real native iso of dragon?

[Brian Boyer]

Rate a DRAGON at ISO 320 and set the lens T5.6 and rate a DRAGON at ISO 1280 and set the lens to T5.6 and you will be recording the EXACT same image - because the same amount of light is hitting the sensor. The images will be, by default of the camera generated meta data, processed differently - but the RAW file is the same.

BUT choosing an ISO does effect your EXPOSURE METHOD - which is whatever method you decide to use to choose your exposure. Lightmeter, histogram or just looking at the monitor.

Mark, perhaps I'm getting hung up on semantics and I'm sorry to nitpick but I was thrown a little by your use of the word "rate." In my mind, a rating is based on a standard of measure. In this case, that standard would be middle gray. Whenever I've heard a DP talking about rating a camera or film stock, I've always associated that with using a light meter because they were actually taking a reading.

Shooting a scene at ISO 320, T5.6 and then bumping the ISO to 1280 at T5.6 on the same scene does produce the exact same image. But can you really call that rating the camera?

Setting a camera to a given ISO and eyeballing the rest based on the histogram and/or monitor is inexact and inconsistent, and so doesn't seem to qualify as a rating to me.

Maybe that was a distinction I shouldn't have made?

To my thinking, rating Dragon for ISO 320 at T5.6 using a light meter would yield a different image from rating at ISO 1280 because the light meter would instruct you to alter your T-stop to compensate for the ISO change. The only way to get the exact same image would be to ignore the light meter and keep the T-stop at 5.6.

I've noticed other Redusers also reserve the term "rate" for referring to light meter readings. Maybe "rate" has a broader usage I wasn't allowing for. Even if it does, I think I prefer it with the narrower association to a light meter.

If we can agree that rating your camera for an ISO is a more precise endeavor than simply setting your camera's ISO and eyeballing the rest, I guess I would count "rating a camera with a light meter" as one of the exposure methods you mentioned and reword your ISO 320 example to read "Shoot a Dragon..." in place of "Rate a Dragon..."

What do you think?

 

[Marc Rochon]

yah I feel like were beating this to death. shooting at 32O ISO and then shooting at 1280 ISO is NOT the same in my mind. In my mind if I do that I have put a faster film in my camera and that will yield different results. I don't plan on pulling out my old light meter anytime soon so if I have a good histogram and the image looks good than I'm good.

ISO low smoother. ISO higher brings grain.

Give this sensor/film enough light and it will do amazing things.

 

[David Battistella]

FWIW. When I use rate I mean this.

My meter is set to that ISO and I create light ratios based on that ISO

to be honest here is what I look at more than anything.

shutter angle
T/F stop
left goalpost
right goalpost
stoplights
the onboard monitor

If those look "right" at 320-800ISO, then it's all pretty much ok. Anything above and I am using ISO to "see", when I am doing this then I know i will be doing something to recover an underlie image in situations I can not control.

The base ISO is the ISO you chose build your ratios on between about 250-2000 ISO. at these ISO's the sensor is capturing the same DR. But how you do the things above will determine how you are spreading that DR above and below middle grey. Many situations, many changes in light and creative choices can affect your choices and should.

One thing I will say is that with Dragon I am not at all as worried about highlight protection or letting things blow out as I was on the MX sensor, MUCH smoother rolloff in the top end, especially with the STH, which is back in my camera now.

I agree with Mark on his LL assessment of the OLPF's

Battistella

 

[Brian Boyer]

yah I feel like were beating this to death. shooting at 32O ISO and then shooting at 1280 ISO is NOT the same in my mind. In my mind if I do that I have put a faster film in my camera and that will yield different results. I don't plan on pulling out my old light meter anytime soon so if I have a good histogram and the image looks good than I'm good.

ISO low smoother. ISO higher brings grain.

Give this sensor/film enough light and it will do amazing things.

To be clear, my comment was not about ISO so much as it was about the terminology used to describe a process.

When someone says "rate" I think light meter and middle gray. If that's not necessarily the case, I'd like to know. I'd prefer rate to imply light meter usage but, if that's not how it's generally understood, I want to adjust my thinking accordingly.

