would be nice to know exactly what ISO setting is the most accurate, 320, 300, 250 or else. Maybe the various test could give us a clue.
Thank you all testers, much appreciated here.

Emmanuel

[Kevin Halverson]

So far, ISO 250 is what I got last time (320 gives you a bit of headroom). I will be confirming this again on a different camera body, but from what I have learned, they adjust each camera for this property in non user accessible portion of the calibration setup so I expect it to be very closely matched camera to camera.

Kevin

[Sanjin Jukic]

I give you a new link (http://photoshopnews.com/2006/12/06/tethered-shooting-in-lightroom/) about tethered shooting in photography before to continue.

Now I will turn to the HD video acquisition and give you photo that shows the text about "Lighting to monitor"

from the book "High Definition Cinematography" by Paul Wheeler (http://www.amazon.de/High-Definition-Cinematography-Paul-Wheeler/dp/024052036X),
Second edition 2007, Focal Press, Part 4, Chapter 16 - Lighting and exposing for HD, Page 83.

It is very important for shooters that are used to work with Cinealta, Genesis, Viper, Varicaam...etc, HD camera to get used to RED.

The idea is about to make an effective workflow and fusion between sort of tethered checking test shots with a RAW app in computer (like in photography) and

together with the HD experience of using monitor on set to adjust lighting and movie picture desired look.

It is about if REDCINE could load (???) (LIVE stream) from camera tethered to a computer using USB or FW cable.

Or to write a new little app for the RED LOOK tethered capture (equivalent to tethering possibilities in Ninkon NX, Canon's Camera Window, Capture Pro or Bible for digital photography).

A single RAW frame (RED LOOK R3D file) can be quickly brought into REDCINE, bypassing the need for a camera card.

If this is not possible then simply load RED LOOK R3D from CF or RED DRIVE for a test look to your laptop or desktop.

With that option RED cinematographer can have immediate look at the picture (what should be a real one) in a RAW app (REDCINE).

Simultaniously watching: at the first picture on set, the second picture through EVF, the third on camera LCD and the forth picture in REDCINE on computer lcd (laptop/desktop).

It could be a total control of the picture acquisition.

And the picture ADJUSTMENT could finally start:

It would be interactive and something like What You See And Adjust Is What You Get.

No surprises later in post about getting a wrong exposure, noise, highlights, shadows, etc...

She/he could fine tune simultaniously in camera (exposure/ISO/shutter angle/speed, white balance, etc...

all acquisition's basic image parameters) and then with the lens and lighting do further adjustments...

It is an INTERACTIVE way of working with camera and it would be very close to Writing with light (Storaro) (http://www.theasc.com/cgibin/store/acsstore.cgi?user_action=detail&catalogno=11247).

It already showed us that a lot of RED users do camera adjusting (doodling) in this direction but all with a different result, good and bad.

Also some of them are quite nervous about that present situation.

One RED expert user finally could set up and publish all parameters and then the (RED) users can use it easy.

Maybe this idea looks GREAT in theory but if it would become reality in RED praxis, all the confusion and doubts about shooting with RED could be suddenly gone with the wind.

http://homepage.mac.com/sanjinjukic/RED/lighting-to-monitor.jpg

[Adrian T.]

I must say that this wohle issue is the worst news about RED ONE I got for a long time.
It simply doesn't make sense the way it's implemented. The indicators are virtually useless.

RED people, wizards and pixies! :help:

Please show us indicators based on the actual RAW data!

Please bring the RAW workflow to perfection!

[Hans von Sonntag]

Interesting thread.

One thing that come to my mind is the fact that I never had the chance to meter a scene and then to check immediately on a monitor the result simply because looking into the negative is impossible.

With RED this is possible. Now there is a uncertainty to the metering one does (at least I would have) that was never around. Personal habits and feelings to light metering are now put into question. The gold old self-assuredness concerning exposure is gone. Bummer.

Nevertheless I am very interested into the upcoming results, just a thought...

Hans

[Kevin Halverson]

It's really not bad news at all, just an important concept to understand. I have already sent on of the RED team members a pair of ideas that could make for a really simple solution. I am sure that they have some great ideas of their own too. Nothing to worry about, really all that we need is a bit more testing and well have a configuration that can be used with high confidence.

