Exposing + Native ISO

[Adrian Jebef]

Amen.

And again WYSIWYG is NOT the way to learn the fundamentals that will serve you better even if you decide to use WYSIWYG as an exposure method. And it's the LAST exposure method I would personally choose.

When you shoot RAW - YGMTWYS (You Get More Than What You See).

I agree. When you shoot raw, or basically full range DR, YGMTWYS.

But here's the thing: if you can implement a proper Log/LUT workflow not only do you get to shoot raw, you also get to "view raw" (RGB Log camera output) and you can even creatively develop your "look" (via a LUT on-set) all at the same time. It's kinda awesome-sauce.

 

[Gunleik Groven]

Just a side note here but we should stop referring to the ALEXA as a camera that "bakes in it's ISO". It shoots raw just like the RED cameras. We should stop assuming that it can only shoot HD ProRes. BTW: If you actually test ISO ratings on the ALEXA between ProRes Log-C and Arriraw you'll find the differences surprisingly small. That camera has a very smooth response curve.

And again: Agree.

This discussion is better without assumptions about other systems.

 

[Blair S. Paulsen]

(snip)
This is the other major thing, I think many people are using ISO as gain instead of as a basis for measuring light.
Battistella

Indeed. Based on the electronic cameras that have comprised 99% of the market over the last 30 years many shooters assume that dialing up the ISO is adding gain to the signal via amplification of voltage values - typically in front of the A-D converter.

RED doesn't use gain in camera, but not everyone gets that.

Cheers - #19

 

[Gunleik Groven]

I agree. When you shoot raw, or basically full range DR, YGMTWYS.

But here's the thing: if you can implement a proper Log/LUT workflow not only do you get to shoot raw, you also get to "view raw" (RGB Log camera output) and you can even creatively develop your "look" (via a LUT on-set) all at the same time. It's kinda awesome-sauce.

Again:
Agreed.

A while back I started a thread called "what is raw" or something to that respect, as that is what many refer to as some kind of magic.
And in the sense that we never see, touch, grade or deliver it, it is. But it is not a totally uncomprehensible "magic" aspect. It is just a way of storing non-debayered files, in REDs case, linearily, probably pre-matrix.

 

[Bill Anderson]

Back in the day, there was a story going around that went something like this: during a photography workshop Minor White, known for his 'spiritual' approach to photography, asked Ansel Adams, with a critical tone, "still using the Zone System?" to which Adams replied, "yes, are you still using the Zen System". Implying that we ALL use "systems".

 

[David Battistella]

Just a side note here but we should stop referring to the ALEXA as a camera that "bakes in it's ISO". It shoots raw just like the RED cameras. We should stop assuming that it can only shoot HD ProRes. BTW: If you actually test ISO ratings on the ALEXA between ProRes Log-C and Arriraw you'll find the differences surprisingly small. That camera has a very smooth response curve.

yes. I know. I always forget to include; "Unless shooting raw". ARRI cameras produce lovely images with excellent back end LUT support. Baked in only applies to non raw capture on Alexa. Pro res 4444 is a very decent image container.

Battistella

 

[Mark L. Pederson]

But here's the thing: if you can implement a proper Log/LUT workflow not only do you get to shoot raw, you also get to "view raw" (RGB Log camera output) and you can even creatively develop your "look" (via a LUT on-set) all at the same time. It's kinda awesome-sauce.

100% with you Adrian. It's beyond awesome.

And yeah ... even that doc camera AMIRA has 3D LUT support.

BUT - it's much more powerful if the person choosing exposure understands the basics first - as they might not have a DIT as good as you on the set

3D LUTs is "advanced class" - coming soon to Reduser and I hope you will help me with that thread.

 

[Mark L. Pederson]

RED doesn't use gain in camera, but not everyone gets that.

Cheers - #19

It DOES out the monitor path - it just doesn't record that.

 

[Adrian Jebef]

3D LUTs is "advanced class" - coming soon to Reduser and I hope you will help me with that thread.

I'll be there.

It DOES out the monitor path - it just doesn't record that.

A very important distinction for those with less experience.

 

[Scott Crawley]

A very important distinction for those with less experience.

