Red v30 exposure tools won't match lightmeter

[Nigel Smith]

There are some new changes coming with the new FLUT Science camera builds... which will start with Build 30. Your meter now matches the camera. Incident metering will match mid grey.

Jim

 

I just came across this quote from Jim on a now closed thread. I was struck by it's relevance to a problem I'm having with our film school Red.

 

I cannot get the camera's metering tools to match up with readings from my Sekonic L608 meter.

I posted on Reduser regarding this and was advised by Jim that I should try build 30 as the problem had been fixed. David Mullen also assured me that the new Mysterium-X sensor and build 30 were matching his lightmeter. So I duly loaded the new firmware, but the problem has not gone away.

 

I posted this earlier on Reduser:

"I just carried out a similar test on Build v30.4.1, and I'm still getting around a stop difference between my L608 and the RAW level meter. To get the green bar lit up when filling the frame with an 18% grey card I need T4, while my light meter tells me I need T5.6. Both incident and reflected readings on the Sekonic agree.

This test was using Arri studio cools with daylight tubes, camera set to 5600k and EI 320. Red was in 2K 16:9 mode, using a Canon T2.4 8-64mm lens.

I'm really hoping there's something I'm missing or doing wrong, so any advice is appreciated. "

 

Just to be clear - I bought a Red one, and have great respect for what Jim and his team are doing. Many things in build 30 are terrific, including the fact that almost all exposure tools now show clipping and exposure levels at the sensor. But that's of limited use if it doesn't match my meter. Which should I trust?

 

Apologies for the long post - I'm posting here in the hope of advice, because I find that sometimes reduser kind of ignores 'problem' posts.

So does anyone have suggestions as to what I'm doing wrong?

Edited March 18, 2010 by Nigel Smith

[John Sprung]

First thought, are you using a camera with the old chip and build 30 firmware? I believe that build 30 is for the new MX chip. The last build number for the old chip was in the low 20's.

 

I'd think that the old and new chips would be pretty much the same where the bits come out, the improvement would be in getting from photons to bits. Does the firmware know which chip it has?

 

 

 

 

-- J.S.

[Nigel Smith]

First thought, are you using a camera with the old chip and build 30 firmware? I believe that build 30 is for the new MX chip. The last build number for the old chip was in the low 20's.

 

 

 

 

-- J.S.

Thanks for the reply Jon.

Build 30 is for both sensors.

Mr Jannard himself suggested I try it, and he will know that there are very few mysterium x sensors out there ATM.

I have just come across Mr Natress effectively rubbishing Sekonic meters [particularly ones that do incident and reflected light readings] on the Arri Alexa thread, so perhaps he has a point and it's my meter?

Mind you it matches my studio Deluxe Mk2 :lol:

[John Sprung]

Mr Jannard himself suggested I try it, and he will know that there are very few mysterium x sensors out there ATM.

 

Ah, that's the beauty of renting. We have two pilots on Red, both MX. Probably half a dozen bodies, two working and a spare for each show.

 

 

 

 

-- J.S.

[Graeme Nattress]

Thanks for the reply Jon.

Build 30 is for both sensors.

Mr Jannard himself suggested I try it, and he will know that there are very few mysterium x sensors out there ATM.

I have just come across Mr Natress effectively rubbishing Sekonic meters [particularly ones that do incident and reflected light readings] on the Arri Alexa thread, so perhaps he has a point and it's my meter?

Mind you it matches my studio Deluxe Mk2 :lol:

 

Not rubbishing them as they're not rubbish, just not 100% accurate.

 

Are you incident metering or spot metering?

 

In RGB false colour what colour is the grey card?

 

A meter won't ever tell you how to expose the RAW as it doesn't understand that. What we do is check that 18% grey incident metered will produce an RGB level (on the 0 to 100% scale) of 44% when the ISO and lens is set to the meter's recommendation.

 

Graeme

[David Mullen ASC]

What I discovered is that when I exposed a SMPTE 11-step chip chart so that the crossover point on a waveform was at 50%, with an equal number of steps above and below that, switching between RedColor and RAWView, I had to use a 320 to 400 ASA setting in RedColor so that when I switched to RAWView (which ignores ASA metadata), the chart was exposed the same.

 

It's an easy test, just light a face and switch between RAW and RedColor, adjusting ASA, until they look the same, exposure-wise.

 

This doesn't mean that this is the best way to expose however, RAW images, like Log images, should be a bit lower on the waveform to give you more overexposure headroom. I seem to recall that a grey scale, when viewed in Log on a Genesis, is something like 40% or 45% for middle grey and 70% for white.

 

A waveform is a more accurate meter than a separate meter, but I don't know if your meter is set correctly in terms of shutter speed, for example.

 

If you want to line-up your meter to match RAW, I guess I can understand that if you feel you can spot meter for the clip and crush points, but I don't think it's the best method for exposing midtones and faces.

 

If you were using the older sensor, I can understand more how your meter would match RAW more in the 160 to 200 ASA range, a lot of people reported that, but with the new sensor, it seems like it should match more in the 320 to 400 ASA range.

