Determining ISO...

[Dylan Macleod CSC]

I know this has been discussed before, but I am wondering what peoples thoughts are now that there a few more cameras in circulation.

I did a test back in December with "Madrid" (I believe it was build 13?) and was interested in exposure latitude and getting a handle on how to rate this camera.

I used a waveform read from the 720 video output and based my "normal" exposure on the gray scale (blacks at around 7, white chip at around 100).

When I later "processed" the footage in REDCINE the "normal" exposure was WAAAAY "under" on the histogram. I would say, by about two stops. I didn't cross reference the waveform reading with a light meter, but if I had I would be able to determine the ISO from that.

From what I have read now in the forums, this 720 video output is not a reliable way to get at measuring the signal that is being recorded.

How DO I measure the signal that is being recorded then? And how do I "see it" when I am shooting.

I don't want to rely on a "reccomended" ISO. I wouldn't do this with film. I test and figure out how I want to rate it.

I am testing a new "X" body next week and this is my primary goal...to determine how to accurately measure the sensitivity of the camera in a measurable way "LIVE".

How do I do this?

Thanks in advance.

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

 

[donatello b]

from my own observations ..
i rate my meters for RED at
320 daylight ....
250 for 3200k tungsten ...
200 if scene color temp is below 2500k ..
i use a modified zone system ..
i have no scientific way that i measure ...
more of a trial /error ..
getting zones where i want them and have it be repeatable ...
all above ISO/ASA could change with firmware 14 ...

 

[Miguel "Macgregor" De Olaso]

125ISO. +4 stops above. -6stops under. Build12.

 

[Ken Willinger]

Finding ASA rating for video camera

Finding ASA rating for video camera

I just recently asked the question of determining a video cameras ASA rating from Rob Draper ASC. This was his reply and I think it could work with the RED Camera if you use a B4 mounted broadcast video lens:

Take an 18% grey card and light it to 40 footcandles.

Set your camera to zero dB with auto knee off and camera in factory preset (no internal adjustments).

Frame the card up so the card completely fills the frame.

Go to auto iris and see where the iris sits. (This is why you would have to do this with a B4 mounted broadcast video lens...they have an auto iris setting).

If the iris is on F4 then your camera is 500ASA Normal exposure at 40fc on 500ASA is f4 (there are charts available online and in the American Cinematographer Manual)

eg

500ASA

T Stop 2 2.8 4 5.6 8 11 16 22
f/candles 10 20 40 80 160 320 640 1280


If the iris reads say 2.8 (at 40fc) then the camera is one stop slower ie 250ASA
if it reads 5.6 (at 40fc) then it is one stop faster ie 1000ASA

etc

 

[David Mullen ASC]

125ISO. +4 stops above. -6stops under. Build12.

Which suggests: 250 ASA = +5 stops over, +5 stops under

Though in practical terms, detail existing at 5-stops overexposed at 250 ASA doesn't seem right, neither does 4-stops over at 125 ASA. My brief testing at 320 ASA showed clipping occuring above the 3-stops overexposed range. But then, that's not based on a Stouffer's wedge chart...

 

[Dylan Macleod CSC]

Much appreciate the feedback.

But still wondering... how can one get accurate and immediate feedback out of the camera. Is the 720 video feed the only option? If it is, from what I have understood from reading previous posts, it is not accurate read of the signal being recorded.

Basically, I am hoping to use a waveform to expose the grayscale in the "middle", then cross reference with a meter to determine the ISO. But if the signal that the waveform is receiving is not a true represention of the signal that is being recorded, this is a pointless exercise.

It also makes me wonder if a waveform, the way we would traditionally use it in HD, is a useless tool with the RED?

If it is not delivering a true representation of the signal being recorded, then what is the point of this signal? Is it just to put a viewable image on the screen? Then why include the histogram?

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

 

[David Mullen ASC]

Wel, I think what most people are saying is that the 720P output, which is used to generate waveforms, false color info, etc. is an accurate way to determine exposure IF you are shooting close to the actual sensitivity of the sensor.

It's only when you are selecting a ASA rating that is further from the true sensitivity, like 500 ASA let's say, where you may be capturing more highlight information in RAW than what the 720P output is showing you (when using a rating that is more faster than the actual sensitivity of the camera. And the reverse is true the other way.)

Of course, in this example, shooting at 500 ASA, it's not a terrible thing that you may be getting more highlight information on the RAW file than you think based on the waveforms and false color warnings.

The thing is that if you see information, high or low, on your on-set monitor, you are definitely recording that information, just that you may find even a little more information on the RAW file. But it is unlikely that there would be a situation where you saw less information on the RAW file than what you saw on the set looking at the monitor.

In other words, if it's there on the set monitor, it will be there for you to work with in post; whereas if it's not there on set monitor, it may still be there for you work with in post, or it may not. So it's only when you see a problem on the set where you may want to check what's getting recorded to RAW to see if the problem is less bad than you thought (because it is unlikely to be worse than you see on the on-set monitor, exposure-wise.)

And one way of checking, according to some ideas posted here, is to switch to around a 250 to 320 ASA rating if you aren't doing so already, so that the waveforms and whatnot are closer to being accurate.