 

[David Battistella]

Give this sensor/film enough light and it will do amazing things.

good summary. I've been saying this for many moths it seems.

 

[Mark L. Pederson]

Mark, perhaps I'm getting hung up on semantics and I'm sorry to nitpick but I was thrown a little by your use of the word "rate." In my mind, a rating is based on a standard of measure. In this case, that standard would be middle gray. Whenever I've heard a DP talking about rating a camera or film stock, I've always associated that with using a light meter because they were actually taking a reading.

If more people used light meters - we'd see a lot better exposures

Shooting a scene at ISO 320, T5.6 and then bumping the ISO to 1280 at T5.6 on the same scene does produce the exact same image. But can you really call that rating the camera?

I agree with you - an ISO "RATING" is the ISO you choose for your EXPOSURE METHOD - what you plug into your light meter if your were using one.

The issue that so many people seem to have an issue wrapping their head around - is that changing the ISO setting ON THE CAMERA does not change the image CAPTURE.

It DOES change the IMAGE PROCESSING - and of course - it DOES effect your EXPOSURE METHOD.

Setting a camera to a given ISO and eyeballing the rest based on the histogram and/or monitor is inexact and inconsistent, and so doesn't seem to qualify as a rating to me.

AMEN.

If we can agree that rating your camera for an ISO is a more precise endeavor than simply setting your camera's ISO and eyeballing the rest, I guess I would count "rating a camera with a light meter" as one of the exposure methods you mentioned and reword your ISO 320 example to read "Shoot a Dragon..." in place of "Rate a Dragon..."

What do you think?

AMEN.

 

[Robert Ruffo New]

The correct ISO is whatever gives you the results you like, anything else is just academic. There is a point where you can oversimplify things when explaining how things work.

Agreed - in the end the ISO you light, ND and/or iris for is an artistic choice and trade-off, with many complex aesthetic implications and even suggested meanings and story tones that come form different ISO looks. But I think it is important to clarify the mechanics behind it for some folks.

Here's my way of explaining it to people who rent our Red cameras:
It's not like picking film stocks as nothing is permanently chosen as an ISO, and you have the exact same choices for ISO on set as in post. I think rather than "shooting at" we should say "preview at" and "meter for" a given ISO. If you meter at a high ISO (and thus will tend to give the sensor less light), obviously this will give a different look, when you later fine tune the ISO or apply curves in the grade, than if you meter for a low ISO, and thus chose to give the sensor more light during the shoot. All ISOs have advantages/unique charms and there is no absolute correct answer, and there is no real native ISO because all light signals are processed by the computer inside the camera and in the later grading computer, no matter what. That said, previewing/metering/lighting/ND-ing/iris-ing for low ISOs will tend to protect darks more, and doing so for high ISOs will have a tendency to protect highlights more, but you can be careful of both ends at any ISO, or choose to not be, and even this is not exactly a steadfast rule.

With the rise of "Vimeo cinematographers" and the wrongful notion that no one needs film school (or any school) too many things are getting over-simplified. There are no magic bullets in cinematography, and the more I learn the more I find myself realizing how much I still don't know. Even something a seemingly simple as choosing an ISO on a Red as a baseline for exposure choices is not really as simple as it may appear at a first, superficial glance.

 

[Les Hillis]

There was a discussion here on Reduser a while back about renaming the in-camera 'ISO' label to something that was more RED camera specific and more explicitly descriptive of what that 'ISO' label is actually referring to when seen in camera. Could be worth revisiting the idea, maybe as a potential option in the camera menu's.

Seems to me on one hand the 'ISO' label kind of fits and is a good shorthand descriptor of something people are generally familiar with, but on the other hand that same familiarity results in ongoing confusion about what 'ISO' actually means in the specific context of shooting with a RED camera.

I'm not personally fussed by it either way, but it does seem strange when I think about it. It's kind of like film 'ISO' (or 'ASA'), but not really. And it's also kind like digital 'ISO' gain/sensitivity, but we're shooting RAW (mostly, I assume), so again, not quite the same.

