ISO of Red

[BigLu]

My Rule of thumb for shooting to all my DPs.

Indoors rate at ISO 200 or 250 but you must stay cautious of protecting your highlights.
Out doors 250 or 320 also keeping in mind those highlights.
Depending on the shot or scene and how it is lit will dictate what I recommend and how to help improve the quality of the shot.
Other factors that come into play is shooting Daylight or Tungsten.
The daylight or if shooting HMI is more forgiving in the noise when slightly under exposing.


Results are pleasing.

 

[Graeme Nattress]

Another very important point is light meter accuracy. The best light meter for the RED is the RED itself - raw false color view mode, bar graph and histogram.

I've just shot some charts and my light meter spot on the grey chip of the macbeth said 70cd/m2, but my highly accurate calibrated spectroradiometer says it was actually 56.79cd/m2. That's like 20% different!

I know that if you rate for 320 you have a good starting point for low noise and great highlight protection. If you wish to differ from that, get really comfortable and familiar with the RAW metering in camera and make sure that those meters are what you trust when it comes to measuring the light hitting the sensor.

Graeme

 

[Lauri Kettunen]

People still need a way to tie by to it visually

Perhaps one shoud precisely give up of that.

Let me give an example and use Red One as the reference/nominal level of sensivity. Let's rate Red One by saying its highlight sensitivity is 5.6 or 320 (the reason for this becomes apparent in a moment) and that its dynamics is 11 stops. (Whatever is the correct dynamics is irrelevant.)

Next, it has been said that one can set Epic-X to ISO 800 and get the same results as what one gets with Red One with ISO set to 320. However, this does not say anything about the sensitivity of Epic-X. That is, if Red One clips at level 5.6 (or 320), will Epic-X clip at the same 5.6 or already at somewhere around 9 (or 120, resp.). Effectively this means, that if a certain amount of light makes Red One to clip at stop f/5.6 (or at ISO 320), Epic-X clipped already at f/9.0 (ISO 120, resp.) and thus had a more sensitive sensor. But, saying Epic-X ISO 800 corresponds to Red One ISO 320 may mean as well that Epic-X has the same sensitivity but has just 1 1/3 stops of more dynamics.

Putting all together, if the highlight clipping is rated with respect to some nominal level and also the dynamics of the sensor is given, that pretty much characterizes the sensor by telling the sensitivity and distance to noise. Everything between is linear and thus also the mid grays are in the eye of the beholder.

Those who like to talk in terms of real world tests, the question is, if you shoot a (mid) gray cat againts a (mid) gray wall, how do you set the exposure? To squeeze the best out of the camera, one should exposure just below clipping. Mid grays get a meaning only in post.

 

[Eric Haase]

Another very important point is light meter accuracy. The best light meter for the RED is the RED itself - raw false color view mode, bar graph and histogram.

I've just shot some charts and my light meter spot on the grey chip of the macbeth said 70cd/m2, but my highly accurate calibrated spectroradiometer says it was actually 56.79cd/m2. That's like 20% different!

I know that if you rate for 320 you have a good starting point for low noise and great highlight protection. If you wish to differ from that, get really comfortable and familiar with the RAW metering in camera and make sure that those meters are what you trust when it comes to measuring the light hitting the sensor.

Graeme

Frequently I will have a rigging crew working a set or two ahead and have to give instruction on light levels and ratios. It's impractical to have the rigging crew include a camera department for the sole purpose of lugging a Red around to use as a light meter. So for me, it's important to be able to know what I will be getting out of the camera at certain predetermined light levels. It's always been disconcerting that rating the camera at 320 on my meter and setting the metadata to 320 does not give the expected results in false color. This is the issue David Mullen was talking about.

 

[CJ Roy]

If we have our gaffers light to their meters at 320, we'll end up with an underexposed and usually noisy image. For two years now we've known this issue and we've been compensating by lighting for a slower stock, which yields much better and beautiful results. This was the case before Redspace was invented, and it's the same now that Rec 709 has more accurate color.

It's impractical to always have the camera setup for lighting. Is Red's official position to rate the camera @ 320, and for everyone's meters to be set to 320?

Thanks.

-CJ

 

[Bill Anderson]

Old film references are often the cause of a bad decision.

