ISO - Useless while shooting and in post

[Rui Guerra]

Hi all,

Thinking about what ISO change does or not, it looks like a pretty much useless control (if shooting RAW).
Do you agree ?

Well, let's see:

WHILE SHOOTING:

1 - We agree that choosing an ISO value in camera while shooting is the same as choosing it in post, if everything else stays the same, since ISO is just metadata that can always be changed afterwards.
2 - We also know that lowering the ISO leads to more stops of dynamic range in the shadows (less noise and more shadow detail) and higher ISO values will keep more stops of dynamic range in the highlights (protection against burned highlights). Dynamic Range is always the same, it's just displaced towards one end or the other.
3 - But the statement in number 2 is always valid, no matter if a particular ISO number is chosen while shooting or in post.
4 - Although that many people follow the thinking that choosing a low ISO while shooting in low light will lower the image noise, that is only true if besides the ISO lowering, something else also change (lighting, aperture or shutter speed).

Conclusion A - If everything above is correct, changing only the ISO (while shooting), doesn't do anything to the image. If so, any change in image quality are only made when one physically change the amount of light that reaches the sensor (by changes in aperture, shutter angle, lighting or ND filters).

IN POST:

5 - Again, what is said in number 1.
6 - We know that to make a good color correction and grade in post, for example in DaVinci Resolve, we should start with a flat log image (IPP2/RedWideGamma/Log3D10).
7 - If we look inside the camera RAW settings inside the Color tab of DVResolve, we can see that ISO is set to 320, regardless of the type of shooting conditions (bright or dark) or the ISO value that we've chosen while shooting.
8 - While we advance in the color correction and grading of our image, we never go back to those camera RAW settings and adjust ISO. We just keep using the primaries and secondaries adjustments, taking advantage of the very wide dynamic range of that initial flat/log image, that contais all the light and color information that a RED sensor can record.

Conclusion B - Is everything above is also correct, changing the ISO in post is useless, as is also useless any time spend to choose a specific value while shooting.
Conclusion C - The displacement of the Dynamic Range according to the ISO value (explained in number 2) is also completely irrelevant in post, since we just use as a starting point the flat/log image.

GENERAL CONCLUSIONS:

D - We can further conclude that the ISO control is completely useless, both in camera and in post. So all we need is the camera's tools to check the amount of light reaching the sensor (goal posts, traffic lights, GIO Scopes, etc) while shooting, and a flat/log image while editing. We can disregard ISO value completely in both stages.

E - The displacement of the Dynamic Range towards the shadows or towards the highlights (also frequently referred to as the level of mid grey), or in fact the knowledge about that fact, is also irrelevant because 1) we don't need to set ISO at any specific value while shooting and 2) we we'll not use it in post anyway.

F - ISO can be used just as a reminder of some light controlling decisions that should be made while shooting, but if one knows how to use the camera's exposure tools, there is no need to change ISO for that.

So, the final big question: do you agree that ISO is an irrelevant control that we can forget about, or there is any specific scenario (while shooting or even in post) where you think that ISO plays a main role?

All the best,

 

[Björn Benckert]

It works like this. Iso is digital gain added as a parameter post capture.

So for example the DP shoots a couple of wide shots at a certain stop. Then when he moves in to shoot his closeups he might want less focus falloff, so he close down two stops. Then to make close look the same as his wide in therms of brightness he can upp the iso for the closup by multitude of 2. If the wide was 800 he would simply change iso to 3200 for the closeup.

Now if the colorist does things right he import the footy with camera setting so in his timeline the wide and the tight mach up. The got the same brightness values, one is less exposed and has higher iso and the other vice versa.


But yes when you shot iso irrelevant, what matters is how you expose the sensor and iso does not do thing to your exposure, at that point is like brightness of the screen. But the iso and color temp etc that you set tells the colorist / dit or who ever is next to work on the material what your intentions where.

 

[Rui Guerra]

Hum... if the wide shoot is well exposed, following that procedure will make the close-ups 2 stops underexposed.

