I S O settings on MX

[joshua csehak]

Yeah, the chip is rated at 800, and I've seen a post by Jim somewhere that 800 is the officially recommended setting.

I don't really do the whole "push it to 1000-1600 to protect the highlights" thing. I keep it at 800 and check the raw sensor data false color view before almost every shot so I know exactly what's getting clipped, and I'll just stop it down or put some ND on if I want less of it. I think they're two different ways of getting the same exact footage. I think the reason to set it higher to protect the highlights is so it gets to the editor with the exposure you want, so he doesn't think you want it to be all dark. But since I'm usually my own editor and grader, I feel more comfortable getting the highlight protection only on the shots I want.

 

[joshua csehak]

AND if you go 1600 ISO with an FLUT of -1... You have just cancelled yourself back to 800.

makes not much sense to me...

I think they're not the same. ISO is just metadata, right? The camera gets what the camera gets. So if you take a shot of a scene metered at T5.6 at 800 ISO, and then re-rate the camera at 1600, your meter will tell you to set it at T8, right? But since ISO is just metadata, your raw footage will have one extra stop of highlights (and one less stop of shadows). So if you later FLUT back up, the highlights will roll off much more nicely, though you'll have a bit more noise in the shadows.

 

[Stephen Strangways]

ISO and FLUT are the same, they are both just metadata, and they both affect the image in the exact same way. It's just that FLUT is more finely adjustable in smaller increments than ISO.

As Gunleik said, if you go to 1600 ISO with a FLUT of -1, you are back to 800, and if you are using a light meter to set your lens aperture* then you should have your meter at 800. But the camera will be reporting ISO 1600... and potential mistakes ensue.

*(which is rather silly, actually. Use a light meter to check contrast ratios when lighting, but use the camera's built in meters to measure what the sensor is actually seeing, as that's the only thing that matters for proper exposure.)

 

[Gunleik Groven]

If you go back to RAW view, you'll see that that corresponds well with 320 and that all RAW metering is based on that.

BUT at 320 you have a headroom in the highlights at exactly 0 in daylight, less if you rate the camera for tungsten.

0 is a hard value to balance properly

To get that headroom, you'll have to set the ISO higher.

Thus: 800 is a good plan.

Where is the "native" ISO of the sensor?
Where middle gray hits the middle of the exposure latitude. But that is a theme that can be discussed indefinitely as people tend to disagree as to what is acceptable, and even for the same person that might be a bit project dependant.

http://205.234.135.241/forum/showpost.php?p=664150&postcount=504

 

[joshua csehak]

ISO and FLUT are the same, they are both just metadata, and they both affect the image in the exact same way. It's just that FLUT is more finely adjustable in smaller increments than ISO.

As Gunleik said, if you go to 1600 ISO with a FLUT of -1, you are back to 800, and if you are using a light meter to set your lens aperture* then you should have your meter at 800. But the camera will be reporting ISO 1600... and potential mistakes ensue.

*(which is rather silly, actually. Use a light meter to check contrast ratios when lighting, but use the camera's built in meters to measure what the sensor is actually seeing, as that's the only thing that matters for proper exposure.)

I don't think it's silly at all. There are two schools of thought on exposure. Some want the fattest negative possible, and it doesn't matter where the key falls. For that it doesn't matter where you set your iso (except for preview purposes), just use the histogram and false color views. But other people, mostly those who are used to a film workflow, prefer to expose as they would for film (meter the scene and set the aperture to what the meter tells them). If you're doing the latter method, if you rate your camera at 1600 and bring it down to 800 in post, you'll have one stop of extra highlights and one stop higher noise floor than if you were to rate it at 800.

Gunleik, as I understand it, you've got it spot on.

Graeme and a new guy talk about it in this thread:
http://reduser.net/forum/showthread.php?t=50632

It's pretty confusing. I'm going to start a new thread that hopefully can turn into a basics-of-how-to-meter-on-the-Red tract.

 

[Paul Harding]

I read an early thread 2008 I think which stated RcX clips at iso 200. Is this true for the current build
Paul

 

[Pablo Villegas S.]

I've used ISO 800 on the RED MX, in day exteriors, to extend the latitude on the top of the curve, but, is 800 ISO really usable on night exteriors and interiors where there is little light? The ammount of noise is acceptable?

I usually shoot at a lower ISO at night to protect the toe of the curve from excesive noise, and to gain more detail in the shadows, at least that's what I did with the old sensor. But this Will be my first night shoot with the new sensor. And I will need the extra speed, so, is the MX really usable at 800 ISO? With acceptable noise levels?

Thanks in advanced.

 

[Gunleik Groven]

It doesn't hurt the image to shoot at high ISO. You may consider pushing it back in post, though.

If you test the current colorscience and MX sensor, you'll see that ISO doesn't really gain noise anymore, it just lifts what is in the image to viewable levels. This surprised me a bit, but pretty much explain the low noise of the MX images in general.

It is very impressive.

 

[Pablo Villegas S.]

Thanks! I guess i'm gonna have to try it.

I'm really impressed with this sensor, but I haven't used it at night.

 

[Costelloe Michael]

I've used ISO 800 on the RED MX, in day exteriors, to extend the latitude on the top of the curve, but, is 800 ISO really usable on night exteriors and interiors where there is little light? The ammount of noise is acceptable?