Graeme's point should be re-read by all.

Shooting ISO 160 is not a good idea except in low contrast situations.

Shooting ISO 320 will give you the best starting place for highlight protection and low noise for a RED ONE under most conditions.

ISO Rating is a good term. "Native ISO" is a very dangerous term.

From Asia...

Jim

OK, playing the devil's advocate, and still working with the
remnants of an old film brain - although I get on perfectly
with my DSLR- why would one choose 320 because it has
highlight protection? After a typical period of new gear
acclimatizing, who needs highlight protection with all these
accurate RED scopes and meters? A "stop to a stop and a
third protection" isn't highlight protection, that's underexposure.
If I understand this correctly: choices are good- but recommended
"ISO" should be bang on, and the highlight protection ISO (320) choice
should be a training wheels sort of alternative, one that is not quite
optimal- ??????

 

[Bill Anderson]

Another very important point is light meter accuracy. The best light meter for the RED is the RED itself - raw false color view mode, bar graph and histogram.

I've just shot some charts and my light meter spot on the grey chip of the macbeth said 70cd/m2, but my highly accurate calibrated spectroradiometer says it was actually 56.79cd/m2. That's like 20% different!

I know that if you rate for 320 you have a good starting point for low noise and great highlight protection. If you wish to differ from that, get really comfortable and familiar with the RAW metering in camera and make sure that those meters are what you trust when it comes to measuring the light hitting the sensor.

Graeme

Oops, Graeme answered that perfectly. We're back on track lads ( :

 

[Dan Hudgins]

Not ASA, Not ISO, use EI

Not ASA, Not ISO, use EI

Before when you use using ASA to refer to exposure values in the American system I mentioned that ASA was a film exposure standard based on the foot of film's curve and was not the right way to refer to a sensor's exposure.

Since then you now seem to be using ISO, but as pointed out the ISO "standard" would "produce blown out highlights" as most video cameras have, and that ISO is for "processed" data?

What you should have been using all along was EI. Kodak changed from ASA to EI when they went off the ASA standard to give higher film speeds on some of there high speeds stocks, speeds with no underexposure latitude.

EI stands for Exposure Index, which is what you are giving in the META-DATA you are giving the camera and Exposure Index, you are not changing the sensors reach for light like push processing would. So just call it EI from now on and people will not argue about "standards" since there is nothing "standard" about the way REDCODE works.

On the point about flesh tone being a patch on the color chart, that was not my point, my point was that specular reflections off oil and sweat on actors skin would be one, two, or more stops ABOVE 90% white card, so if you do not want to clip the "flesh" tone you need to stop down one or two stops past the white card being at 100% signal (12 bits high from the ADC).

Also since the green clips before the red and blue, you need one or two stops less light for the green sensor to not clip the green before white balance. Under other K light the red can clip before the green, so the EI needs to be larger so that you stop down enough to avoid cliping ANY of the colors on specular reflections on faces and other semi-gloss subjects.

Also rim-lighting is hot at least two stops, so you do not want the sensor in clip, since that can cause off-color fringe on the rim-lighting and other blooming artifacts do to the un-even clipping of the primary colors in a Bayer sensor, you want data for all three colors in the specular reflections and the rum-lighting etc.

 

[Sanjin Jukic]

Before when you use using ASA to refer to exposure values in the American system I mentioned that ASA was a film exposure standard based on the foot of film's curve and was not the right way to refer to a sensor's exposure.

Since then you now seem to be using ISO, but as pointed out the ISO "standard" would "produce blown out highlights" as most video cameras have, and that ISO is for "processed" data?

What you should have been using all along was EI. Kodak changed from ASA to EI when they went off the ASA standard to give higher film speeds on some of there high speeds stocks, speeds with no underexposure latitude.

EI stands for Exposure Index, which is what you are giving in the META-DATA you are giving the camera and Exposure Index, you are not changing the sensors reach for light like push processing would. So just call it EI from now on and people will not argue about "standards" since there is nothing "standard" about the way REDCODE works.

On the point about flesh tone being a patch on the color chart, that was not my point, my point was that specular reflections off oil and sweat on actors skin would be one, two, or more stops ABOVE 90% white card, so if you do not want to clip the "flesh" tone you need to stop down one or two stops past the white card being at 100% signal (12 bits high from the ADC).