And the whole subject is actually more about how to use the information that's being labelled and quantified anyway, so who cares, when ultimately the names and numbers don't matter beyond the practical use you find for them.

And of course, if seriously considering such an option (a menu option to ditch the 'ISO' label and numbering), you'd have to factor in how changing something people are familiar with could itself cause confusion, and that you still have to fit into a larger media ecosystem.

But honestly, if you were starting from scratch, without all that history, would you use the term 'ISO' and that set of numbers? Couldn't there be a better, more helpful and instructive way of labeling and quantifying what it is that's being measured and adjusted, when using a RED camera and its particular way of recording?

 

[David Mullen ASC]

This is getting a bit silly, rename "ISO 800" into, what? PROCESSING INTENSITY VALUE (PIV) 800? Do we really need one more label after ASA, ISO, and EI that means the same thing? And if you change the numerical values for a whole new system, you'd need to build and sell new light meters that used the new system. Besides, every other digital camera manufacturer, still and cinema, is using the current nomenclature and values.

This is where getting too hung up on semantics and definitions becomes impractical, when all that really needs to happen is for people to learn to expose an image. It can't be that hard, once you make a few mistakes because you underexposed too much, you probably will stop doing it.

 

[David Mullen ASC]

The real issue, that has always existed, is the conflict between what you have and what you want, and the mistake of confusing the two. What you have, let's say, is some camera where testing has revealed that the sweet spot is a rating of ISO 400... but what you need and/or desire is a rating of ISO 3200. People will always run into that wall, between reality and desire, and will deal with it in different ways. Some will be strict and shoot at ISO 400 no matter what, some will shoot at ISO 3200 and hope they get lucky and have to fix the footage in post perhaps, and others will find some way to compromise and find an approach they can live with.

 

[David Battistella]

I agree David.

The ISO setting on the RED camera is a basis for which to calibrate to something like a light meter. When used in this way with an ambient or spot meter, the results are consistent, predictable and acceptable.

Deploying exposure strategies and setting a meter and the camera to the same ISO even as a simple reference tool works.

Results vary based on style, testing and the depth of understanding of how the camera sensor performs under various conditions.

In the past, I have been hung up on base ISO, like red needed to state one as if it were some law.

I've discovered through testing, reading and understanding that the base ISO of dragon is what I set on the camera and what I set on the meter in the working range between 250-2000 ISO.

Are the images going to be the same if I light and rate at the extremes of this range? heck no, nor should they be.

That's the artistic part of what we do when crafting images. Like rating at 500 and processing at 800.

Film was also unforgiving until one could grasp important concepts. Mostly images are unforgiving in the absence of light and understanding.

Battistella

 

[Les Hillis]

Well yeah, people could just learn and have the discipline to expose properly, regardless.

And no, I wasn't suggesting just swapping one set of labels and numbers for another in defiance of established practices and standards.

I do think though the reliance on the use of 'ISO' is problematic, even though the way exposure information is presented in-camera is totally workable and fine by me personally.

Was just looking at the subject and asking questions in a more abstract way. Call it silly and impractical if you want, but I find it interesting.

 

[David Battistella]

Fwiw

arri don't use ISO on alexa, but EI and ASA.

Side by side they meter pretty much the same.

 

[joshua csehak]

This is where I'm confused. Admittedly my testing method is primitive and unscientific (layered sheets of ND in front of a Kino), but I'm getting the same results you are -- 13/14 stops DR. So where's the 16 stops?

 

[Phil Holland]

This is where I'm confused. Admittedly my testing method is primitive and unscientific (layered sheets of ND in front of a Kino), but I'm getting the same results you are -- 13/14 stops DR. So where's the 16 stops?

Process all of that out as REDlogFilm and REDgamma4. Also, run exposures rated from ISO 250-6400 and read those patches. The stuff past ISO 2000 is mostly there to inform you what's going to happen when you pass the tipping point of having a noisier image.

If you are using a pass through source method rather than reflected make sure you are accurately reading 1 stop increments.

If you have both OLPFs it's worth testing both of those as you can see some interesting information.

