Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

Best Practices. Obtaining Proper Exposure. Share your methods please.

The reason for sticking at 800, is that, since it is the recommended "native" setting, you should get the best idea of clipping/noise while shooting.

You can change the ISO in post. You can alter it in RedCine-X as soon as the footage leaves the camera. The reason for setting it at 800 and leaving it there is to remind you that it's JUST meta-data. You are not, at all, changing what the sensor is seeing. You are not, at all, altering your raw data. If something looks Good at 320, or lower, you may get it into post and realize you clipped some things. If something looks good at 1200 or higher, you'll load it up and see a bunch of noise. If all you have is a small monitor, it is very easy to either clip or get noise. I know it's easy, because I've seen the bad results of professional crews switching up or down their ISO, thinking it's actually changing what the camera sees.

It's not.

I can certainly see making it a habit to set your ISO down a stop or two in order to limit noise. I would not go farther than 320. I can also see bumping up your ISO when you're in a low-light situation and you don't have much choice. But! Understand that when you do this, you are not adjusting what the sensor is capable of picking up and recording. All you're doing is helping you see, while you are on set, what the camera can detect. And the end result will be noisey.


Because, again, METAdata. Do not think you are actually altering what the sensor is detecting in-camera. This is a dangerous line of thought. Which is why, for any beginner, I stress: set it to 800, and forget it's there. Because if you use it as a tool without actually knowing what you're altering, you are going to end up with bad footage.
 
Gavin Greenwalt said:
Native if you want to use every last bit. But I disagree. The bottom stop is too noisy in my opinion and I can play with the highlights more at 320iso. They clip 'unnaturally' soon at 800. I'll be shooting 800 and the histogram will be registering white. I'll only change it to 320 iso and they'll drop below white. That tells me that 800 is a slight gain beyond what the sensor is actually registering.
Click to expand...

In RC-X and redcolor2, it seems like white clip level falls just below 100% at ISO800, but as you adjust ISO lower, the clip level drops, but it does not seem to reveal more highlight detail... What gamma mode/camera are you using? What you are describing does tend to occur using R1 and redgamma1, but not using redgamma2. In redgamma2, ISO320 should have less compressed highlights but less highlight range than ISO800.

Personally I have found redgamma2 works pretty well at ISO800 and it scales the camera's usable range fairly well for about +-1.5 stops but falls apart a bit beyond that as far as maintaining shadow and highlight levels without throwing anything out and looking natural. I am am considering using Redlogfilm which seems to scale and use the camera's range a bit better, and it would be easy to use additional user curve look options to correct/de-log it to use it for a great looking image in camera. I plan to test and tweak this strategy more fully when I receive my Scarlet.
 
Nick Pasquariello said:
The reason for sticking at 800, is that, since it is the recommended "native" setting, you should get the best idea of clipping/noise while shooting.

You can change the ISO in post. You can alter it in RedCine-X as soon as the footage leaves the camera. The reason for setting it at 800 and leaving it there is to remind you that it's JUST meta-data. You are not, at all, changing what the sensor is seeing. You are not, at all, altering your raw data.
Click to expand...

When changing ISO on RED you are not changing what the sensor is seeing, but when you change ISO/gain on just about any digital camera ever made you are not changing what the sensor is seeing. What is important is that when you change your ISO on RED cameras, you are changing what YOU are seeing, and for many people who light to the monitor and for many people who with good reason want to get as close to their intended look in camera, it is very important to know how to use this to your advantage and to often change your ISO to get the most out of the sensor and get the image closest to what you intend, making less work for the colorist, less likelihood of your images going through post and looking completely different from what was intended on set, etc.

It may be metadata, but that doesn't mean you shouldn't use it and configure it properly on set. Would you leave your WB alone and just shoot everything at 5000k or 5600k even if all your lights are <3200k? No, you wouldn't except in very extreme circumstances, you want to light and shoot and monitor with an image as close as possible to the image you want on screen because that gives you and the director and the client and whoever else the confidence that it can look that way and that they wont have to spend unnecessary amounts of time and money to fix it in post, and it gives you confidence because it means that if they decide to do some transcoding workflow where they just export your R3Ds to Prores/DNxHD/DPX/etc without any adjustments and without your supervision, you know your images will still look the way you wanted and if they don't you did everything you could to make it an easy process to get the proper look from your images no matter the post workflow.

Yes metadata means you CAN change it in post, but it doesn't mean it wont affect the way you light for your camera on set, and it doesn't mean you or your production will have the time or the money or the resources to do so properly in post all the time. If you are shooting and editing and grading all your own material, feel free to do what you want, but most people are not working that way.
 