Also since the green clips before the red and blue, you need one or two stops less light for the green sensor to not clip the green before white balance. Under other K light the red can clip before the green, so the EI needs to be larger so that you stop down enough to avoid cliping ANY of the colors on specular reflections on faces and other semi-gloss subjects.

Also rim-lighting is hot at least two stops, so you do not want the sensor in clip, since that can cause off-color fringe on the rim-lighting and other blooming artifacts do to the un-even clipping of the primary colors in a Bayer sensor, you want data for all three colors in the specular reflections and the rum-lighting etc.

Dan,

you are totally right.

Also ARRI Digital web page when talking sensitivity and ASA refers that is BASED ON EI (Exposure Index).

So back to the school again:

"Current ISO system

The current International Standard for measuring the speed of colour negative film is called ISO 5800:1987[1]
from the International Organization for Standardization (ISO). Related standards ISO 6:1993[2] and ISO 2240:2003[3]
define scales for speeds of black-and-white negative film and color reversal film. This system defines both an arithmetic
and a logarithmic scale, combining the previously separate ASA and DIN systems. [4]

In the ISO arithmetic scale, corresponding to the ASA system, a doubling of the sensitivity of a film requires a doubling of
the numerical film speed value. In the ISO logarithmic scale, which corresponds to the DIN scale, adding 3° to the numerical
value that designates the film speed constitutes a doubling of that value. For example, a film rated ISO 200/24° is twice
as sensitive as a film rated ISO 100/21°.[4]
Commonly, the logarithmic speed is omitted, and only the arithmetic speed is given; for example, “ISO 100”.[5]


Exposure index

Exposure index, or EI, refers to speed rating assigned to a particular film and shooting situation in variance to the film's actual speed.

It is used to compensate for equipment calibration inaccuracies or process variables, or to achieve certain effects.

The exposure index may simply be called the speed setting, as compared to the speed rating.

For example, a photographer may rate an ISO 400 film at EI 800 and then use push processing to obtain printable negatives in low-light conditions.
The film has been exposed at EI 800.

Another example occurs where a camera's shutter is miscalibrated and consistently overexposes or underexposes the film; similarly,
a light meter may be inaccurate.

One may adjust the EI rating accordingly in order to compensate

for these defects and consistently produce correctly exposed negatives."

LINK>>>


Now you could all that "translate" to your RED1 shooting situation.

Also follow precisely the recommendation or suggestion that Graeme Nattress wrote about the R1 exposure topic.

 

[David Mullen ASC]

I know all of this, but my question is: if RED recommends 320 ASA and you set the camera to 320 ASA, why do the False Colors recommend that you overexpose the key by a 1/2-stop compared to a light meter?

So does RED recommend I believe the False Colors, in which case the camera is behaving more like 200-ish ASA in terms of the key light on a Caucasian face, or that I treat the camera as if it were 320 ASA and not believe the False Colors? My question is not what the "native" sensitivity is, but why, when the camera is set to 320 ASA, that False Colors don't correspond to that rating? I'm particularly referring to the Caucasian fleshtone "Pink +0.3" false color, it's darn near impossible to get pink to show up in a Caucasian face unless you open up the iris a 1/2-stop compared to what an incident meter reading gives you at 320 ASA.

I just want to know more about False Color, because this was the first time I tried using it and I found it to keep recommending that I overexpose compared to my meter readings at 320 ASA.

 

[Sanjin Jukic]

I know all of this, but my question is: if RED recommends 320 ASA and you set the camera to 320 ASA, why do the False Colors recommend that you overexpose the key by a 1/2-stop compared to a light meter?

So does RED recommend I believe the False Colors, in which case the camera is behaving more like 200-ish ASA in terms of the key light on a Caucasian face, or that I treat the camera as if it were 320 ASA and not believe the False Colors? My question is not what the "native" sensitivity is, but why, when the camera is set to 320 ASA, that False Colors don't correspond to that rating? I'm particularly referring to the Caucasian fleshtone "Pink +0.3" false color, it's darn near impossible to get pink to show up in a Caucasian face unless you open up the iris a 1/2-stop compared to what an incident meter reading gives you at 320 ASA.