[Jendra]

So far, ISO 250 is what I got last time (320 gives you a bit of headroom). I will be confirming this again on a different camera body, but from what I have learned, they adjust each camera for this property in non user accessible portion of the calibration setup so I expect it to be very closely matched camera to camera.

Kevin

I concur more or less, but am trying to find a confident method of determination. What was your methodology? There are so many variables involved, especially how you post it.

Mr. Mullen, what method would you use? For my thinking, with all the variables, I want to arrive at the same results from at least two different pathways to minimize errors/oversights.

[Kevin Halverson]

Hello Jendra,

I will post my findings and methodology once the confirmation testing is completed, which should be sometime this eveing. Just about to start testing again, we are getting the setup finished now, should begin in about an hour. I assume that I don't need to go to the extent of providing tracability certs on my instrumentation, just the method and results.

Kevin

[David Mullen ASC]

Mr. Mullen, what method would you use? For my thinking, with all the variables, I want to arrive at the same results from at least two different pathways to minimize errors/oversights.

I suppose one testing idea would be to put an 18% grey card next to a white card and a deep black card, and then shoot the three cards in one-stop increments to determine when on the RAW file, there is no separation between the white and 18% grey (it's clipped) or black and 18% grey (it's crushed) so that you can know from a spot meter reading how many stops over your shooting stop can the reflectance of an object be before it goes to pure white. Assuming you want to be spot metering hot spots on the set when you shoot (it's not a habit I ever got into when shooting film, and with a normal HD camera, there are zebras in the eyepiece to tell you that info.)

The other testing method would have to determine how to get the histogram and false color reading to tell you how the RAW file is recording the information, maybe by checking it at a particular ASA like 250.

[Kevin Halverson]

Thanks to Kip (RED #99) we got to do another round of testing last evening. I am processing the data now and will post the results later.

The conclusion that I have come to is that any of these would be useful solutions.

1) As soon as 'looks' are enabled, then creating a look that as closely as possible matches the raw data (no clipping headroom, linear, etc).

2) If the histogram feature (in camera) could have an option to select raw data as a source, that would be great.

3) My real wish list item is an on camera spot meter with a variable spot size (perhaps 20 photosites) that could be navigated around the image and display three corresponding numeric values (demosaiced RGB). This would allow for a really quick way of verifying every aspect of the exposure.

Kevin

[Sanjin Jukic]

Thanks to Kip (RED #99) we got to do another round of testing last evening. I am processing the data now and will post the results later.

The conclusion that I have come to is that any of these would be useful solutions.

1) As soon as 'looks' are enabled, then creating a look that as closely as possible matches the raw data (no clipping headroom, linear, etc).

2) If the histogram feature (in camera) could have an option to select raw data as a source, that would be great.

3) My real wish list item is an on camera spot meter with a variable spot size (perhaps 20 photosites) that could be navigated around the image and display three corresponding numeric values (demosaiced RGB). This would allow for a really quick way of verifying every aspect of the exposure.

Kevin


Thanks Kevin.

Great news, and if it is all like you said then SOON no more headache and nightmare for our busy RED DPs.

By the way I am a cinematographer and I do have another approach to a problem.

[Kevin Halverson]

Hi Sanjin,

I am always interested to learn of different approaches that other cinematographers take, so feel free to share your idea.

Kevin

[Sanjin Jukic]

Kevin.

all my thoughts are pure theory (now) because of lack in praxis.

But after Jan/Feb winter time frame when I am expecting RED delivery could be more positive or negative proofs for my theories.

Go on Kevin with your work and mostly do not take me so serious because there are just my pure theories what ever I writing here.

[Hans von Sonntag]

The conclusion that I have come to is that any of these would be useful solutions.

1) As soon as 'looks' are enabled, then creating a look that as closely as possible matches the raw data (no clipping headroom, linear, etc).

2) If the histogram feature (in camera) could have an option to select raw data as a source, that would be great.


Good Idea, Kevin.

The idea of implementing looks to get an idea what the picture will look like after debayering and grading might be tempting. In real life it is much more interesting to know what actually happens in-camera.