 

[Dylan Macleod CSC]

"In other words, if it's there on the set monitor, it will be there for you to work with in post; whereas if it's not there on set monitor, it may still be there for you work with in post, or it may not. So it's only when you see a problem on the set where you may want to check what's getting recorded to RAW to see if the problem is less bad than you thought (because it is unlikely to be worse than you see on the on-set monitor, exposure-wise.)

And one way of checking, according to some ideas posted here, is to switch to around a 250 to 320 ASA rating if you aren't doing so already, so that the waveforms and whatnot are closer to being accurate."

Wow. I don't like the sounds of that! "may or may not"? Yikes!

So with an F900 I would be shooting a scene with a person sitting in front of a window that looks outside into daylight. I need to light the interior to a level that allows me to see detail outside. I look at my waveform and I can do this quite easily.

With film, based on testing, I use my spot meter to determine lighting levels to balance the inside and outside.

How do I do this with the RED?

It sounds like, what your saying David, is that I light to the tolerance of the 720 output of the LCD monitor (or waveform), even though there is more "range" in the RAW file.

But don't we all want to light to the tolerance of the RAW information?

I am getting the impression that this is not possible and there will still be a little "interpretting" to be done based on playing with footage in REDCINE. Basically, find out what this "extended range" is and pack that knowledge away.

In a way, like shooting film and having to explain to people on set who are looking at the video assist..."oh, there is way more information on the negative"...

Interesting...

Thanks.

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

 

[Jon Corcuera]

I work with Macgregor but, but not for that i agree with him, we have done separate latitude tests with prety similar values, my test consisted in placing a grey card mesuring with spot and the value was 11 1/3 while the correct grey checking file in post of the image was 8. It was done with tungsten and i didn´t balance in post. So i will like to know, if you dont mind, how was youre post David because is the only point were doing a fotometer mesaurement the things my change.

Jon
AC/DIT

 

[David Mullen ASC]

So with an F900 I would be shooting a scene with a person sitting in front of a window that looks outside into daylight. I need to light the interior to a level that allows me to see detail outside. I look at my waveform and I can do this quite easily.

With film, based on testing, I use my spot meter to determine lighting levels to balance the inside and outside.

How do I do this with the RED?

You may be overthinking this...

You can do it either way. I mean, with the F900, if your monitor is properly set-up, you're probably going to see the same clipping happening on the screen that the waveform is telling you is happening - the waveform is just a double-check to some degree, a properly set-up monitor can be a pretty good judge of what you are getting on tape.

Or, as with film, you can test the RED camera for its sensitivity and dynamic range and use your spot meter in the same way. Not that I shoot film that way (I'm not much for metering everything.)

The thing is with color negative is that where it tends to burn out to white is where you want it to burn out to white, so generally you figure the hot highlights will reproduce more or less how you see them on the set. The overexposure latitude is quite good with film. It's really the shadows where you sometimes worry about losing detail and might double-check with a meter if you are playing things on the edge of visibility.

Exposing the RED image properly is not rocket science. The only issue to watch for is bright highlight detail when you are worried about it, like when it looks too clippy on the monitor -- and there are workarounds to check it (worst case scenario, open up the file on set on a laptop if you are really worried. Otherwise, set the ASA rating to around 250 to 320 and use the waveform, false color reading, etc.)

Exposing so it looks good on the monitor and knowing you have a little more information to play with in post using the RAW file is not the worst method either. We often have more information on a film negative than we end up using in the final image.

 

[Chris Nuzzaco]

I'm curious as to why everyone is still basing sensitivity readings off 18% gray. An important thing to remember about light meters is that they are basically "stupid" instruments. You can rig a modern light meter in many different ways. I personally base my ratings off of highlight clipping, and then meter off of highlights for exposure settings etc.. (though NOT specular highlights). I'm essentially "exposure compensating", but not via any special built in meter functions that do that sort of thing. I hope that makes sense...

You can choose to work in the middle and think in terms of latitude over and under if you wish. I used to do that, but given the harsh highlight ceiling RED and all digital cameras have, it seems to make more sense to work from the brightest highlight with detail on down.

Whatever method works for you, use it.

 

[Dylan Macleod CSC]

"I'm curious as to why everyone is still basing sensitivity readings off 18% gray. An important thing to remember about light meters is that they are basically "stupid" instruments."

Both reflected and incident light meters are calibrated to measure the light in a scene and produce a medium gray tone. If you spot meter a white phone it will give you an exposure (in black and white terms) that produces a gray phone.

They are "stupid". We know that this is what they will do. So we use things like the zone system to our advantage. We know how to make the phone look white by interpreting the "stupid" meter.

A waveform or histogram represents the entire tonal range. Monitoring the peak whites as your basis for determining the tolerance of exposure is absolutely the right way to do it.

In my original post I mentioned gray scale, but pointed out that I based my "normal exposure" on where the white (and the black) chip on that scale landed.

I agree. Monitoring where the 18% gray falls is less useful than watching "highlights".