I just want to know more about False Color, because this was the first time I tried using it and I found it to keep recommending that I overexpose compared to my meter readings at 320 ASA.

David,

also I still have to make extensive false color tests with skin tones (hopefully on weekend) but there are some guys here that are pretty experienced with it:

"We set our light meters to 250 ISO all the time to be on the safe side and verified the exposure using mostly false colors.
We tried to light to a 2.8 1/2 - 2/3rds on ISO 250 all the time so if we run the lens at 2.8 the key would be that much over. We left the camera
on it's default settings though and shot everything on 5000K as well. Colorwise we lit to the image we saw at 5000K, as we thought this would be
the most efficient workflow to follow through by getting what we wanted into camera as much as possible.

We had a directors monitor (cheap dvi lcd) and for some stuff we had an Astro with waveform on set, which we used a couple of times,
but we're quite used to the false colors now after 3 features, so that was more of a reference than the waveform.

What also helped is that we graded a project completely ourselves in Apple Color before that so we know what we can do with the RED material
and what we can't. Obviously those are the nice shots but there are of course a few underexposed things as well where you get rushed by schedule
and other stuff and just can't add a light after a bit of staging changed.

I think that we've found our sweet spot for a skin-keylight to be 1/2 - 2/3rds a stop over, which is pink anyways on the RED false colors since build 20.
So as soon as we had spots of pink and green on the key side of the faces we were happy . No red no purple as well

We're quite happy to see that we get more and more shots where we nail the exposure, we used to have lots of underexposed shots when relying on a monitor instead of false colors."


XiaoSu Han a.k.a Xax, a member of this forum>>>

 

[Shawn Nelson]

I know all of this, but my question is: if RED recommends 320 ASA and you set the camera to 320 ASA, why do the False Colors recommend that you overexpose the key by a 1/2-stop compared to a light meter?

So does RED recommend I believe the False Colors, in which case the camera is behaving more like 200-ish ASA in terms of the key light on a Caucasian face, or that I treat the camera as if it were 320 ASA and not believe the False Colors? My question is not what the "native" sensitivity is, but why, when the camera is set to 320 ASA, that False Colors don't correspond to that rating? I'm particularly referring to the Caucasian fleshtone "Pink +0.3" false color, it's darn near impossible to get pink to show up in a Caucasian face unless you open up the iris a 1/2-stop compared to what an incident meter reading gives you at 320 ASA.

I just want to know more about False Color, because this was the first time I tried using it and I found it to keep recommending that I overexpose compared to my meter readings at 320 ASA.

Bingo!!! Using false color makes the camera behave at ISO 160 is what I found as well. And since False Color is the easiest tool to use, it's what I use. So...it's a conundrum.

 

[Bill Anderson]

Not to highjack David's query- I'm sure we'd all like to hear that answer:
Is the false color behavior, as per David's query, linear?
Is it constant across a range of ISO's?

 

[Bill Anderson]

Sanjin, that doesn't answer David's post, it simply gives someone's workaround. As valuable as that might be, I think DM has that part well covered. The question is why is this workaround, this "1/2 to 2/3 stops" compensation necessary when false colors are there to provide an accurate reference in the first place?

EDIT: or, to be more accurate, read DM's post again.

 

[Ryan E. Walters]

Old film references are often the cause of a bad decision.

Graeme's point should be re-read by all.

Shooting ISO 160 is not a good idea except in low contrast situations.

Shooting ISO 320 will give you the best starting place for highlight protection and low noise for a RED ONE under most conditions.

ISO Rating is a good term. "Native ISO" is a very dangerous term.

From Asia...

Jim

Thats why I tried to differentiate between shooting and metering. I've been shooting with my camera set to ISO 320- however in order to get my meter to correspond to what False Color says I have to adjust the meter to around ISO 160- which means I either need more light or I end up underexposing the skin tones according to false color ...

 

[Ryan E. Walters]

Ideally everything I do will respond to the ISO setting. So if I am at 160 - I'll light the image to look good at that setting. This ensures that I have the room to grade things with some flexibility. If we need to push things up a little the noise floor has been "exposed up" a little to accommodate that.