I agree with just about everything you just said Noah, but altering the Kelvin a bit is quite different than altering your ISO too much, and thus ruining your footage. And that's the key here: if you alter the ISO too much, you may think what you're seeing on set is great! But then, post production gets it, and it's . . . not. I know this can happen because it's happened twice in the span of 2 months to us, once with the ISO too high, and once with it too low.

And so, I caution people against changing the ISO on a Red camera because it's too easy to fall into the mentality that it's actually altering your "film stock" when it really isn't. My personal word of advice is to do as much as you can without altering the ISO. For me, it's the last thing I would tweak to get the look I'm going for.
 
I've been thinking about this issue with respect to the MX's native white balance too.
To preserve the most information, I should be rating the sensor at 800EI 5000k and using 80A filters to balance tungsten light, otherwise I could be clipping the red channel. I'd do it that way with film. Seems to be the same principles at work. In practice though, i can imagine the light loss of the filter might outweigh the correction across colour channels and result in more noise where i can't compensate with more light. I'll be keen to test this when my Scarlet arrives in a couple of weeks. Has anyone been white balancing in camera with filters for the above reason or have any thoughts on this?
 
while I am not sure we have settled everything here (yet), esp for our poor OP, I appreciate this discussion! Coming from a video background (and non-digital 3CCD Betacam in my case) I am realizing how much there is to learn here. And I have been doing this for 25+ yrs. Thanks guys. keep it coming. My scarlet is due next week...this is fab.

Stu Aull
Alaska
 
No kidding Stu. LOL I think I'm even more confused now.

JK
 
Nick Pasquariello said:
I agree with just about everything you just said Noah, but altering the Kelvin a bit is quite different than altering your ISO too much, and thus ruining your footage. .
Click to expand...

I would argue that it is not really very different, one adjusts your RGB curve and the other adjusts your R/B curves primarily. If you adjust your WB to 3200k you are pushing your blue channel 1.5 stops to compensate for underexposure. Even more in candle light (2000-2500k). So imagine what that means if you shoot ISO800 3200k, your blue channel may be operating at the equivalent of ISO2500, more if you go lower than 3200k or higher than ISO800.
 
While I do shoot at ISO 800, my meter reads correctly when set to iso 320. Also, when you hit the RAW button, it goes to ISO 320.
The reason to shoot at 800 is for highlight protection. Expose using the histogram, false color, and stop lights and you are good to.
 
Troy Bakewell said:
No kidding Stu. LOL I think I'm even more confused now.

JK
Click to expand...

So sorry. :( It was difficult for me to comprehend as well, until I began to shoot more and more and even learn a bit by trial and error.

To make things simpler, I'd suggest you start shooting as if you were rating the camera at 800 ISO. Shoot shoot shoot, and see what you are finding.
 
Uli mentioned back on the first page that he recommends to "expose to the right and check the histogram..." - Could someone expand on what expose to the right means and why we should do it. Also what does a correctly exposed histogram look like on and Epic?
 
Mike Van Laanen said:
Uli mentioned back on the first page that he recommends to "expose to the right and check the histogram..." - Could someone expand on what expose to the right means and why we should do it. Also what does a correctly exposed histogram look like on and Epic?
Click to expand...

The right side of the histogram is toward the highlights. When dealing with digital image sensors, the primary concerns are shadow noise and highlight clipping. If you overexpose, you will at some definite point lose highlight detail and the sensor will just see bright elements above that luminance level only as the same solid white. If you underexpose, you can get varying degrees of noise which get worse and worse as an element falls more in the dark. The cleanest and most adjustable image in most situations will be one where you "expose to the right", as in tend toward over-exposure, as much as possible but still trying not to over do it so as to protect highlights from clipping and becoming unrecoverable as detail in your image. If you are setting your stop or ND or shutter to adjust exposure, you can usually expose to the right the best by opening up as much as possible but not so much that RED's stoplights turn on (indicative of highlight clipping).

This will work for most scenes. It may not work for scenes with extremely high contrast. If there is too much contrast in your scene and you use this method, you may end up underexposing important parts of your image and exposing them to noise if you protect highlights too much, or you may end up blowing out your highlights if you expose too far to the right. With Scarlet, when shooting in scenes with extremely high contrast, you have the option to shoot with HDRx to capture that wide contrast range, or you will have to choose how to balance your blown highlights and noisy dark areas.
 
A light meter is not obsolete yet! You need to know your scenes contrast range, period.