I usually shoot at a lower ISO at night to protect the toe of the curve from excesive noise, and to gain more detail in the shadows, at least that's what I did with the old sensor. But this Will be my first night shoot with the new sensor. And I will need the extra speed, so, is the MX really usable at 800 ISO? With acceptable noise levels?

Thanks in advanced.

I can't see any noise at 800asa on MX, it shows less noise than M chip at 320. I usually overrate at 1200/1600 on nights as I know we will drop levels in post and loose any noise. I really don't get how not getting info to the sensor will limit noise?

 

[Phil Bates]

I like shooting with 800, for the simple reason that I can go either way in post. It leaves all my options open.

Phil

 

[Gunleik Groven]

I can't see any noise at 800asa on MX, it shows less noise than M chip at 320. I usually overrate at 1200/1600 on nights as I know we will drop levels in post and loose any noise. I really don't get how not getting info to the sensor will limit noise?

Well it doesn't...

Thing is that this is a bit situation-dependant.

If you light your scene - even at night, you "could" rate your ISO lower to "force" the lights up, thus getting more image/less noise/higher contrast.

When you shoot natural light-only, the rating of the sensor doesn''t really affect what is actually recorded, so I'd set the ISO at whatever is convinient to have an ok preview of what I am recording.

 

[Nick Shaw]

I read an early thread 2008 I think which stated RcX clips at iso 200. Is this true for the current build
Paul

When you say clips, I think what whoever said that meant was if you reduce the ISO below 320, you are effectively gaining down, so image data which was clipped will still be 'flattened' by clipping, but will be dropped to a level lower that 100% on the histogram. There will be nothing to the right of this on the histogram, as there can't be any information higher than the clipping level.

This is and will always be the case, although HDRx confuses the issue slightly!

 

[Pablo Villegas S.]

I really don't get how not getting info to the sensor will limit noise?

Well, when the sensor is rated at 320, and you shoot at let say 160, you are overexposing by 1 stop, so there is actually more info getting into the sensor than it should, you are basically moving your medium gray exposure up the curve, so you gain more info in the shadows and less noise.

When you do the opposite, rating the camera at 800, you're under exposing by 1.5 stops, and moving the medium gray down the curve, to gain latitude and information on top, that is OK in daylight exteriors since most of the frame is well lit, but at night that might create noise, since most of the frame will be in shadow or darker. That's why I'm asking if the noise level by underexposing at night is acceptable, and apparently it is.

I'm trying to use available light and practicals as much as I can, so i need the 800 speed.

 

[Pablo Villegas S.]

When you shoot natural light-only, the rating of the sensor doesn''t really affect what is actually recorded, so I'd set the ISO at whatever is convinient to have an ok preview of what I am recording.

It depends, if you're monitoring and exposing on Redcolor, it does.
You're either underexposing or over exposing the Raw file, the Histogram moves to the left or right.

And if you're exposing using RAW, then it really makes no sense moving the ISO at all.

 

[Gunleik Groven]

It depends, if you're monitoring and exposing on Redcolor, it does.
You're either underexposing or over exposing the Raw file, the Histogram moves to the left or right.

And if you're exposing using RAW, then it really makes no sense moving the ISO at all.

It actually doesn't affect what is recorded at all, only how it is displayed.

Do this:

1. Set a scene exposed for say 320 @ 5,6 - expose
2. Crank FLUT and ISO to the full without adding ND or changing aperture or shutter - expose

Take the two exposures into RC-X and set the ISO/FLUT at the same values.

If the images aren't identical as to exposure, I'll send you a crate of beer from Norway, and with the price of beer here, I wouldn't have made that bet if I wasn't sure I'd win...

 

[Clint Lealos]

Gunleik, I like the cut of your jib

 

[Pablo Villegas S.]

It actually doesn't affect what is recorded at all, only how it is displayed.

Do this:

1. Set a scene exposed for say 320 @ 5,6 - expose
2. Crank FLUT and ISO to the full without adding ND or changing aperture or shutter - expose

Take the two exposures into RC-X and set the ISO/FLUT at the same values.

If the images aren't identical as to exposure, I'll send you a crate of beer from Norway, and with the price of beer here, I wouldn't have made that bet if I wasn't sure I'd win...

Well, I've never used RC-X.

But if I move the iris the RAW image changes, at least the Histogram does. Is this correct?

So if I close or open the Iris based on what ISO I'm using to expose, the RAW image will change.

 

[Costelloe Michael]

Well it doesn't...

Thing is that this is a bit situation-dependant.

If you light your scene - even at night, you "could" rate your ISO lower to "force" the lights up, thus getting more image/less noise/higher contrast.

When you shoot natural light-only, the rating of the sensor doesn''t really affect what is actually recorded, so I'd set the ISO at whatever is convinient to have an ok preview of what I am recording.

Dude depends on your post route and who is looking at what you do. You shoot for yourself?

 

[Paul Harding]

It actually doesn't affect what is recorded at all, only how it is displayed.

Do this:

1. Set a scene exposed for say 320 @ 5,6 - expose
2. Crank FLUT and ISO to the full without adding ND or changing aperture or shutter - expose

Take the two exposures into RC-X and set the ISO/FLUT at the same values.

If the images aren't identical as to exposure, I'll send you a crate of beer from Norway, and with the price of beer here, I wouldn't have made that bet if I wasn't sure I'd win...

That,s my conclusion after testing this week
Spot on
Paul