If the DP needs to gain depth of field for the close-ups and wants to maintain a "brightness" level for all shoots, he should than correct the light/exposure for the close-ups (more natural/artificial light or lowering shutter speed), avoiding that way a noisier underexposed close-up images.

Using ISO that way will be like using it as a "post-it note", but a dangerous one, regarding image quality. Don't see any advantage in that.

 

[Audy Erel]

Constant ISO number only needed to keep the noise/grain consistency across the entire film.. Other than that it only act like a 'monitor brightness setting'.

 

[Rui Guerra]

Other than that it only act like a 'monitor brightness setting'.

Indeed, that's a good description.

But then, if in camera ISO is just a "post-it", a reminder (so it exacerbates the need to take measures regarding shadows or highlights) or just a "monitor brightness setting", then why there are so many talk about the advantage of RED ISO been a range with constant DR, and the "important" practice of lowering it in low light situations to gain more stops below middle grey (and the opposite for bright conditions).

In the end, it looks like all that information is pointless since changing it in camera doesn't do anything per si and in post what we need is just a flat/log image.

 

[Audy Erel]

I think the 'low ISO for clean shadow' principle only works if we have full control of all the lighting.. Don't confuse "low key" (a creative decision) with "low light" (lack of light) though, totally different beasts. We need MORE light to produce deep & clean low key images (hence, the darker monitoring or lower ISO).

The 'higher ISO for highlight protection' principle on the other hand, is more applicable to wild documentary/available lights only situation where we could not help with the blown out windows or bright skies but to reduce the light hitting the sensor (with ND, Aperture or Shutter Speed) thus making all seems darker as a consequence to have that lovely clouds correctly exposed, so in turn we need to "brighten the monitor up" to make everything else look right, (a.k.a. raise the ISO)!

Given no creative look nor grain consistency is concerned, the best practice with RED or any raw camera is to expose to the right. Go as far right as possible with the GIO scope without overexposing highlight and you will be good. We could always add grain later in post but we could never get the lost highlights back. It's all safely contained in the flat log image.

But remember the usable DR is actually quite limited... It's a constant battle between the noise floor and highlight clipping. We need to choose where to shift the middle grey according to our need/priority.

 

[Rui Guerra]

I think the 'low ISO for clean shadow' principle only works if we have full control of all the lighting.. Don't confuse "low key" (a creative decision) with "low light" (lack of light) though, totally different beasts. We need MORE light to produce deep & clean low key images (hence, the darker monitoring or lower ISO).

Yes but then changing the ISO is just a commodity (to improve monitor visualization).Darkening the monitoring can of course also be done by lowering the monitor brightness (to a lesser degree, depending on the monitor). In either case, we agree that in reality lowering ISO doesn't do anything to the image, and ultimately is not needed.

The 'higher ISO for highlight protection' principle on the other hand, is more applicable to wild documentary/available lights only situation where we could not help with the blown out windows or bright skies but to reduce the light hitting the sensor (with ND, Aperture or Shutter Speed) thus making all seems darker as a consequence to have that lovely clouds correctly exposed, so in turn we need to "brighten the monitor up" to make everything else look right, (a.k.a. raise the ISO)!

Again, the raison is just to make some things right, not a technical reason indeed. Just by looking at the traffic lights we know when something will be burned out and take measures to prevent it, as you said. No need to change ISO. BTW, rising the ISO so "everything else look right" will again display the sky almost white. So this is just a personal choice of what we want to see better in the monitor. So, no technical reason here also.

Given no creative look nor grain consistency is concerned, the best practice with RED or any raw camera is to expose to the right. Go as far right as possible with the GIO scope without overexposing highlight and you will be good. We could always add grain later in post but we could never get the lost highlights back. It's all safely contained in the flat log image.

I could not agree more! And that's the main focus of this post. Besides the usual exposure correct practices (ETTR), the bottom line is "It's all safely contained in the flat log image."

But remember the usable DR is actually quite limited...

Do you think so? My findings is that even very high contrast scenes (with limits, of course) will easily stay within the limits of the sensor's DR.

