Native ISO Rating

[Wey Wang]

I've looked around, read the manual, and it seems like the MX sensor is supposed to be rated at ISO800. I recently did some tests where I changed the ISO and then would check the raw.

The only setting where the histogram didn't significantly change was at ISO 320. Am I missing something? Or should I really be metering and shooting at ISO320 so what I'm seeing is closer to what the sensor is seeing?

 

[Jake Wilganowski]

Lots of threads on this. What your seeing is correct. 800iso basically makes you stop down and protect your highlights. This also affects the creaminess of the highlight rolloff instead of pushing your whites right up to true clipping at 320 iso. You can use both ways to your advantage once you get comfortable.

 

[Wey Wang]

I know that shouldn't affect the camera's actual dynamic range, but how would that affect the perceived dynamic range? Do you know of any dynamic range charts similar to what the Alexa has?

 

[Jake Wilganowski]

I'm trying not to get super technical here. I'm sure someone can point you to charts. Just know that 320 is accurate, but if u expose at 800 you are in essence exposing a stop under in relation to 320, which allows for a smoother transition to white, part of the filmic look. But some people like to use 320 and expose precisely up to true clipping, depends on your style, shooting and grading and personal taste.

 

[Sidney L. Plaut]

I'm trying not to get super technical here. I'm sure someone can point you to charts. Just know that 320 is accurate, but if u expose at 800 you are in essence exposing a stop under in relation to 320, which allows for a smoother transition to white, part of the filmic look. But some people like to use 320 and expose precisely up to true clipping, depends on your style, shooting and grading and personal taste.

I am not sure this is entirely accurate (320 iso accurate) - I seem to remember one of the red guys mentioning that the Raw view basically was a leftover from the red1 m? (you are right ofcourse about the exposing 320 vs 800 and how much head room you have)

 

[Mike P.]

It "increases" the perceived dynamic range because you're protecting the highlights which tend to visibly clip first. I don't know of any alexa-like charts, but like Jake said, having the ISO at 800 (which is recommended because it's the highest ISO that still has clean shadows/noise floor) forces you to physically reduce brightness via shutter, f-stop, or ND on location. Then, in post, when/if you lower the metadata to ISO320 (for example) and apply a suitable curve, you'll see way more details in the highlights (because you physically reduced their brightness on location) and the shadows will still be clean when boosted up a bit to compensate for the meta-ISO reduction (via the necessary pulling up of the shadow due to the fact that you lowered the ISO to 320.)

I mean, it's simple enough to test. Go outside and try to capture cloud detail at ISO 320 and then again at ISO800 (adding ND/shutter/stopping down to keep the highlights from clipping at ISO800). Now in post (RCX or what-have-you) correct the image of both captured-ISOs to get the cleanest/best exposed images possible. The ISO800 shot will presumably look better.

 

[Wey Wang]

So if raw view is a leftover from the R1MX, should we even be using it? Because the way it used to work was that the histogram would show the RAW, even if you changed your ISO. Or should rely more on false colors to see what's actually clipping?

As for charts, in the Alexa manual, it shows graduated gray chips and which point is 18% gray at what ISO rating. It helps give a sense of how many stops you have above and below 18% gray.

 

[Jake Wilganowski]

The histogram is god. So I like to look at it as the true judge of exposure at 320. Everything else is a tool to use for whatever occasion you wish. My 2 cents.

 

[Jake Wilganowski]

Somebody is about to write in with some uber technical talk about how there is no true iso for raw and then start talking about sensor technology math stuff that is way over my head. I work in practicalities, so practically speaking, 320's histogram is the most accurate.

 

[Arnold F.]

Can the gain on the camera be changed? Is a change in ISO on set essentially a pre-selection of ISO in post?

 

[Jake Wilganowski]

Your iso selection only affects your monitoring path and can be changed in post as it is only metadata with no consequences.

 

[Jake Wilganowski]

But yes, you can have redcine default to whatever you recorded at in the field

 

[Noah Yuan-Vogel]

I work in practicalities, so practically speaking, 320's histogram is the most accurate.

ISO320 histogram is most accurate for images shot at ISO320

 

[Mike P.]

So if raw view is a leftover from the R1MX, should we even be using it? Because the way it used to work was that the histogram would show the RAW, even if you changed your ISO. Or should rely more on false colors to see what's actually clipping?

Honestly, I'm not really "qualified" enough to answer that definitively. As Jake mentioned again, it depends on what you're comfortable with. I'm talking theoREDically (see what I did there? I should trademark that.) Theoretically, viewing at 320ISO in RAW should give you the most realistic representation; if highlights aren't clipped and shadows aren't crushed when viewing 320/RAW it'll be 'what you see is what you get', which will look great if you aren't really futzing with it all that much in post. That said, you'll probably be adjusting it in post, so it may not be the best option given the conditions of the shoot. On the other hand, theoREDically, you'll be capturing footage that has a much more usable adjustment range (that is, footage that has the cleanest shadows with the most highlight detail) if exposed for ISO800 on set/location independent of your viewer settings. Sorry if that seems somewhat vague, I just don't want to tell you 'do it like this' and then find out I completely botched your next shoot.