Reality is - it isn't always practical to light a night scene to 160 ISO - but theoretically - if you can do it - you are favoring more info in the shadows. Which is generally what you want in night scenes. Highlights are less of an issue - but need to be checked (i.e. when setting practicals) knowing that you are rating the camera to "expose more".

Short answer is - everything would get processed with the in camera rating of 160 ISO because that is what I have lit for and set the stop for.

Interesting- it looks like I need to do some more experimenting with this ...

It was my understanding that the sensor performs the best when set at ISO 320, and that processing the image at that setting delivered the best results. So while I would meter, rate, and expose for a different ISO depending on the scene at hand, I would process at ISO 320. I think that may be the source of some of my confusion. That, and the fact that false color does not match what my meter says at the given ISOs...

 

[Eric Haase]

I know all of this, but my question is: if RED recommends 320 ASA and you set the camera to 320 ASA, why do the False Colors recommend that you overexpose the key by a 1/2-stop compared to a light meter?

So does RED recommend I believe the False Colors, in which case the camera is behaving more like 200-ish ASA in terms of the key light on a Caucasian face, or that I treat the camera as if it were 320 ASA and not believe the False Colors? My question is not what the "native" sensitivity is, but why, when the camera is set to 320 ASA, that False Colors don't correspond to that rating? I'm particularly referring to the Caucasian fleshtone "Pink +0.3" false color, it's darn near impossible to get pink to show up in a Caucasian face unless you open up the iris a 1/2-stop compared to what an incident meter reading gives you at 320 ASA.

I just want to know more about False Color, because this was the first time I tried using it and I found it to keep recommending that I overexpose compared to my meter readings at 320 ASA.

If you meter for 320 then you are underexposing the image and bringing it up in post. Graeme has indicated that RED recommends this setting so you have highlight protection. False color won't match because you are in fact underexposing, only to grade it later. Why doesn't 320 match up on the camera to your meter and with false color? Because the camera is slower than 320.

Why was the number 320 chosen as the camera metadata setting that closely resembles the light levels in RAW (when viewed in rec709 B20)? It seems like 200 should be the number assigned to what is currently 320. Then if you set the camera on 320, the image would be brightened from what is on the sensor and would match your meter and false color. I suspect the reason the number 320 was chosen was a marketing one, but maybe there is a better technical explanation.

 

[Florian Stadler]

I find 250 for Tungsten and 320 for daylight on my meter pretty accurate. I do use REDSPACE and not Rec709 for monitoring and dailies processing. (I don't like the skin tone rendition in 709, even in the 20+ builds, although highly improved).
I think key is using Redspace. Rec709 rides way too close to RAW View exposure wise to protect highlights imho.

 

[Dylan Macleod CSC]

Frequently I will have a rigging crew working a set or two ahead and have to give instruction on light levels and ratios. It's impractical to have the rigging crew include a camera department for the sole purpose of lugging a Red around to use as a light meter. So for me, it's important to be able to know what I will be getting out of the camera at certain predetermined light levels. It's always been disconcerting that rating the camera at 320 on my meter and setting the metadata to 320 does not give the expected results in false color. This is the issue David Mullen was talking about.

Based on the some of the conclusions in this thread - have your rigging crew put their light meter at 160 ISO and when you get your camera over there (at 320) you should be fine, no?

 

[Eric Haase]

I find 250 for Tungsten and 320 for daylight on my meter pretty accurate. I do use REDSPACE and not Rec709 for monitoring and dailies processing. (I don't like the skin tone rendition in 709, even in the 20+ builds, although highly improved).
I think key is using Redspace. Rec709 rides way too close to RAW View exposure wise to protect highlights imho.

Yes REDSPACE really lifts things way up. Check it out:

http://dl.dropbox.com/u/904379/Build 20 Tests Round 2.pdf

You can see the relationship between REDSPACE, REC709, and RAW. Your meter will more closely match if you use REDSPACE.

But REDSPACE will show you red false color when the RAW reading is really only 90-93. To me that's a huge difference and means you are not taking full advantage of the latitude of the sensor. I've found footage shot and metered with false color in REDSPACE to have less flexibility in grading since it is more underexposed than if REC709 were used to shoot and meter with false color.