*edit* - I should add if you can start at one of two places when it comes to clipping. Either right at the point of clipping or right at the point of retention. If you have the ability to measure 1/10 through 1/3 stops that can help when finding where that last drop is.

Also, on my calibrated display I can from right to left count to 16 and see visible separating in your patches. Though I have no idea what ISO that is and all that.

 

[David Mullen ASC]

I count 14 distinct steps on my computer screen, but I suppose if I lifted the image, I may squeak out one more step in the blacks.

 

[Blair S. Paulsen]

I think a key source of confusion is that many cameras change their signal processing based on the ISO setting. RED changes the monitor/preview signal but does NOT change the signal level in the RAW r3D files. Many folks don't tip to this since the first time they open the r3d in RCX (or virtually any other tool) it displays with that same preview/monitor ISO. I'd love to do a survey and see how many people are even aware that they can simply adjust the ISO reference in processing with a click.

When I'm wearing my DIT hat, I sometimes "game" the on set process by choosing a monitoring ISO that keeps people conditioned to WYSIWYG systems from making exposure choices that put the colorist up against the wall in the grade. Frankly, it's way easier than trying to explain that ISO is part of a metadata formula that determines the development parameters applied to the RAW image data, but doesn't actually change what is - or isn't - in the data record. As Mark wisely notes, the real issue is how ISO settings impact exposure choices made by folks who don't use more traditional meter based methodology.

Cheers - #19

 

[joshua csehak]

Process all of that out as REDlogFilm and REDgamma4. Also, run exposures rated from ISO 250-6400 and read those patches. The stuff past ISO 2000 is mostly there to inform you what's going to happen when you pass the tipping point of having a noisier image.

If you are using a pass through source method rather than reflected make sure you are accurately reading 1 stop increments.

If you have both OLPFs it's worth testing both of those as you can see some interesting information.

*edit* - I should add if you can start at one of two places when it comes to clipping. Either right at the point of clipping or right at the point of retention. If you have the ability to measure 1/10 through 1/3 stops that can help when finding where that last drop is.

Also, on my calibrated display I can from right to left count to 16 and see visible separating in your patches. Though I have no idea what ISO that is and all that.

This is another thing that's confusing. Technically, I shot it at 800 ISO, but I was monitoring in RAW mode, so ISO had no bearing on it. Indeed, changing the ISO while monitoring RAW had no effect. I brought it in and assigned DRAGONcolor and REDlogFilm to it. (REDlogFilm and REDgamma4 are mutually exclusive, no?) So if I get more dynamic range at ISO 2000, it should reflect that as I move the slider in RCX, right? But it doesn't, it just brightens or darkens the image. This seems to be verified by your ISO test. THE DR in your chart seems to be the same across the board, from 250 to 2000, just shifted over. BUT: your light bulb pics tell a different story. They all seem to be exposed the same, but the higher ISOs have more DR. Or is it that you exposed them differently, and adjusted them to look the same in post?

I agree about starting at the edge of clipping, that's why I'll say 14, if I count each patch as one (vs. counting the steps). But I wouldn't be surprised if there was a flaw in my method of stacking NDs. But I definitely can't see 16.

I count 14 distinct steps on my computer screen, but I suppose if I lifted the image, I may squeak out one more step in the blacks.

I tried bringing up the gamma both on my screen and in RCX, and I can maybe squeeze out a little of 15, but not really. This is all academic of course -- the important thing is that it looks awesome, which it does. But it's kind of a fun test.

I think a key source of confusion is that many cameras change their signal processing based on the ISO setting. RED changes the monitor/preview signal but does NOT change the signal level in the RAW r3D files.

Yeah, that's huge. On the C300 for instance, a given shot at 2.8 and ISO 850 pushed 2 stops in post looks a hojillion times worse than the same shot at 2.8 and ISO 3200. And you never go below 850 ISO. Red is exactly the opposite.