I suppose this kind of LUT stuff on set will slow things down a lot - not a real improvement.

When clients complain about dull pictures I do some telecineing with the monitor, brightness, contrast, saturation... much easier and faster to accomplish.

Hans

[Emmanuel Cambier]

To me, a better solution would be to be able to choose different LUT on the different monitors.
Like the DP and camera op would get the "Raw" signal while the director and the producers would get a less flat image on hdsi monitors.
I always thought it would be so, by the way.

Emmanuel

[David Mullen ASC]

I'm still not clear on this, but is a RAW image even viewable? Doesn't it have to be converted to RGB to be viewable?

If so, the question is whether a conversion to a flat 10-big LOG RGB image would make the exposure range being captured more obvious, for anyone who doesn't mind looking at a milky LOG image.

When Lucas put up the converted "RAW" image on his Scratch station during L.A.R.T., it always looked really dark for material shot at 320 ASA. So if that's what RAW is, would it be confusing to look at? Would you be tempted to overexpose if looking at a dark image like that? Or would a low-con LOG image make the extreme exposure ends more visible on the monitor?

[Kevin Halverson]

The "RAW" image is "viewable" but not representative of anything that would be of value, it has to be de-mosaiced in order to be useful. The other thing to consider is how to map the 12 bit linear output of the sensor/quantizer to a luminance range that is useful. I proposed a 'look' that would only be useful to help see the exposure extremes since the dynamic range of the imager is greater than either the LCD or EVF. I would never use this as the only tool for setting exposure or making lighting choices, rather, its just another one that would be useful to be able to cycle through. This isn't the best solution, its just one that should be easy to add once in camera looks are enabled. What I want is a few quickly accessible functions that would allow for in camera exposure decisions that take into account the downstream flexibility, but don't abuse it to the point of being overly optimistic.

I still would like the histogram to be assignable to just after the de-mosaic step and ignore all other in camera settings, particularly the "ASA" choice. This would prevent someone from making an exposure choice based upon anything other than what is actually being captured.

Personally, my wish list idea is the electronic spot meter that I proposed (assignable number of photosites and three numeric values corresponding to the primary colors). When I used to shoot large format color transparency (4"x5") I used a spot meter nearly exclusively and found it a great tool. Having one that I could fly around the image and get luminance values for all three color channels would be an ideal tool as far as I am concerned.

Would love to learn what others think about the usefulness of any of these proposed tools.

Kevin Halverson

[Steve Sherrick]

Kevin,

This is a huge help. Much kudos for doing the testing and to others for thinking these issues through. I truly believe that by the end of 2008, there will be a lot of finely tuned methods of shooting with this camera with a lot of the major issues worked out and improvements happeing continuously by the Red team. We are at the beginning and I think a lot of this is to be expected for a brand new camera design. Without a lot of you guys doing this testing, it would leave those without cameras at the moment in the dark. Thanks!

Steve

[red1225]

500T-320D w/85
100T-64D w/85

...and so on and so on.

We never had this problem with film and it is a very simple math. Meters work just fine.

Look...the only way to figure this out is by trial and error. Your meter should be used for levels only anyway.

TRUST YOUR EYE!!!

[Sanjin Jukic]

Could be a useful tip from DSLR experience>>>

QUOTE:

"When shooting in RAW mode under tricky lighting, sometimes it is useful to set a slight negative exposure compensation (0.3-0.5). This decreases the chance of clipped highlights (http://www.cambridgeincolour.com/tutorials/histograms1.htm), yet still allows one to increase the exposure afterwards. Alternatively, a positive exposure compensation can be used to improve the signal to noise ratio (http://www.cambridgeincolour.com/tutorials/noise.htm) in situations where the highlights are far from clipping."

1. Clipped highlights (http://www.cambridgeincolour.com/tutorials/histograms1.htm)

http://www.cambridgeincolour.com/tutorials/graphics/tut_hist_lowkey_auto_hist.png

http://www.cambridgeincolour.com/tutorials/graphics/tut_hist_zoomlowkey_clipped.jpg


2. Signal to noise ratio (http://www.cambridgeincolour.com/tutorials/noise.htm)



MORE>>> (http://www.cambridgeincolour.com/tutorials/camera-metering.htm)