This is especially true I think for folks like me who's native language is film based photography, particularly still photography, where the end result is a paper print that is created in an entirely secondary process.

None of the display of the image happens in camera, so I tend to draw a hard and fast line in my mind as to where exposure happens and where display happens, if you get my meaning. I sometimes need to be reminded of what is applied, where it is applied and where that is written, or fixed to the image. By that I mean my Red does record ISO data and that is part of the image, but it also is not part of the recorded image. It resides in a sidecar file that may be better described as a part of the digital lab and not part of the image per se. It is part of the print but not the negative. It just happens to be written at the same time as the negative.

It's not that I don't know or understand, but I have to translate it, so to speak, and I don't often think about it. I kind of take these basics for granted, I guess.

 

[Elsie N]

I have a question... is it necessary to use all the DR available?

 

[Gunleik Groven]

I have a question... is it necessary to use all the DR available?

Nope. Not at all. Often it is good not to. It is not a moral thing.

 

[Mark L. Pederson]

I have a question... is it necessary to use all the DR available?

One rarely does

 

[Hrvoje Simic]

Or ... you can just accept that because of subjectivity to the noise floor - it's a RANGE

I'm okay with that.
I can see what I need to see, but many people like exact figures for a reference, which is understandable.

Yes yes, but captured noise is simply an incorrectly sampled pixel. It is the result of an insufficient sampling technogy and/or insufficient exposure.

Yes, but 16 bits (or any amount of bits) are not necessarily making it "sufficient" if they are sampling noise (i.e. not usable data).
These graphs based on Nikon D610 measurements give more insight into the effect of higher ISO values.

These two graphs show the offset in camera's ability to distinguish colors.

Edit:

Oh, and we can absolutely create noise in post, in addition to the other things you describe. We do not just amplify pre-existing noise.

Yes, but we are talking about different type of "noise", the one induced by underexposure, regardless of how many bits are sampling it.

 

[Scott Crawley]

Yes, but 16 bits (or any amount of bits) are not necessarily making it "sufficient" if they are sampling noise (i.e. not usable data).

I understand your point as it applies to poor exposure. I do, but just to be picky, we do not sample noise. It doesn't exist in the world. It is created by a flawed sample. 16 bit is not enough but it is exponentially better than 12 bits. I am certain you understand this, I am just trying to clarify my position so that you or anyone can correct me if I am wrong, and all may benefit.

My current theory is that one day we will be able to sample enough bits. Some number of bits will be enough that even our most underexposed images will be free of noise. They may be free of detail and color, but theoretically they could be noise free.

These graphs based on Nikon D610 measurements give more insight into the effect of higher ISO values.

Those are great. I will give them their due consideration. Thanks!

Yes, but we are talking about different type of "noise", the one induced by underexposure...

We are, and with most of your points I agree. But as I indicated, increased sampling bits benefit us both on the capture end as well as in processing. I have to tell you I am smitten by Dragon's bits. I can't wait to wield them, but it's going to be a while for me unfortunately.

 

[D Fuller]

This is especially true I think for folks like me who's native language is film based photography, particularly still photography, where the end result is a paper print that is created in an entirely secondary process.

None of the display of the image happens in camera, so I tend to draw a hard and fast line in my mind as to where exposure happens and where display happens, if you get my meaning. I sometimes need to be reminded of what is applied, where it is applied and where that is written, or fixed to the image. By that I mean my Red does record ISO data and that is part of the image, but it also is not part of the recorded image. It resides in a sidecar file that may be better described as a part of the digital lab and not part of the image per se. It is part of the print but not the negative. It just happens to be written at the same time as the negative.

It's not that I don't know or understand, but I have to translate it, so to speak, and I don't often think about it. I kind of take these basics for granted, I guess.

My recent epiphany is that the translation is more direct than I had thought. I really find the processing of the r3d very much like printing. And there are two steps that are analogous to developing and printing. The raw settings (what you would use in RedCine, or in the Camera Raw settings of Resolve) work to give you the base image you planned for at the time of exposure. And the RGB manipulation, with power windows and sharpening, diffusion, etc. is a lot like the printing step. The sidecar file seems to me to be analogous to a really good set of notes about what you did at the time of exposure. You can change all of those assumptions in the development stage except for the actual exposure decision. In that, too, it is more like film shooting than not.