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

 

[Dylan Macleod CSC]

"Exposing the RED image properly is not rocket science. The only issue to watch for is bright highlight detail when you are worried about it, like when it looks too clippy on the monitor -- and there are workarounds to check it (worst case scenario, open up the file on set on a laptop if you are really worried. Otherwise, set the ASA rating to around 250 to 320 and use the waveform, false color reading, etc.)

Exposing so it looks good on the monitor and knowing you have a little more information to play with in post using the RAW file is not the worst method either. We often have more information on a film negative than we end up using in the final image."

From what I understand from this post - there is no other way to really "know" what you are getting, other than to open the file in REDCINE and see what THAT histogram has to say?

That is a bit of a drag, because on those rare occasions where you don't have time to do that it would be good to have a way of knowing what the exposure tolerance is....beyond what the waveform is telling you.

The waveform has always been a bit of a "holy grail" in the sense that when it says it is clipped at 120...you KNOW there is no information on the tape.

It is just interesting to discover that, in fact, the signal the RED sends out is not the whole story...

An interesting gray area to explore.

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

 

[David Mullen ASC]

Maybe it's just me... but I shot eight features in HD without any scopes or waveforms and they came out fine, exposure-wise. I've spent the past several years trying to stop being so anal-retentive about exposures. Sure, I don't want to make mistakes, but there's going to be a limit to how much I effort I want to spend thinking about exposure. My natural tendency is to be over-prepared and over-precise, so I'm trying to be much looser about it, do more by eye, etc.

 

[fightordie]

Maybe it's just me... but I shot eight features in HD without any scopes or waveforms and they came out fine, exposure-wise. I've spent the past several years trying to stop being so anal-retentive about exposures. Sure, I don't want to make mistakes, but there's going to be a limit to how much I effort I want to spend thinking about exposure. My natural tendency is to be over-prepared and over-precise, so I'm trying to be much looser about it, do more by eye, etc.

Beautiful. A breath of fresh air David.

 

[JD Holloway]

Hi Dylan,

The 1080p record firmware build will probably help with this. The 720p output will simply be a rescaled 1080p recorded image...But nobody anal enough to use a vectorscope will run this build IMHO...

Hope things are well,

James

 

[Dylan Macleod CSC]

Maybe it's just me... but I shot eight features in HD without any scopes or waveforms and they came out fine, exposure-wise. I've spent the past several years trying to stop being so anal-retentive about exposures. Sure, I don't want to make mistakes, but there's going to be a limit to how much I effort I want to spend thinking about exposure. My natural tendency is to be over-prepared and over-precise, so I'm trying to be much looser about it, do more by eye, etc.

I hear ya! Don't get me wrong. I forgot my meter on a job the other day (a career first-I had just come off of an HD job!). It was a breath of fresh air. I lit by eye, then grabbed my gaffers meter to set the stop...I know what you mean! :biggrin:

What I am trying to get at is the "learn the rules first" part of the saying..."learn the rules first, then break them". I want to know what the RED's sensitivity is so I can then "play". I am sure it took you a few shoots, looking at a waveform and a calibrated monitor, then doing your final color corrections to get to the comfort level that you are describing.

I am trying to figure out how to do that with RED.

It just seems that it is going to be more difficult ascertaining what the true sensitivity of the RAW data is - "live". What I am trying to figure out is if it is possible to "measure" the signal "live". It seems, from this thread, that it is not possible.

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

 

[Dylan Macleod CSC]

Hi Dylan,

The 1080p record firmware build will probably help with this. The 720p output will simply be a rescaled 1080p recorded image...But nobody anal enough to use a vectorscope will run this build IMHO...

Hope things are well,

James

Hey James

Life is good. I am getting my RED soon!

Do you think this will help get at a more accurate reading of what the RAW data is recording? Again, from this thread, I don't think it will. It will just be a higher resolution image to look at.

I hope you are well too.

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

 

[Rich Schaefer]

This is an intelegent thread! Good stuff Guys.

Dylan, The answer is test it until your comfortable!

I think of my RED's sensor like a DSLR sensor. To me the holly grail is the Histogram. I have tested the 720p output and I can use a waveform effectively but I know Black is 0 and clip is 109.

I am an engineer's nightmare because I love to slam the clippers if it looks good! The RED has no Knee so it is more unforgiving is this regard so, test how she handles the highlights. If you think it looks good do it! I know you know this, you are being paid for your eye!

I understand the history of the 18% card, but I agree more with Ansel Adams in that: find out what your capture media can do and center your scene into that window. The way you handle highlights is personal tast/art.

Regarding rating RED I set my camera to 320. The internal meters are not accurate at lower isos and she gets noisy in higher isos.

Like Jim, I reserve the right to change my opinion on this at any time, due to me doing more testing!

Cheers,
Rich

 

[David Mullen ASC]

My feeling is, pick an ASA rating that you feel is practical, shoot some tests, decide if you like the way it looks, and if not, adjust your rating.

Personally, even if someone proves, let's say, that 125 ASA is the "optimum" rating for the RED, it's not like that's a practical choice for most interior applications; I'm still going to have to figure out how to make a more reasonable ASA rating work.

So if you think that 320 ASA is what will work for you, I'd test that and see where the limitations happen, what the look is.