It's a constant battle between the noise floor and highlight clipping.

Yes, of course. But my point is that battle will take place more in the realm of other light controls (lighting itself, aperture, ND, shutter speed), and during post, rather then by any change in ISO (either in camera or in post). Don't you agree?

We need to choose where to shift the middle grey according to our need/priority.

I find this last sentence a bit in contradiction to the other ones. You you agree that changing ISO is like a A)"monitor brightness setting" and B)"It's all safely contained in the flat log image.", then what's the purpose of changing ISO?! Middle grey, more then anything will be set in post while color correcting/grading which, BTW, is done without the need to change the default ISO, right?

 

[Ryan Sims]

The only time I have found adjusting the ISO up away from your base (like 800 ISO or whatever your base is) useful on set is in low light situations to allow the AC or operator to see the image better on the monitor to nail focus.

 

[Les Hillis]

I think for good practical reasons there's a necessary place for 'ISO'-like adjustments in the image-capturing and processing chain.


However, imo the further along in time we get with digital image capturing, the more anachronistic the old analogue film term 'ISO' becomes.

It made sense during the initial transition from analogue to digital to use that sort of short-hand, familiar terminology that everyone sort of 'got' without having to give it too much thought. But that was then...


I've mentioned this before, but I think the actual ISO (International Standards Organization) should create something like a Standardized Viewing LUT (SVL) for use with digital imaging systems.

It would be a display LUT made by each camera manufacturer, for use in each of their cameras, set to a particular target determined by the ISO.

It would be nothing more than a colour & gamma transform of the RAW (or other) sensor data, with positive and negative decimal values that the end-user can adjust like they currently do with 'ISO'.

The only difference is, when set to SVL 0 (Zero), what you see in your monitor is the same thing (in terms of 'exposure') for every camera that complies with the SVL standard (much like analogue film was, or was supposed to be).

Apart from the standardization itself, the benefit is, the terminology used makes it clear the SVL adjustments have nothing to do with your actual exposure but are purely how you're choosing to view that exposure.


Obviously there's not much chance of such a standard being implemented, so the suggestion is a bit ridiculous, but imo, so is using the term 'ISO' in modern digital cameras.

 

[Mark A. Jaeger]

test

 

[Dermot Shane]

6 - We know that to make a good color correction and grade in post, for example in DaVinci Resolve, we should start with a flat log image (IPP2/RedWideGamma/Log3D10).

We do not know that, not really...
transforming RAW into display refered is really not the only way to work, and really not always the best way...
to make an awesome looking image from a RAW file you can work in ACES /Truelight /RCM2 with output of the sensor in linear light
just say'n...

 

[Michael Tiemann]

As long as sensors have finite dynamic range and non-zero levels of noise in the darks, then the principles of the Zone System (first articulated by Ansel Adams, but practiced and evidenced by so many great image-makers) apply. At the same time, cinematography is more the just calculating and obsessing over exposure details of every single frame for hours. It is helpful to look at a monitor that shows in real-time the promise of WYSIWYG (what you see is what you get).

The practice of treating ISO as a variable for a given image capture medium goes back to the late 19th century. RED cameras provide a digital negative that allows evaluation in real-time, which is amazing. Treating ISO as irrelevant entirely misses the point of the DSMC concept.

Here on REDUSER people are of course welcome to ask and answer however they like (provided they follow the code of conduct). But do not mistake rhetoric for soundness nor boldness for truth. Misconceptions abound, and it is only experience that teaches us the real what and why.

 

[Nick Morrison]

I personally think ISO is very helpful, even when shooting RAW.

Filmmaking is teamwork, and ISO is a common standard that can be shared with your gaffer, your colorist, your AC, your DIT, etc.

Even if you are familiar with the intricacies of RED and shooting RAW, they may not (and often don't need to be).

 

[Aaron Green]

I could be wrong about this, but dont RED sensors record at 800 iso, and every adjustment (lowering/raising ISO) is just like turning an exposure wheel in post?

 

[Phil Holland]

I could be wrong about this, but dont RED sensors record at 800 iso, and every adjustment (lowering/raising ISO) is just like turning an exposure wheel in post?