[Nigel Smith]

Not rubbishing them as they're not rubbish, just not 100% accurate.

 

Are you incident metering or spot metering?

 

In RGB false colour what colour is the grey card?

 

A meter won't ever tell you how to expose the RAW as it doesn't understand that. What we do is check that 18% grey incident metered will produce an RGB level (on the 0 to 100% scale) of 44% when the ISO and lens is set to the meter's recommendation.

 

Graeme

Thanks a lot for the reply Graeme, and David too.

I know you must both be very busy.

 

I metered in both modes, as mentioned above.

I was reacting to the comment that your Sekonic meter was 20% out, which really surprised me.

 

I don't really understand the point you and David make about light meters not matching up with RAW.

i'm confused by the fact that the RAW exposure meter is measured in stops, according to the manual, and also by RAW viewing mode - surely that is an RGB path?

I suppose I don't have a good enough grasp of the science.

 

Thanks to david for the suggestion re ISO. Perhaps rating the camera differently will help.

I'll try some more tests, including false colour this morning.

[Graeme Nattress]

Any raw metering is for exposing the raw. Methodologies there include "ETTR" expose to the right, or protecting highlights, or ensuring you're not buried in the noise-floor. They are not how light meters work. A single meter incident reading has no concept of "scene", yet that is what the camera records.

 

Graeme

[Nigel Smith]

A single meter incident reading has no concept of "scene", yet that is what the camera records.

 

Graeme

?

Ok so you will likely get less detail in shadows and highlights that you can see with your eye, dependent on recording medium and dynamic range. But surely an incident reading gives you an aperture setting at which mid tones will be exposed correctly, and what you see is what you will get?

[Graeme Nattress]

Yes, it will get mid-tones where you want them, but it has no concept of where the highlights are, where the shadows are, and which are important and how important to the shot. Obviously if the scene DR is less than camera DR, and you're exposing for mids, and you want the mids in the middle, all is fine. However, that is not always the case.

 

Graeme

[Nigel Smith]

Yes, it will get mid-tones where you want them, but it has no concept of where the highlights are, where the shadows are, and which are important and how important to the shot. Obviously if the scene DR is less than camera DR, and you're exposing for mids, and you want the mids in the middle, all is fine. However, that is not always the case.

 

Graeme

Thanks Graeme. I see your point.

 

I have done some more tests and I am still finding differences.

Green in false colour mode gives a reading 1 + 1/3 stop open from my lightmeter reading.

 

On the other hand: I use the grey scale side strip option in EVF and LCD to warn me when my angle of view is off for the LCD. I found that if I matched the 18% grey card up with that, I DO get the same exposure as my light meter suggests!

 

Looking at the tests in Redcine X, using REDcolor and REDGamma, the light meter exposed shots look correct, and the false colour metered shots look overexposed.

 

So I guess I'm going to trust my meter rather than the false colour display, and use the RAW tool to avoid clipping in the highlights.

 

BTW testing as per David's advice, the camera does seem to rate at 320 ASA.

Edited March 19, 2010 by Nigel Smith

[Graeme Nattress]

Yes, use the raw clip overlay to look for red fizzle warning you of clipping, and the purple fizzle that suggest when things might be a bit too dark. That does work very well.

 

Graeme

[John Sprung]

Yes, use the raw clip overlay to look for red fizzle warning you of clipping, and the purple fizzle that suggest when things might be a bit too dark.

 

Bottom line, fuggedabout light meters. Save your money. The Red is its own best light meter.

 

 

 

 

 

 

-- J.S.

[David Mullen ASC]

So I guess I'm going to trust my meter rather than the false colour display, and use the RAW tool to avoid clipping in the highlights.

 

BTW testing as per David's advice, the camera does seem to rate at 320 ASA.

 

But my advice is not to actually rate it at 320 ASA -- because of the low noise of the MX sensor, you are better off at a rating closer to 800 ASA, 500 ASA if you really want to be conservative, and thereby get a more natural roll-off into overexposure. You're under the false impression that you're supposed to expose the image so that it looks normally bright in RAW.

[Graeme Nattress]

But my advice is not to actually rate it at 320 ASA -- because of the low noise of the MX sensor, you are better off at a rating closer to 800 ASA, 500 ASA if you really want to be conservative, and thereby get a more natural roll-off into overexposure. You're under the false impression that you're supposed to expose the image so that it looks normally bright in RAW.

 

Yes indeed - if you're M, start at ISO320. If you're M-X start at ISO800.

 

RAW exposure is to look for clipped highlights, not to see a "reasonable" picture, although it could very well look reasonable, it's not guaranteed to, and will look less that right in raw as you move further away from ISO320.

 

Graeme

[Nigel Smith]

But my advice is not to actually rate it at 320 ASA -- because of the low noise of the MX sensor, you are better off at a rating closer to 800 ASA, 500 ASA if you really want to be conservative, and thereby get a more natural roll-off into overexposure. You're under the false impression that you're supposed to expose the image so that it looks normally bright in RAW.

