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 :)

Native Dragon ISO…once and for all!

AndreeMarkefors said:
Really? I could understand if there was an option to display an ISO corrected histogram, but it should be possible to get the raw histogram as well…

Anyway, the goal posts are always raw, right? Or is it the stop lights?
Click to expand...

one thing you can do is switch to raw view as a baseline. This shows you 800iso rg2 and RLF so it does show you at an ISO 800 reference what is hitting the sensor as a representation of ISO 800, if you were using a light meter.
Exposure tool and stop lights give you sensor clipping and noise, but it is (I think) based on that ISO 800 raw view, which was until very recently about 320 ISO.
 
sorry for being a nitpicker, but i kinda hate the "raw view" term, as it is anything but raw....

if i was to guess, the stoplights, goalposts and new "zone tool" are pulled from a "base" linearlight setting, and thus without iso or gamma correction. that "base" is not neccessarily "native" ISO.
 
I think I understand all of this now and it was kinda what I thought it was. I remember when the Red One first came out and local BBC TV show shoots
where doing just that; underexposing lighting and exposure whilst ramping ISO and then complaining the camera was far too noisy.

What threw me was the notion of high ISO's on high key lighting seemed a little counterintuitive.
I would have thought that ETTR would have been more appropriate on high key shoots than increasing ISO.
Am I right then that increased ISO on high key lighting helps with the sensor capturing midtones or have I got the wrong end of the stick?
 
ISO doesn't affect what the sensor captures

ISO is post-exposure and post-capture metadata in the signal chain, even when it is used as an exposure tool the way David suggests
 
Gunliek,

i agree, but I am just using the REDs terminology. I think the move to RG2/RLF 800 ISO was a smart move as default exposure values. It paves the way for LUT implementation and other important things, like post houses have a default view to apply to the footage pre-grading or pre LUT application. Gavin's LUTs in resolve are a huge help in establishing base looks (just as one example).

it probably should be renamed to something else like default or REDlog, it would be nice if raw view just showed linear light, but I think that would create more confusion on set.

What I think we really need to know more than anything is:

1.what data is clipped in the whites (unrecoverable)
2.what data is clipped in the shadows (unrecoverable)
3.what is the percentage of pixels in noise in the frame

The onboard RED tools do a great job on this, goal posts and stoplights, and exposure tool. The current raw view (baseline) is a great starting point for any grading, but if cinematographers are rating or lighting differently, the the processing step needs to be considered in terms of overall effect on the image, DR, noise levels and all those lovely factors.

Bottom line, if, as a very general rule, if people meter light and rate the camera at 800iso in just about any situation and when doing this you expose to stoplights and wind it back to eliminate them, have a low noise goalpost(50% or less) and a wide middle or right exposed histogram, then you can bepretty certain to have very acceptable images, that are highly gradable with good elasticity.

Battistella
 
Last edited:
David Ellison said:
What threw me was the notion of high ISO's on high key lighting seemed a little counterintuitive.
I would have thought that ETTR would have been more appropriate on high key shoots than increasing ISO.
Am I right then that increased ISO on high key lighting helps with the sensor capturing midtones or have I got the wrong end of the stick?
Click to expand...

depends where you want the majority of your stops over or below middle grey. So you pick a rating that gives you the stops you want or need over middle grey, balancing all other exposure factors.

ISO has no effect unless you use it as an exposure measuring tool.

Battistella
 
David Battistella said:
ISO is digital gain.

RATING at specific ISO is using ISO as a lighting tool. That is to say you are using ISO ona light meter as a basis for your lighting ratios.

if I rate at 250 ISO (I'm talking metered scenes here not run and gun, then I am bringing in the combination of light, aperture and shutter to move a histogram (a fat wide histogram for that rating) and more light hitting the photosites.

i can also rate at 800iso and set my meter to that rating. Then it's a different combination of light aperture and shutter to get that SAME fat histogram BUT (less light hitting the sensor)

i can also rate at 3200iso, still a different combination of light, shutter and aperture BUT (far less light hitting the sensor)

Less light = more noise

more light = less noise

moving ISO to "see" better is like turning up the monitor brightness. IT HAS NO EFFECT ON EXPOSURE WHEN USED IN THIS WAY

(You could use the brightness or exposure controls to accomplish the same thing)

if you don't have a meter and don't assign a scene a rating, then you are lighting by eye or available light and using ISO as a monitoring tool and not an exposure tool.

Battistella
Click to expand...

And that basically ties it all together. Thanks!

