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 :)

Epic at 320?

"In fact, experience with DSLR's tells me that if anything, you get less dynamic range in higher ISOs." - the RED is not a DSLR. On a typical DSLR DR is ISO dependent. On a RED the DR is constant on change of ISO.

Graeme
 
Graeme Nattress said:
http://www.red.com/learn/red-101/exposure-with-red-cameras

Graeme
Click to expand...

Okay, take this image:
HighlightProtection.png


Let's say you've got your image exposed correctly with the "midtone at lower ISO." Now you move the ISO up two stops to the "midtone at higher ISO" spot. Now you're overexposed by 2 stops and you need to stop down your iris to expose correctly again. Isn't that the exact same image, except with a little more noise?

The page implies the same thing: "In general, minor underexposure is acceptable and recoverable, whereas minor overexposure sometimes isn’t. Many therefore err on the side of less light in order to protect against highlight clipping." So all you're really doing is underexposing on purpose? Which also flies in the face of ETTR, no? (But that's a whole 'nother subject)
 
If you're ETTR can only talk about highlights. If you're under-exposed for the whole image then you're not ETTR. If you're ETTR on the highlights you could very well be under-exposed for the rest of the image though.

Graeme
 
Wait, ETTR is overexposing, right? So if you've got an image that's darkish, and none of the highlights are clipping, then you shift the histogram to the right side of the goalposts to give yourself more shadow detail, knowing that you'll bring it back down in post, right? Or do I have things backwards?
 
Your not clipping though. It's "Expose To The Right" without clipping. The point is to get the most exposure without clipping. NEVER CLIP unless your trying to...even then I would do that in post and not capture it like that in camera.
 
Matt Ryan said:
Your not clipping though. It's "Expose To The Right" without clipping. The point is to get the most exposure without clipping. NEVER CLIP unless your trying to...even then I would do that in post and not capture it like that in camera.
Click to expand...

Okay, glad I got at least *some* terms right ;) ISO is more confusing though. The way I see it, there's only two ways it can work: a) the sensor has a native ISO and the sensor gets what the sensor gets, and with an iris (and ND, etc) you can change what it gets, but ISO only changes your perception of what it gets. Or b) the ISO setting actually changes the sensor's sensitivity to light. Or maybe there's a third way? I thought it was "a", but now I'm just confused.
 
ISO referred to the speed of film. To change ISO meant to change film. Sensors are fixed. Their "speed" can be changed via "gain" and there's two ways to do that -analogue gain and digital gain. Neither change any property of the sensor, but how the electrons read off it are dealt with. Analogue gain amplifies the electrical signal before it gets digitized, which can in some designs lead to lower noise at higher gains at the expense of reduction in dynamic range, whereas digital gain amplifies the captured numbers and leads to constant dynamic range at all gains.

Graeme
 
Graeme Nattress said:
ISO referred to the speed of film. To change ISO meant to change film. Sensors are fixed. Their "speed" can be changed via "gain" and there's two ways to do that -analogue gain and digital gain. Neither change any property of the sensor, but how the electrons read off it are dealt with. Analogue gain amplifies the electrical signal before it gets digitized, which can in some designs lead to lower noise at higher gains at the expense of reduction in dynamic range, whereas digital gain amplifies the captured numbers and leads to constant dynamic range at all gains.

Graeme
Click to expand...

Thanks, I didn't know that! So which way does Red do it? And which ISO setting is applying no gain? I thought it was 800, but it sounds like maybe it's 320? When you set ISO to less than that, is it applying negative gain, and is that worse, in theory? (I can't seem to tell a functional difference.)
 
Well..... RED uses digital gain, but it's not a "pure gain" function like an analogue amplifier, but a digital gain that is bendy - the famous FLUT curve that keeps highlights as the gain increases.

That means it's very hard to define a zero-gain, and what we say is that with ISO800 you have a great starting point for Epic images, just like back in the old days of original RED One, we said ISO320 was that point.

Graeme
 
Mark Toia said:
becuase I never ever look at my histogram.
I just look at the monitors... If the whites look blown out, I pull back....
Click to expand...

I have a silimar question about ETTR. I shot a few tests last week and we threw a bit of mist into the room for this particualr one. Firtly I looked at the monitor by eye to expose for a nice looking image, but the histogram read way under even though the image looked ok on the monitor. I was particualrly looking at the skin tones. With a small tweak in the curves - to me this is what looked nice in Premeire although I knew it was reading underexposed when we shot it.

http://www.darklineonline.co.uk/uploads/under.jpg


Now using what the histograms said were correct I opened up the iris and took another exposure of the same image, but on the monitor it just looked horrible, even though the histograms read correctly.

http://www.darklineonline.co.uk/uploads/normal.jpg

Finally I took that image in premiere and applied a monsterous curve, to see if would match the other exposure I had taken - and it did :

http://www.darklineonline.co.uk/uploads/normaladjusted.jpg

So in some respects that's great as indeed I'm crushing the bottom image so much that it's super clean, better than trying to brighten the underexposed image....So clearly you should use the histogram - but why does it look so wrong as shot? I was with my cameraman and he said he'd never have pushed record with the image looking the way it did on the monitor. Everything was whiting out.