My camera doesn't have the M-X sensor, only build 30.

one day, maybe...

So I followed this advice -

"It's an easy test, just light a face and switch between RAW and RedColor, adjusting ASA, until they look the same, exposure-wise."

- only I used a grey card instead of a face.

that's what gave me an ISO of 320.

 

I take your point that a RAW image should look underexposed on the camera to capture most data.

But I made the choice of correctly exposed image in Redcine-x, looking at the files in REDcolor and REDgamma.

The one that looked best was the one based on my Sekonic's incident reading, which was exposed a stop under what false colour suggested.

 

Thanks to you both for your responses and advice :)

Edited March 20, 2010 by Nigel Smith

[Chris Dalton]

just to throw another opinion in the mix, I like 500 as a good Base for the MX Sensor. Red lists 800, but they did that as soon as Arri announced the Alexa. 800 is still a clean enough signal to work with, but I think 500 is the right answer. What ever you do Protect your highlights.

[Bruce Greene]

I just came across this quote from Jim on a now closed thread. I was struck by it's relevance to a problem I'm having with our film school Red.

 

I cannot get the camera's metering tools to match up with readings from my Sekonic L608 meter.

I posted on Reduser regarding this and was advised by Jim that I should try build 30 as the problem had been fixed. David Mullen also assured me that the new Mysterium-X sensor and build 30 were matching his lightmeter. So I duly loaded the new firmware, but the problem has not gone away.

 

I posted this earlier on Reduser:

"I just carried out a similar test on Build v30.4.1, and I'm still getting around a stop difference between my L608 and the RAW level meter. To get the green bar lit up when filling the frame with an 18% grey card I need T4, while my light meter tells me I need T5.6. Both incident and reflected readings on the Sekonic agree.

This test was using Arri studio cools with daylight tubes, camera set to 5600k and EI 320. Red was in 2K 16:9 mode, using a Canon T2.4 8-64mm lens.

I'm really hoping there's something I'm missing or doing wrong, so any advice is appreciated. "

 

Just to be clear - I bought a Red one, and have great respect for what Jim and his team are doing. Many things in build 30 are terrific, including the fact that almost all exposure tools now show clipping and exposure levels at the sensor. But that's of limited use if it doesn't match my meter. Which should I trust?

 

Apologies for the long post - I'm posting here in the hope of advice, because I find that sometimes reduser kind of ignores 'problem' posts.

So does anyone have suggestions as to what I'm doing wrong?

 

Nigel,

 

This has been an interesting topic and replies.

 

I think that some experimentation would be a good idea here in order to learn how you like to expose your RED or even other digital camera. The sensor, under a given colored light source, will have a fixed response to light exposures and a fixed range depending upon your tolerance for noise in the dark areas.

 

In general, you can have the cleanest images by giving the most exposure possible for a scene at the expense of limiting detail in the highlights, or the greatest dynamic range at the expense of noise in the deepest shadow areas. One could expose differently depending upon the scene brightness range to extract the best possible image from each scene, rendering the idea of the "camera ISO" kind of irrelevant.

 

I think though that this is not how most of us work as it may make for very inconsistent dailies and make producers and directors nervous until the final film is color graded. When shooting a movie exposure consistency has it's place.

 

To determine your "normal" exposure I think you might start your test by setting up a scene that contains what would be reasonably seen range of tones, from deep shadows to highlights that would naturally blow out when shooting film. Include in your set a grey scale test chart to represent the middle range of tones that would normally be reproduced in a linear fashion to make a realistic image where the the black patch is close to black and the white patch is bright white, but not as bright as specular highlights.

 

The other side of this test is the de-bayer / RAW processing settings you will set up as your "standard" correction. You will need to come up with settings which render the test scene in a way you like to see it reproduced. To complicate matters, the RAW conversion settings will differ with changes in camera exposure as you will apply different curves with different exposures.

 

So you might just start by exposing your test scene (use a slate!) from ISO 160 to 1200. Bring each ISO shot into your RAW converter and make the best image you can for your chosen ISO settings. You will probably notice that the highlights clip in the low ISO settings and have too much detail in the high ISO settings and you'll clip them in the RAW converter yourself to make a pleasing image.

 

When you pick out the image processing that you like best for rolling off the highlights vs noise in the shadows, you'll know your "normal" camera ISO and you can use your light meter to expose the camera. When you apply your saved RAW conversion settings, you'll get a consistent look.

 

Do this test for both daylight and tungsten light as the sensor has different sensitivity to colors and a different dynamic range to match.

 

Since you'll be shooting your project RAW, you'll have an opportunity to fine tune the look if you do your color grade from the RAW files. If the production never revisits the RAW files, at least your editing files should reflect your basic vision of the movie.

 

Best of all, you'll get a real feel for how changing the exposure of your camera will effect the range and look of your image possible when color grading the final product.

 

If anyone else is able to clarify my suggestion or add to it, please chime in!