When you use a meter, and prefer a certain ISO as being the "native" ISO of Dragon (which may match or differ from RED's own declaration), and control the light hitting the sensor to get a fat histogram with that ISO setting, you are in a certain group.

Some, like myself, are in a different group and don't use a meter (although I'm starting to think I should), often work with available light, and understand that ISO doesn't essentially add a 'curve' to the RAW data that is irreversible.

What I do carry away from this convo is to stay at a lower ISO even when monitoring via a monitor, so that if the scene looks too dark/bright, the first instinct should be to adjust light hitting the sensor any which way besides raising/lowering ISO.

I'm oversimplifying, but this did clear up the whole 'shoot at a lower ISO' general statement made on these forums. Lot more context around it now. Thanks all!
 
Regarding Redcine-X: Can someone explain the difference, if there is any, between adjusting your image via ISO, FLUT, Brightness/Exposure or Curves? Is it just a preference, or does each tool affect the image differently? That is, are some tools better for maintaining grain structure, color saturation, latitude, etc?

Thanks,
Mike
 
David Battistella said:
2.what data is clipped in the shadows (unrecoverable)
3.what is the percentage of pixels in noise in the frame
Click to expand...

I think part of the problem here is that #2 and #3 are not precise, at least when compared to clipped highlights. Lows are never clipped, they just get forced into more or less noise, and one person's noise is another person's texture. Graeme made this pretty clear in his post about why the low goalpost value changes depending on RAW view status.

This topic is way too complex for many users of RED in my opinion. Notice I did not say "redusers". There are a ton of people who shoot RED and have no interest in this website or the minutiae discussed here. This was made very apparent at Reducation in LA, where many attendees had been shooting RED for some time but had no observable comprehension of how the exposure (and tools) really worked. I've since seen the same in users in the field.
 
very interesting discussion here.. great read btw on redcine, too bad more people don't check out those tutorials, they are great!
 
I'd like to make my earlier post in response to Jarek a little clearer. He asked why a person would counterintuitively use a low ISO in low light and a higher ISO in bright light. I don't really follow that general rule and shoot more like Phil Holland described, but I tried to give a possible reason based on how RED cameras behave regarding middle gray when exposed at different ISO values. Knowing how the dynamic range shifts around middle gray is useful information. It's something I think a lot of run-n-gunners haven't necessarily paid much attention to because the best way I know to do that is with a light meter - more on that later.

It seems to me that questions about the native ISO arise because people believe it's the one that will give the best results from the sensor. Fair enough - who wouldn't want to shoot at the point where image quality is maximized?

The problem is it's not quite that simple. This is where the term "exposure strategy" comes into play. Understanding what the sensor wants to see and the various ways of achieving that, given the shooting situation, is crucial in getting the best looking (and most flexible) image you can.

It's also important to understand how the camera behaves in response to different strategies.

Let's say, for the sake of argument, the native ISO of Dragon is 250. Knowing that only gets you so far before other questions invariably get asked. Should I shoot everything at ISO 250? What are the other ISO values there for? When should I use them? Are they just there for monitoring? And the big one these days: What do I do if the scene is dark and I don't have/can't use lights? I think people get frustrated and/or confused when they ask these questions and are given an "it depends on what you're shooting" type of answer...but it really does.

Over the years, this discussion has come up numerous times on Reduser and the best answer is always the same: get more light on the sensor.

This is often met with the internet equivalent of a blank stare. Many people refuse to accept it. I think it's because they've seen what they would consider good looking shots come out of situations with barely any light on other (often cheaper) cameras, so they feel light can't be the answer. There HAS to be something they can do with the camera to get clean, well exposed images in low light.

There is no secret sauce, special feature or hidden menu to be discovered. Every image capturing device needs light to do what it does. Some sensors are better at gathering light than others for various reasons but that doesn't change the fact that they give their best results when they have enough light to work with.

Beyond the limits of the light gathering power of sensors, there's gain. You can have gain with noise reduction or gain without it. RED does digital gain in the form of ISO without noise reduction. I think there are still many people who don't know this.

How one goes about getting light to the sensor is a matter of preference...or even habit. Speaking of which, I think for some, the way they've shot video for so long might actually be getting in the way of understanding how to properly expose on RED (or any digital cinema) cameras.