The other general consensus is that you use the goal posts to ensure your aren't clipping or crushing regardless of how you're viewing the footage. With that in mind, setting to ISO800 seems to be the safest bet no matter what you're viewing the capture at. If you're really apprehensive about it, try switching back and forth between RAW and RG2 or 3, to get an idea of what's happening both in the image and in the histo/exposure posts... Honestly, I feel that's the only way that'll give you peace of mind (I wasn't joking about the cloud test either.)

EDIT:

ISO320 histogram (aka RAW histogram) is most accurate for images shot (aka monitored) at ISO320

Everything I wrote above means essentially the exact same thing that Noah said here (yeah, I know...) I added a couple of brackets for clarity... hopefully you can see the relation (which I admit isn't easy.)

Can the gain on the camera be changed? Is a change in ISO on set essentially a pre-selection of ISO in post?

All the camera ISO does is change what you're viewing/monitoring, it has no real bearing on the footage EXCEPT in how you physically expose the footage on set/location which is based off the ISO you choose to monitor at (and the physical exposure obviously has a bearing on the footage's characteristics.) SO, for example, lets say you VIEW at ISO320 which requires f2.8 to get proper exposure of the mids/skin-tones. HOWEVER if you VIEW at ISO800 it'd require f5.6/8 to get the same proper mid-tone exposure on the screen. Now in post, the f2.8 image is going to have clipped highlights (but brighter shadows) where as the f5.6/8 shot will have way more highlight detail captured (but much darker shadows.) The difference being that you'll have an easier time regaining shadow detail than you will highlight detail because a.) if the highlights are actually clipped they are gone (property of all digital sensors) and b.) ISO800 on an MX has a decent enough noise floor that you can pull up a fair bit detail in the shadows before it starts looking like complete crap.

By the way, this is a very blunt way of looking at it and is probably not very realistic in practice, I'm just trying to help clarify what people mean when they say "ISO is just metadata in RAW."

 

[Arnold F.]

Sometimes it's easier for me to understand if I've put it in my own words. Can you tell me if the following statement is true?:

The sensor has one sensitivity, 320 ISO, and all others can be thought of as "virtual" ISO's. No real gain adjustments can be made in-camera like they can in my HVX or 5D. Regardless of what "ISO" I shoot at the sensor accepts it at 320. If I choose 1600 I am, to a degree, pre-choosing that gain/ISO in post because all other exposure parameters (shutter speed, aperture, etc.) were chosen based on that ISO. ISO is "chosen" on set and "confirmed" (with some latitude available) in post. The histogram will change with ISO changes based on the assumption that that ISO will be chosen in post. However, what remains the same is that a gain increase, although now in post, still results in added noise like it always has.

 

[Mike P.]

Sometimes it's easier for me to understand if I've put it in my own words. Can you tell me if the following statement is true?:

The sensor has one sensitivity, 320 ISO, and all others can be thought of as "virtual" ISO's. No real gain adjustments can be made in-camera like they can in my HVX or 5D. Regardless of what "ISO" I shoot at the sensor accepts it at 320. If I choose 1600 I am, to a degree, pre-choosing that gain/ISO in post because all other exposure parameters (shutter speed, aperture, etc.) were chosen based on that ISO. ISO is "chosen" on set and "confirmed" (with some latitude available) in post. The histogram will change with ISO changes based on the assumption that that ISO will be chosen in post. However, what remains the same is that a gain increase, although now in post, still results in added noise like it always has.

Yeah that's right.

The only confusing part is that ISO800 offers the "same" (negligible difference) noise-free cleanliness as ISO320, so by setting your monitoring path as such, you inherently capture more highlight detail because you stop down to compensate for the "higher-than-necessary" ISO of what you're viewing at.

 

[Jake Wilganowski]

Yes

 

[Joseph S.]

Mike P. And Jake explained it well.

Basically if you make it look right at 800 DURING the shoot, you get a lot of leeway in post (I think). ?

 

[Nathaniel Kramer]

SHOOT 320.. ALWAYS.. 800 is just mental masturbation.. Check your goal posts.. Check your exposure mode.. If its too purple, open up or add light.. If its too red, close down, ND. I personally can live with clipped highlights -as usually that is the effect i want..I pay more attention to the blacks as thats where noise is most visible... my one cent

 

[Jon Thomasberg]

In other words, expose for 320 and make sure you don't clip your highlights, but make sure you give yourself a stop or more of headroom or else you will have truly lost the clipped pixel data in RAW, from which you will not recover. The 'set for 800' thing is to give you a 'failsafe' or 'spill-over zone' so that if you expose for 800 you will still be protecting your highlights by 1.3 stops, thereby still giving your highlights an extra stop of headroom before truly clipping. Stated differently, if you are on the borderline of clipping at 800, you still have a stop left before true clipping, but you are treading in the danger zone after that. However, like others I have tested this and RAW = 320. Anything you adjust it to in camera is just a monitor path for producing over/under look, and stores that as just metadata in the R3D, however it doesn't affect the RAW image capture, which is still captured at 320 regardless of what you set your ISO to in camera during capture.

Armed with that info, I shoot for 320 (or RAW mode) and leave 1-1/2 stops headroom in the histogram@320 or regret it later. Couple that with checking your noise meter, clip meter and stoplights -- they are there as a sanity-check for that reason. Reds MX sensor IS awesome, but it is also very unforgiving in that when you are clipped, you are clipped....no failsafe.

/FIN