I have to say Ive also noticed it on other footage, you might see a nice blue sky and you're historgrams show no clipping, but when you look at the monitor it's pretty milky and approaching white. You have to do some serious curve work to get the sky looking like it did to your eye when you shoot.

Ps This was all shot with the new colour/gamma but I think premiere uses RG2.
 
Graeme Nattress said:
Well..... RED uses digital gain, but it's not a "pure gain" function like an analogue amplifier, but a digital gain that is bendy - the famous FLUT curve that keeps highlights as the gain increases.

That means it's very hard to define a zero-gain, and what we say is that with ISO800 you have a great starting point for Epic images, just like back in the old days of original RED One, we said ISO320 was that point.

Graeme
Click to expand...

Thanks Graeme, I thought 800 was the sweet spot. Those guys were scaring me with all that talk of 320 ;) Of course, if it looks good, it looks good.

Karl, I share your doubts about ETTR. I did a test once in the pre-MX days just like the one you linked to, except no mist. Took one shot properly exposed, and another overexposed a couple stops (but without any clipping). Except when I brought it into post, I couldn't get the overexposed one to look as good as the properly exposed one. Now, I only ETTR if the scene's supposed to be dark, so when they bring it down in post, there's less noise. Which I guess would be more like Expose To The Middle...
 
Yeh Joshua I have no technical grounding for what Im about to say but it seems as if there is some kind of tail-off which is now built into the gamma curve at the top end which never used to be there in the old colour science back on the R1. So anything caught in it is given a very washed out look. Personally I always found that when FLUT was introduced too. I steer clear of it because anytime I use that to brighten my image, even if I'm nowhere near white, the image starts to lose saturation and depth (for lack of a better word) . I have always stuck to ISO and curves. I see the same sort of visual effect here in the correctly exposed image (before I darkened it in post) that turning up the FLUT gives me - very milky bunched up tones that all push towards white but don't clip.

In theory if you don't clip you 'should' be able to get the same look by dropping exposure in post (which is why I did the test) - but I can't figure out why it looks so bad on the monitor if this is indeed 'correct' exposure for the image. I'm glad someone else has had a similar issue but it's worrying that you found you can't always replicate the look of the underexposed clip, even when you haven't clipped.
 
Karl H said:
Yeh Joshua I have no technical grounding for what Im about to say but it seems as if there is some kind of tail-off which is now built into the gamma curve at the top end which never used to be there in the old colour science back on the R1. So anything caught in it is given a very washed out look. Personally I always found that when FLUT was introduced too. I steer clear of it because anytime I use that to brighten my image, even if I'm nowhere near white, the image starts to lose saturation and depth (for lack of a better word) . I have always stuck to ISO and curves. I see the same sort of visual effect here in the correctly exposed image (before I darkened it in post) that turning up the FLUT gives me - very milky bunched up tones that all push towards white but don't clip.

In theory if you don't clip you 'should' be able to get the same look by dropping exposure in post (which is why I did the test) - but I can't figure out why it looks so bad on the monitor if this is indeed 'correct' exposure for the image. I'm glad someone else has had a similar issue but it's worrying that you found you can't always replicate the look of the underexposed clip, even when you haven't clipped.
Click to expand...

If you need to get the details in highlights back in post, take your shot exposed at 800 down to 320 in Premiere or RCX and then apply your curves. All that highlight info, color, and detail will be there if you didn't clip. You will just need to bring the rest of the image up slightly which shouldn't introduce noise as long as you don't push it too far.

Your sample images above look off. What color and gamma space were you viewing in? It almost looks like redlogfilm since it is so flat, but at the same time it seems too bright for redlogfilm.

The main reason you ETTR is to prevent noise and capture as much exposed material as possible which you later in post expand and play with. This gives you the greatest latitude in post since you a) have no noise in the lowlights or areas you want noise free and not crushed, and b) you haven't clipped your highlights so all the detail and information is there. This gives you the flexibility to switch your iso down to 320 in post and then work your color magic. Your full range of data is there at your disposal. It's not necessary to drop the iso in post, but it's easier/faster to get a better image that way since you instantly get your highlights back.
 
Hi Matt

I understand your points completely, the image above (the washed out one) was shot redgamma/colour 3 and nothing was clipping. It just looked horrible on the monitor when I expose as far to the right as I could. In prem it's displayed in these images as redgamma2 as they dont have the updated colour science yet, but there was little difference really. It looked just as bad on my LCD while shooting, as it does here.