As for ETTR (Expose To The Right), it aims to maximize bit usage and eliminate noise. It can be used to great effect and everyone ends up doing it to a degree at some point, but it has what I see to be several major drawbacks to using it as a primary method of exposing:
1) Lack of accuracy - you never know how much light is actually present. Not only that, you don't know how bright an area is compared to another, also referred to as the lighting ratio. Saying a scene has "plenty of light" is just not very helpful.
2) Lack of consistency - without an accurate measurement or frame of reference, you're basically eyeballing everything or going by the video false color, which changes with ISO, which itself causes confusion, which leads to threads like this one...you get my point.
3) Lack of repeatability - you can't recreate lighting values later on if need be. This isn't much of a problem for one-off, run-n-gun type jobs but people should also know where a strategy is deficient.
4) Offers no solution for low light shooting - Expose to the right implies it's always possible to fill up your histogram on the right side just shy of clipping. It should actually be called "expose as far right as you can" because, when the scene is dark, the lens is open all the way, the shot is still underexposed and you don't have/can't use lights, exposing to the right isn't really an option.

In this age of digital video and HD monitoring, suggesting people use a light meter is basically tantamount to asking them to drink poison. I honestly don't know why everyone doesn't own one. Incident meters aren't that expensive. So many people are saving their pennies for whiz-bang equipment like handheld or flying gimbals and playing fast and loose with something as important and fundamental as exposure.

Light meters aren't the end-all, be-all and incident-only meters won't work in every situation. But, when appropriate, they're useful for understanding the light in your scene in a way your camera also understands it (middle gray, f-stops, ISO, frame rate and shutter speed). For those people having trouble with exposure, I'd suggest they at least consider buying a a cine capable incident meter (preferably with a spot meter attachment option), learn how to use it properly and watch the consistency and quality of their exposures improve. If nothing else it will tell by how much the scene is lacking light or overexposing so you can make a more informed decision about how to compensate. If you don't like it, you usually have a minimum of 2 weeks to return it.

If you need to shoot in low light, can't add light, are at maximum aperture and you can't alter shutter speed, you could simply use a different camera. Maybe a RED camera isn't the best option for the type of environment you're shooting in. There's no harm in saying that.

Ultimately, I think the best way to evaluate the camera's behavior and ascertain what's going to work best for a given situation is to test, test, test. No amount of us talking about it will substitute for that. It's the only way you're going to know what's acceptable to you and what isn't.

The key to understanding how the camera behaves is to test and the best way to test is to understand how the camera behaves.
 
excellent points... but going to your last point...if other cameras have in camera noise reduction on their gain/asa/iso and red does not...as we both know...

other then hiring a DIT to paint it with noise reduction id gather its hard to actually know when something is "too off to be recoverable"... i agree with red that having control of gain/grain addition/subtraction is better then them building a one size fits all solution that slightly softens everything.. (like other cam makers do)

but for low light shoots knowing when something is too off seems to be the problem...and hiring a Pomfort/live grade noise reduction feed on a shoot that can't afford light kits aint exactly an option is probably the reality

so for low light i guess sony/canon/panasonic is the way to go... another thing i think confuses people is they want to shoot with black blacks with noise vs shooting with less dark shadows and crushing the blacks in post...much better to do the latter... but getting people to understand these things vs them seeing finalized graded red results and coming from a video background are very different things. i came from a video background, but i think i still had some concept of the negative effects of digital gain, but this negative effect on noise when it comes to cinema cannot be understated cuz i dont really see red or alexa as great low light beasts when ive seen them pushed too far...


ps check out some low light stuff i shot on scarlet MX
https://www.youtube.com/watch?v=HA3LOhBRvp8&index=10&list=UUULBwUCMwe09ON3ghBDEH2w
https://www.youtube.com/watch?v=xGhgNkn2sNM&index=11&list=UUULBwUCMwe09ON3ghBDEH2w

something sort of interesting to note with this horrible t3.8 night time no light kit fireworks shooting is the noise/macroblocking changes significantly depending how much light is hitting the sensor...no noise reduction and straight out of camera...

the footage sucks i know, but interesting how much it can improve relative to how much foot candles are being thrown off by the fireworks...another interesting thing to note is how much noise there is as i cropped in...both videos shot same location/lens/aperture but one at 2k and one at 1k so i was using it basically to see how the fireworks would differ speed wise, but it helped the framing i thought in the reflection shot... in either case i think i needed a varicam 35 or sony a7s this day haha (or leica noctilux too, but considering this was a 8mm i would of needed to back up significantly more)
 
Last edited:
Does post-degraining get the Dragon into the same ballpark as the other "night" cameras? Like the A7S?
 
not sure, but id love to see people test this...i mean the A7S has a host of other issues from form factor to needing an offboard recorder to get a better master...that being said i think the new low light king on the way might be the varicam 35... id love to see a test that pits all three under the same lighting situations (both with graded results and straight out of camera results)
 
You must log in or register to reply here.
Back
Top