I would actually rate the Dragon sensor at 1600, b/c if I switch from DragonColor to RAW at 1600, the image looks basically the same except much less saturated and with about 4 stops less contrast. But for me, 1600 is a bit too noisy, so I feel like 800 is the sweet spot. 1600 if you don't mind some noise, or 400 if it needs to be super-clean. I look at ISO as kind of a "Brightness and Noise" slider. Lower is less B&N, and higher is more. But truth be told I just got the upgrade a couple weeks ago. This all may change after I've been shooting on it for a few months.

 

[Phil Holland]

This is another thing that's confusing. Technically, I shot it at 800 ISO, but I was monitoring in RAW mode, so ISO had no bearing on it. Indeed, changing the ISO while monitoring RAW had no effect. I brought it in and assigned DRAGONcolor and REDlogFilm to it. (REDlogFilm and REDgamma4 are mutually exclusive, no?) So if I get more dynamic range at ISO 2000, it should reflect that as I move the slider in RCX, right? But it doesn't, it just brightens or darkens the image. This seems to be verified by your ISO test. THE DR in your chart seems to be the same across the board, from 250 to 2000, just shifted over. BUT: your light bulb pics tell a different story. They all seem to be exposed the same, but the higher ISOs have more DR. Or is it that you exposed them differently, and adjusted them to look the same in post?

I agree about starting at the edge of clipping, that's why I'll say 14, if I count each patch as one (vs. counting the steps). But I wouldn't be surprised if there was a flaw in my method of stacking NDs. But I definitely can't see 16.

I tried bringing up the gamma both on my screen and in RCX, and I can maybe squeeze out a little of 15, but not really. This is all academic of course -- the important thing is that it looks awesome, which it does. But it's kind of a fun test.

Yeah, that's huge. On the C300 for instance, a given shot at 2.8 and ISO 850 pushed 2 stops in post looks a hojillion times worse than the same shot at 2.8 and ISO 3200. And you never go below 850 ISO. Red is exactly the opposite.

I would actually rate the Dragon sensor at 1600, b/c if I switch from DragonColor to RAW at 1600, the image looks basically the same except much less saturated and with about 4 stops less contrast. But for me, 1600 is a bit too noisy, so I feel like 800 is the sweet spot. 1600 if you don't mind some noise, or 400 if it needs to be super-clean. I look at ISO as kind of a "Brightness and Noise" slider. Lower is less B&N, and higher is more. But truth be told I just got the upgrade a couple weeks ago. This all may change after I've been shooting on it for a few months.

I don't know all the variables you have on your side but I have now tested this on a Xyla-21, reflected, and pass through. All cameras I test go through the reflected method as it's a good way to discern captured detail versus just noise. The samples in my chart are literally the patches from that reflected test. I've also had to run this test 5 times over the course of OLPFs and Color Science enhancements. All my data syncs up pretty well and also with the 3 other gentlemen who have done Xyla-21 tests.

The light bulb shot you're referring two was exposed for each rated ISO at Middle Gray via light meter and confirmed with the in camera exposure tools. Can't really get more precise than that. This is why the highlights retain more detail in the higher ISO shots. A nice and quick way to test that is expose shots properly at ISO 250, 500, 1000, and 2000 then overlay them. It's pretty visible.

Red's "RAW VIEW" is a fixed ISO 800 REDcolor2/REDlogFilm. This is what we more or less consider as the "native ISO". For me it's actually in the 800-1280 arena. It's more or less semantics at that point as we shoot actual images of moving subjects and not charts, but the charts inform us of the hard numbers. I rolled last month at ISO 1000 for a narrative gig. This last gig which we wrap next week has been everywhere from 250-1280 due to the massive variety of content.


I'll have to fire off another one of these tests now that DRAGONcolor2 is out, which I'll execute soon and it will have to be with both OLPFs as well. I do mine with a base of Middle Gray as well as coming down from the moment of retention of highlight.

 

[David Battistella]

Just oust another way of describing raw view...


"RAW" view is a guideline.

It is
800 ISO
RC2
redlog film

that is RAW Views default curve.

The stoplights and goalposts and gioscope are telling you more about what's going on at the sensor level.

RAW view is technically not RAW, but let's you see an RLF gamma curve at 800ISO.

Just for clarity.

Battistella