The test above was to illustrate exactly what you said, can you retrieve the info and make it look like the underexposed image. But my question is why the image looks terrible when its correctly exposed? I had to really underexpose the shot (as in my first example) to get the image looking 'nice'. Or alternatively as you suggest, drop the ISO of the second image in prem to get a much cleaner image - but I don't understand why it looks so bad in the first place. I don't know if this was because the room was full of haze, adding to the problem? I don't see this behaviour normally but I do notice anything near to the right gets a washed out look (like bright blue skies), so if you have a lot of image detail falling in that part of the histogram then it's pretty ugly.

Mark said he exposes by eye and I wonder if this is one of the reasons why.
 
" It just looked horrible on the monitor when I expose as far to the right as I could." - exactly ETTR is about capturing data for grading, not for making a nice looking image directly. If you drop the ISO, as long as nothing is clipping in your ETTR you should once again get a "nice" image.

Graeme
 
This may already have been said, and if so, please forgive me, but your monitoring devices are usually not calibrated properly, so judging exposure off of that can be risky. I'd go by the histogram or scopes to evaluate your exposure. Compose your image with the monitor, but not your exposure. Also, trust your histogram.

Von
 
Karl H said:
Hi Matt

I understand your points completely, the image above (the washed out one) was shot redgamma/colour 3 and nothing was clipping. It just looked horrible on the monitor when I expose as far to the right as I could. In prem it's displayed in these images as redgamma2 as they dont have the updated colour science yet, but there was little difference really. It looked just as bad on my LCD while shooting, as it does here.

The test above was to illustrate exactly what you said, can you retrieve the info and make it look like the underexposed image. But my question is why the image looks terrible when its correctly exposed? I had to really underexpose the shot (as in my first example) to get the image looking 'nice'. Or alternatively as you suggest, drop the ISO of the second image in prem to get a much cleaner image - but I don't understand why it looks so bad in the first place. I don't know if this was because the room was full of haze, adding to the problem? I don't see this behaviour normally but I do notice anything near to the right gets a washed out look (like bright blue skies), so if you have a lot of image detail falling in that part of the histogram then it's pretty ugly.

Mark said he exposes by eye and I wonder if this is one of the reasons why.
Click to expand...

As Von says, don't judge the exposure based on how it looks on the monitor. Judge based on the histogram and you will be fine. That's something you'll have to get used to with shooting raw and ETTR. Judge the video feed the way you would a film tap, no one judges exposure off a film camera tap. Exposing by the monitor is also how you get inconsistent exposure from shot to shot, which even though it's raw, can lead to an inconsistent contrast ratio throughout a scene.
 
Matt Ryan said:
Exposing by the monitor is also how you get inconsistent exposure from shot to shot, which even though it's raw, can lead to an inconsistent contrast ratio throughout a scene.
Click to expand...

Not sure how use of monitor leads to inconsistencies, any more than using the histogram and being a slave to ETTR. Histograms don't provide any insight as to where you are placing neutral/grey & skin-tone in a scene. For me, when matching contrast ratios are critical nothing beats a light meter. With a meter, I can peg exactly where I am placing skin-tone. Not saying I don't also use a histogram. But, I tend to work more like Toia than what it seems is generally expoused as gospel around here with regards to the dogma to ETTR. This of course, does not mean that when I am using my meter I am not also thinking about the histogram, and ISO and headroom, shadow detail, contrast ratio, etc.

I've shot plenty of things at 320 and been VERY happy with the results. Some shots thats just what makes the most sense, particularly when I am concerned about shadow detail and noisy darks.
 
PHP:
Paul Nordin said:
Not sure how use of monitor leads to inconsistencies, any more than using the histogram and being a slave to ETTR. Histograms don't provide any insight as to where you are placing neutral/grey & skin-tone in a scene. For me, when matching contrast ratios are critical nothing beats a light meter. With a meter, I can peg exactly where I am placing skin-tone. Not saying I don't also use a histogram. But, I tend to work more like Toia than what it seems is generally expoused as gospel around here with regards to the dogma to ETTR. This of course, does not mean that when I am using my meter I am not also thinking about the histogram, and ISO and headroom, shadow detail, contrast ratio, etc.

I've shot plenty of things at 320 and been VERY happy with the results. Some shots thats just what makes the most sense, particularly when I am concerned about shadow detail and noisy darks.
Click to expand...

What I meant about exposing by monitor is if you adjust your aperture between shots based on what looks good, your going to have inconsistencies not only in contrast ratio but in DR as well. Your talking about using a meter which most people, Toia included, don't use one. That's what I meant above.

As for rating at 320, I do sometimes use it on sets where I have a controlled lighting condition or when I need minimal noise and have more control of my highlights. Other than that, I use 800. Look at Prometheus for example, that was shot at 800 ASA with some scenes being captured at 1000 asa. I didn't see any noise. So for me 800 is the best place to be in order to get the most payoff from the sensor with the headroom fail safe.
 
You must log in or register to reply here.
Back
Top