RED defaults and suggests a Base ISO of 800 as a starting point when Rating your ISO.

Most of the cameras have a recommended ISO Range of ISO 250-3200 with Extended ISOs available. Gemini being a Dual ISO sensor is a bit different, but overall this is what is recommended.

The sensors themselves have a flatline sensitivity in Linear Light, this is a very low number and is indeed the "Native ISO".

To the topic, this is one of those how hard do I disagree with the general statement? A whole bunch.

Fundamentally, and this is true for all digital cinema cameras, how your Rate your ISO and use that information towards your Exposure Strategy is very important. Whether you expose based around Middle/18% Gray or ETTR/Exposing to the Right or even Bucket/Histogram based exposures, where you rate the ISO has an impact on how much Apparent Visual Noise can be seen. It also has an impact on where your values land, i.e. closer to the Noise Floor or Light Ceiling. If choosing Middle Gray as an Exposure Method, what most cinematographers do, shooting lower ISO gets you cleaner shadows, but less stops allocated towards your highlights range. Going for Higher ISO you weight more stops towards highlight information, but your Middle Gray will be noisier. Depending on all things this has implications of where Crushing and Clipping may occur and that's what the RAW Exposure Goal Posts are useful. And honestly even more useful is the GIOSCOPE tool, which really shows you where values land on the sensor and where potential in frame issues arise.

The key with REDCODE RAW and other RAW formats is you have nondestructive control over adjusting your Exposure via RAW controls. As well as Color Temperature. This is actually not the same as general grading controls to do similar tasks. One of the more interesting things I notice omitted from many YouTube grading tutorials is not fully harnessing the power of RAW cameras because many have built upon similar tutorials and workflows presented. Not saying the ways presented are wrong per say, but it's interesting seeing methods actually used somewhat not talked about.

I see a lot of people rate at 800 and call it a day, they end up never changing that. That's cool. More or less like shooting Kodak V3 500T and never pushing or pulling I guess. But I personally switch it up depending on what I'm looking for. I know many who like ISO 1600 for projects from about Dragon onward and there's plenty of Low ISO shooters out there in the 250-400 range. Certainly times where I've had to go up to ISO 5000 or 6400 for sure when the situation calls for it.

If you look at any of my exposure ramp tests over the years, you can see I'm exposing to 18% gray at each ISO Rating to reveal how much visible noise you have at a given output transform/tone map combination. It's how we look at cameras when we test them and also come up with strategies for best practices.

So yeah, my perspective, not useless at all.

 

[Scot Yount]

This will sound (and is) pedestrian after Phil's post, but one thing about ISO I find very useful, is giving the colorist a starting place on multi-cam shoots. I have a Helium Weapon, a Scarlet-MX, and a Komodo. When shooting all three for instance in a cooking tutorial, I usually end up with the Helium at 400, the Scarlet at 800 and the Komodo at somewhere in between, depending on the lighting. That way the colorist who is grading those in a multi-cam project has images that are all about the same brightness, and to my eye, the same level of noise as a starting place. Of course, this varies according to the setup and since everything is raw, he or she can change it all later if so desired.

 

[Rui Guerra]

We do not know that, not really...
transforming RAW into display refered is really not the only way to work, and really not always the best way...
to make an awesome looking image from a RAW file you can work in ACES /Truelight /RCM2 with output of the sensor in linear light
just say'n...

Can you share a bit more about this and how/where the selected in camera ISO value will be respected/relevant?

Thanks!

 

[Rui Guerra]

cinematography is more the just calculating and obsessing over exposure details of every single frame for hours. It is helpful to look at a monitor that shows in real-time the promise of WYSIWYG (what you see is what you get).

Indeed!

The practice of treating ISO as a variable for a given image capture medium goes back to the late 19th century. RED cameras provide a digital negative that allows evaluation in real-time, which is amazing. Treating ISO as irrelevant entirely misses the point of the DSMC concept.