 

[Mark L. Pederson]

My recent epiphany is that the translation is more direct than I had thought. I really find the processing of the r3d very much like printing. And there are two steps that are analogous to developing and printing. The raw settings (what you would use in RedCine, or in the Camera Raw settings of Resolve) work to give you the base image you planned for at the time of exposure. And the RGB manipulation, with power windows and sharpening, diffusion, etc. is a lot like the printing step. The sidecar file seems to me to be analogous to a really good set of notes about what you did at the time of exposure. You can change all of those assumptions in the development stage except for the actual exposure decision. In that, too, it is more like film shooting than not.

RAW is so very, very much like "digital FILM"

 

[D Fuller]

Personally, I think both of us "get it". We've probably both spent enough time with film and digital formats that we're now accustomed to exposing them. The bigger issue is those with less of a background or simply with less experience. Explaining the technical and creative concepts of proper or intentional exposure is very important. Saying that so and so digital camera is amazing and has so much DR and shoots RAW and has a usable ISO range of 200-2000 misses the point. Describing a camera this way makes it seem that there are no compromises when one shoots RAW at 200 or 2000 ISO, when in fact there are many, many compromises.

You said it yourself, "once you accept that development decisions are part of exposure decisions then the craft of exposure becomes almost exactly like film". Exactly! This is why Adams' books have such relevance in the digital age. The difference between you or me and the newbie who has just purchased his first RED camera is that we know this very important distinction. Someone less experienced may take these ISO specs (200-2000 for DRAGON) at face value. They may change camera ISO to suit their scene referenced exposure without realizing that they have now shifted their entire dynamic range to a point where highlights clip or shadows get noisy. And then later, in RCX, they shift development ISO again because they think as long as the numbers are within 200-2000 their shot should look great. Sometimes it works. And sometimes it's a complete mess.

A cinematographer's creative exposure choices can only be made when he/she knows for certain what his camera's Base ISO rating and range are and what additional creative compromises he/she will make in the development phase (which is intrinsically tied to the scene referenced expose / Base ISO).

I totally agree with this. But I'd take up Evin's thought here and assert that the base ISO, while using the manufacturer's label as the starting point has to be derived from your own testing. And that is only partly because of any variation in the device itself. Variation in the "base ISO" is very likely to result from your own peculiar exposure methodology.

 

[Blair S. Paulsen]

It's worth noting that film has a built in log curve due to the chemical response characteristics of silver halide exposed to light. In practice, this meant a certain amount of highlight compression was endemic to the medium so as you pushed up the scale more values got re-distributed from their absolute mathematical analog. There's whole books on these types of "distortions" and how creative shooters use them to craft an image.

My point is that any naturally linear transducer like a digital sensor that generates a linear scale is going to behave differently at the top and bottom than a transducer like silver halide that is bound to its inherent logarithmic response.

Since the human visual system has a logarithmic curve similar to the result of the photochemical imaging process (as well as the Cineon curve) compressing values in the shoulder and toe has been the almost free lunch that made 10bit log DPX such a great balance of size and precision - because it biased the precision toward the portion of the luminance curve humans are most sensitive to.

There are a variety of techniques for re-characterizing the linear values to more closely resemble the log-ish response of human vision. I would add, that IMO, part of why RLF is so popular is that it parks values on the gamma curve in such a way as to facilitate the ability of a colorist to "land" values where they feel "natural". Not to mention that many of the industry leading post tools were built from the ground up for log encoded sources.

IAC, if nothing else comes out of this thread, it does illuminate (sic) the myriad ways different shooters determine exposure.

Cheers - #19

 

[Terry VerHaar]

Indeed. Based on the electronic cameras that have comprised 99% of the market over the last 30 years many shooters assume that dialing up the ISO is adding gain to the signal via amplification of voltage values - typically in front of the A-D converter.

RED doesn't use gain in camera, but not everyone gets that.

Cheers - #19

I can't tell you how often I come across people who are, overall, way more experienced than I am, but who don't understand this. They still think pushing ISO up introduces noise because it is "pushing" the electronic gain.