RED MX shot @ ISO 100

[Brian Timmons]

Test…
For the sake and craft of cinematography the most thorough and reliable means of anyone understanding the best ISO settings for their camera(s) of choice is to test.

800-1200 was normally fine for me with the RED One MX and I remember rating the original RED One anywhere from 250-400.

If I shoot slower motion at a lower resolution (3K or 2K on the RED One MX) I might drop the ISO to 640-400 to help minimize noise at the lower resolution and cropped sensor.

I still take this similar approach with both Komodo and Komodo X.

Still it is best to test to see how the highlights, shadow and noise behave based on the how you plan to shoot, the conditions you might encounter and the look you want.
The end result of that testing is your ISO number or range.

Even with film, a DP would not just dial in a film stocks written ASA (ISO) and call it day.
It was always treated as the manufacturer recommendation. A starting point.
It was common to re-rate a stock based on the look a DP wanted among other reasons.

Years ago I once had to re-rate a film stock from 200 to 160 based on a particular film lab (old times) processing the image slightly darker compared to another lab.

I can say that with some of the advancements in debayering and color science with RED Raw (ie IPP2) that footage going back to even the 1st RED One looks cleaner than the debayering that was available back 15+ years ago so this issue of rating is a bit of a moving target.

There were a few times I experimented with pushing original RED One footage to 800 or even 1600 ISO and thought it looked a bit better (though not perfect ) than I remember it did back in 2009.

Brian Timmons
BRITIM/MEDIA

 

[Liebster]

Test…
For the sake and craft of cinematography the most thorough and reliable means of anyone understanding the best ISO settings for their camera(s) of choice is to test.

800-1200 was normally fine for me with the RED One MX and I remember rating the original RED One anywhere from 250-400.

If I shoot slower motion at a lower resolution (3K or 2K on the RED One MX) I might drop the ISO to 640-400 to help minimize noise at the lower resolution and cropped sensor.

I still take this similar approach with both Komodo and Komodo X.

Still it is best to test to see how the highlights, shadow and noise behave based on the how you plan to shoot, the conditions you might encounter and the look you want.
The end result of that testing is your ISO number or range.

Even with film, a DP would not just dial in a film stocks written ASA (ISO) and call it day.
It was always treated as the manufacturer recommendation. A starting point.
It was common to re-rate a stock based on the look a DP wanted among other reasons.

Years ago I once had to re-rate a film stock from 200 to 160 based on a particular film lab (old times) processing the image slightly darker compared to another lab.

I can say that with some of the advancements in debayering and color science with RED Raw (ie IPP2) that footage going back to even the 1st RED One looks cleaner than the debayering that was available back 15+ years ago so this issue of rating is a bit of a moving target.

There were a few times I experimented with pushing original RED One footage to 800 or even 1600 ISO and thought it looked a bit better (though not perfect ) than I remember it did back in 2009.

Brian Timmons
BRITIM/MEDIA

Thank you Brian, Its beginning to make some sense to me, probably because I am old enough to remember film stock and ASA
What prompted my question is an afternoon not long ago with an Red MX1, filming a panorama under a bright sky with some persistent light diffusing clouds.
It was a slipshot go with this camera; I had no ND's, and had to pull down the aperture drastically to get what looked like a decent exposure while looking at the monitor NOT on raw.
The resulting images were washed out some . I cant recall the ISO dialled in, but since this is just meta data. I am not sure if this had any bearing on the washout.

 

[Brian Timmons]

Thank you Brian, Its beginning to make some sense to me, probably because I am old enough to remember film stock and ASA
What prompted my question is an afternoon not long ago with an Red MX1, filming a panorama under a bright sky with some persistent light diffusing clouds.
It was a slipshot go with this camera; I had no ND's, and had to pull down the aperture drastically to get what looked like a decent exposure while looking at the monitor NOT on raw.
The resulting images were washed out some . I cant recall the ISO dialled in, but since this is just meta data. I am not sure if this had any bearing on the washout.

Hey Liebster,

Glad to be among those helping.
A bit of variables in this.
It might be best to show the image or link to an R3D of the shot.

some questions for clarification.

Was this a high or lower contrast frame?

Describe what the histogram looked like?

Were your exposure tools like the stop lights showing any clipped color channels.

What version of the RED color science and gamma are you using to process the image?

With no NDs on a bright day it seems you were already handicapped in getting the optimal image.
That alone may be causing the issue if you could not close down the iris enough to avoid overexposure.

Brian Timmons
BRITIM/MEDIA

 

[M Harvey]

The resulting images were washed out some . I cant recall the ISO dialled in, but since this is just meta data. I am not sure if this had any bearing on the washout.

Just to piggyback on @Brian Timmons smart reply: in my experience, sometimes the "washed out" quality of an image under those conditions isn't so much the camera as the lens. If you were stopping down past f8 to control your exposure, optical diffraction can cause detail and contrast to suffer.

Another good candidate for washed out contrast with a hot sky is veiling glare, which can make an otherwise punchy lens produce something that looks like a water color!

 

[Brian Timmons]

Here's a Dropbox link to a quick night shot of a doorway filmed
on the original RED One (Mysterium sensor) years ago.

Dropbox

www.dropbox.com

Despite being filmed on the original Mysterium sensor (which for this shot I rated at 320 ISO), this footage
could now easily push to 800 ISO. TBH, 2000 ISO does not look that bad (maybe some noise reduction).
By all means download the file and try it out.

Again, this is with the original RED One Mysterium sensor.

At the time of filming, anything above 400 ISO would likely give a bit of noise particularly in the blue channel.
This was eventually fixed in Build 20 I believe.

I definitely remember filming a scene for a feature in 2009 that after filming I thought I might have lit it too dark as there was more noise than I liked. I planned on applying noise reduction to get by.
By the time we got to color correction months later there was a color science & debayering update in REDcine X that basically made the noise a non issue. The update alone sufficiently rendered the image without most of the offending noise.

This shows a part of raw that does not get mentioned enough that old footage can get enhanced well after the initial filming
as the Debayering algorithms (Demosaicing) improve.

I don't have an original RED One Mysterium around but I do wonder if one could ultimately re-rate the ISO from 320 to something higher like 640-800 just from the many debayering improvements to REDcode.

Brian Timmons
BRITIM/MEDIA

 

[Karim D. Ghantous]

At the time of filming, anything above 400 ISO would likely give a bit of noise particularly in the blue channel.

That's why I assert that the Bayer filter is wrong. It should be RB,BG. I haven't yet heard a good reason for the RG,GB pattern, apart from getting a higher ISO rating.

 

[Phil Holland]

That's why I assert that the Bayer filter is wrong. It should be RB,BG. I haven't yet heard a good reason for the RG,GB pattern, apart from getting a higher ISO rating.

That would be noisier in general and less effective. You could design a baseline sensor technology that would be more sensitive to "blue", but you would also be hitting walls. Green is the best for luminance sensitivity not just based on how our eyes work, but generally how visible light works in relationship to the sensor acquiring the image.