I understand that. My post is really more about the naked technical relevance of in-camera ISO, not so much about the WYSIWYG, creative or cinematographic side of it, which is valuable, of course.

 

[Rui Guerra]

I personally think ISO is very helpful, even when shooting RAW.

Filmmaking is teamwork, and ISO is a common standard that can be shared with your gaffer, your colorist, your AC, your DIT, etc.

Even if you are familiar with the intricacies of RED and shooting RAW, they may not (and often don't need to be).

Yes, I agree with that. But this is a workflow related convenience, and not exactly a technical "must" that will make or brake the final image. Just imagine that you are a one man band, shooting a ridiculous short film (just one clip, for the sake or argument). How will the selection of a particular ISO value while shooting impact the final footage in a way that it could not be done latter in post? Remember that in this scenario you will be the DOP, DIT and Colorist.

So, do you really need to change the ISO while shooting, and why?

 

[Rui Guerra]

Thanks Phil, for your post and detailed explanations. I think that we agree in near everything. The main question is just to understand if there is any difference between changing ISO in camera or just in post.
Not wanting to look arrogant here, or doubt about anything you wrote, just trying to make sense of the intricate technology behind RAW ISO.

You raised some very interesting points, that I would like to mention just the most relevant part of them (hope not to decontextualize it too much) so maybe you could develop a little bit more:

Fundamentally, and this is true for all digital cinema cameras, how your Rate your ISO and use that information towards your Exposure Strategy is very important.

Agree that seeing in the monitor an image darker or lighter due to ISO adjustment, is a powerful reminder of what needed to be done or a closer math of the intended result. But if we assume that we would follow the same Exposure Strategy anyway, if we don't change in-camera ISO, would the end result be different in any way (assuming that we will do that ISO rating only in post)?

If choosing Middle Gray as an Exposure Method, what most cinematographers do, shooting lower ISO gets you cleaner shadows, but less stops allocated towards your highlights range. Going for Higher ISO you weight more stops towards highlight information, but your Middle Gray will be noisier.

But the end image quality will not be exactly the same if that ISO changing was only done in post?

Depending on all things this has implications of where Crushing and Clipping may occur and that's what the RAW Exposure Goal Posts are useful. And honestly even more useful is the GIOSCOPE tool, which really shows you where values land on the sensor and where potential in frame issues arise.

This is a very important aspect of this discussion. Indeed the Goal Posts and GIOSCOPE are both based in RAW data, where ISO have no influence. So when we are using those very important tools to adjust exposure, aren't we in a sense disregarding ISO?

The key with REDCODE RAW and other RAW formats is you have nondestructive control over adjusting your Exposure via RAW controls. As well as Color Temperature. This is actually not the same as general grading controls to do similar tasks. One of the more interesting things I notice omitted from many YouTube grading tutorials is not fully harnessing the power of RAW cameras because many have built upon similar tutorials and workflows presented. Not saying the ways presented are wrong per say, but it's interesting seeing methods actually used somewhat not talked about.

This is a major thing and indeed my initial post was assuming the commonly workflow starting with a flat/log image and using color correcting/grading tools (in my example using DaVinci Resolve), although I have in the Camera RAW tab the RAW adjustments I may need in a clip by clip basis, if needed. This is even what as been suggested as a standard powerful method to work with R3D, for example in this RED TECH | KOMODO | Post Workflow | 4K: https://www.youtube.com/watch?v=a3i3JPU1iLs
But I would be indeed very interested in hear about the "methods actually used somewhat not talked about", can you summarize them (or point to some useful links)? Thank you!

I see a lot of people rate at 800 and call it a day, they end up never changing that. That's cool. More or less like shooting Kodak V3 500T and never pushing or pulling I guess. But I personally switch it up depending on what I'm looking for. I know many who like ISO 1600 for projects from about Dragon onward and there's plenty of Low ISO shooters out there in the 250-400 range. Certainly times where I've had to go up to ISO 5000 or 6400 for sure when the situation calls for it.

I understand those options. Just trying to figure out how they truly impact the quality of the final image, if everything else stays the same.

Cheers,