Exposing to the right is a simplified version I gave. Of course you would not protect a tiny highlight (like a reflection on a small object or a small light source) at the expense of drowning important parts of your scene in noise. As long as the scene contrast is less than the DR of your camera, staying just below clipping is fine to minimize noise. The image may look overexposed at 800, but it isn't. In the end it boils down to this: you got to know your tool completely to get the best images. Test!

To avoid confusion for beginners, I'd even suggest to RED to indicate other ASA/ISO values than 800 in yellow, just like off speeds for varispeed. If I think about it, one might even mark other Kelvin values than 5000 in yellow. If your background is in film, just make one thing very clear: you can't change your sensor like you change a film. If you are changing ISO it's like a push development or under developing.
 
Oh yeah and then there's also the fact that the ADC is linear light and you'll be getting more bit depth (which helps avoid posterization effects especially after post correction) if you capture your image in a "brighter" parts of the sensor's range.

Basically with Scarlet and most digital cameras, the top stop is probably expressed with a depth that amounts to half of the possible values expressed by the ADC's particular depth (probably 32768 values are available for the top stop of detail in Scarlet's 16bit pixels), but the lowest stop (probably just noise) is expressed with a depth of only 2 values. The depth of these portions of the sensor is especially important because this linear light data with a 16bit depth (65536 possible values) has to get converted to a gamma corrected image, which expresses those ranges much more evenly, which means a lot of the depth of the higher stops is thrown out and the depth of the lower stops is really important because it will likely have to get expanded. So just for the sake of discussion, assuming Scarlet's 14th stop below clip is the last stop with any discernible detail not lost to noise, information falling in that luminance range will only be expressed by 8 possible values, and then when it is gamma corrected for 8bit viewing or more it might have to get extrapolated to a depth of 18 values (256/14=~18). You can start to see this at work when you shoot on RED at higher ISOs, where it's really stretching out the low end of the sensor and you will start seeing gaps in the histogram because the ADC has less depth than the 8bit gamma-corrected histogram. It becomes quite visible in the histogram beyond ISO2500.

This might help describe what I am explaining poorly
http://www.luminous-landscape.com/tutorials/expose-right.shtml
 
Gavin,

I agree with you. There seems to be a dogmatic 800 is best. But after shooting my Epic-M 12 hrs a day for 30 day's on my most recent project. The footage simply looks better and cleaner for me when exposed for 320iso. I agree that 800 is still clean and when you shoot exposed for that you give yourself "highlight protection" but really that's saving range for imprecision or mistakes. The theory is that set at 800 the camera tools tells you "uh oh, you're going to clip" but in reality you aren't. So if you do go there, you can recover later. But you still are giving up a range and amount of light that you could otherwise have. This is not a conversation about the dynamic range of the camera (saw a recent post piss Jim off there), nor the usability of the cameras at higher iso (I have gone up to 1600 before), but rather the idea that 800 is best. I equate it to shooting slide film. There really isn't much pushing and pulling in the development of slide film. What you shoot is what you get. If you want to operate with that precision, then you shoot 320iso and you will get all the stops of range the camera has, but if you burn or clip there, you're screwed. Better use false color, stop lights, histogram, all of them. It's gone if you miss. If you want an awesome balance between range and safety (more like negative film stock), set at 800 and go. You can fix some clip mistakes later because there is actually more range up top than some of the tools are telling you there is. It's personally much easier for me to shoot watching for clipped highlights, because I can register what that will look like in the footage, vs. noisy shadows because my brain can't process "how noisy?" until I get back to the shop (the purple predator view just tells you "where").

My advice. Shoot at 320, view exposure mode (recommended), traffic lights (they don't light up unless a certain percentage of the image clips that channel), and histogram. And don't clip anything that you can't lose forever (I prefer to clip nothing when at 320). And you will have the fattest negs out there.

If you are just starting, shooting fast, can't control your lighting (through filtration or otherwise), have large swatches of skies and snow, or just want a little insurance. Shoot 800. It will still look awesome and you will be "highlight protected".

And note when you click raw view, what settings it mirrors :)

What I want to know that I can't seem to find, is that imperially when set at 320 and at 800, what is the distribution around middle gray of the camera's dynamic range? Any help there would be appreciated.



Gavin Greenwalt said:
Native if you want to use every last bit. But I disagree. The bottom stop is too noisy in my opinion and I can play with the highlights more at 320iso. They clip 'unnaturally' soon at 800. I'll be shooting 800 and the histogram will be registering white. I'll only change it to 320 iso and they'll drop below white. That tells me that 800 is a slight gain beyond what the sensor is actually registering.
Click to expand...
 
Last edited:
Yeah guys. The video and especially the Luminous link really help. Thanks.
 
You must log in or register to reply here.
Back
Top