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 :)

Histograms

Hey all,

thanks for the feedback :)

Please note that I wanted to provide a layman's description of how this works, to give 95% of readers an easily digestible version they could understand. I know that there's more to it than what I described and yes, strictly speaking you don't lose half your dynamic range if you underexpose one stop.

But how many cinematographers are computer technicians and video path specialists? I like the KISS principle when teaching so as to give people an easy way to remember in broad strokes how this stuff works. A sentence like "underexposing one stop effectively halves your dynamic range" keeps this simplicity and makes people remember to expose for the way Red works, not the way film works.

HTH and all criticism taken in good spirits! :)

Paul
 
Jones said:
It seems to be that way. Although it would make way more sense to make the histogram showing ONLY what is definitely captured in the RAW image that has been created untouched by the ASA setting, for it´s metadata only anyway.
Click to expand...

Well the Raw view on the camera would show you that. I guess you could also see that in Red Alert by choosing linear light as the gamma space, although it would look very different as the camera has a gamma curve applied when displaying in Raw. I suspect they would show the same clipping point.
 
Ted Chu said:
http://www.luminous-landscape.com/tutorials/understanding-series/understanding-histograms.shtml
Click to expand...

The stuff behind this link is NOT correct, BTW.

"the total range of brightness values encountered in the real world is only about 10 stops..."

- Bull. The difference in illumination in everyday scenes may be somewhat in that range (Day EXT, Night INT), but the differences in brightness values you might come across are far, far bigger (The sun vs. dark object in candlelight).

"This view of the rear LCD on a Canon 1Ds shows a histogram for a particular shot and also the dotted vertical lines that Canon has engraved on the display separating the 5 stops of dynamic range that are available."

- Obviously bull too, i.e. Red one can capture over 11 stops of range, so do DSLR cameras - even home video cameras can exceed 9 stops. The histogram is divided in 5 segments, but those segments are NOT one stop each.

"There wasn't an exposure with today's digital (or transparency film) cameras that could encompass the full dynamic range of this photograph — which is about 8 stops."

- Bull again. In the example images the author has "underexposed" image where he says the shadows are crushed... well, they are not. The shadow detail can be retrieved even from the highly compressed jpeg he uses as an example. He then goes on to make a HDR composite of two exposures - a similar (but possibly better looking) image could have been made from the darker single image with simple gamma curve adjustment (assuming the raw file was available)...

...etc.
 
Ok, so based on several hours reading of these forums I've come to these conclusions regarding metering exposure on the RED. I'd be grateful if people with experience could confirm them, or let me know what I've got wrong.

1) The camera ISO setting does not change the sensors' sensitivity, this is hard baked in to the sensor hardware to be about 320 ISO. Nothing changes it.

2) Shooting at a camera ISO setting that is higher or lower than 320 does not change the recorded RAW image data, it is just a number that is recorded as metadata and applied toe the RAW date in post later, or when viewing the data (in software or from the camera).

3) Unless you are monitoring the image in RAW viewing mode while shooting, the Histograms and zebras DO NOT reflect the RAW image data, they reflect that data AFTER the color space and output LUT's you choose for viewing have been applied to it.

4) The only camera meters that DO reflect the RAW image data all the time are the "Traffic Lights" and the small "RAW meter".
A purple color at the top of the "RAW Meter" would indicate clipping in the highlights of the RAW data.
Any of the RGB "Traffic Lights" being illuminated would indicate clipping in the respective RGB channels individually.

5) The "False Color" meter is NOT currently accurate when viewing in RAW mode, how accurate or inaccurate it is is not known, so you probably shouldn't use it if you're doing critical work.

6) A workable way to accurately judge exposure while shooting is to have one of your buttons setup to toggle to RAW viewing mode, and ONLY trust your Zebras and Metres for exposure purposes when you're in this viewing mode.

Finally, I'm gleaning that the attached pic below is ACTUALLY what I shot at the RAW level, totally clipped in the highlights. Because of lack of official documentation on this subject (there's nothing about it in the user guides) the camera essentially "foooled me" into thinking I was within good exposure limits. The Histograms were not telling me what I logically thought they would be, which is reading the RAW data.

Note that the attached screen grab shows the RedAlert color space set to "Camera RGB", the Output LUT/Gamma Space set to "Linear Light", and the ISO set to 320, with the Zebras on at 95%.

Many thanks for your feedback.
 
All seems correct.

I always leave the camera at 320 because I don't trust the meters at anything else. They may have fixed them recently.

It's not necessarily a good thing to just go on RAW data for exposure. For a simple workflow, you may want to use redspace which will give you nice highlights and lots of highlight protection, as well as more of WYSIWYG image. RAW will get you the most out of the sensor if you have the time to play in post. REC709 is somewhere in between.

Notes:
4) The trafic lights reflect the colour space you are in: REC709, Redspace, RAW.

5) Not sure on that one

6) I set button 1 to "view raw" it saves accidentally taking a white balance when you press side record.
 
RAW view is just a linear LUT applied to the RAW data, should give you a flat, bland image. False color should be working fine with it. Seems to for me? Build 18 beta. False color can be useful at times, like to check 18% grey, flesh tone luma, etc.. But I spend most of my viewing time in Raw View with edge highlight, focus assist and zebras.

In Moggy's shot above, it looks to me like the RAW is definitely clipped and the camera probably didn't show all the info needed because of the 160 ISO setting. Also looks like it should have been exposed more for the highs... ETTR and all that. I'm curious as to what the image looks like on a good monitor or playing with the exposure slider without it being converted to an 8bpc jpeg.

REC709 is the devil unless you're doing the native post route through Adobe CS4, then it's just necessary evil.

I always leave the camera set at ISO 320 and 5000K ~ native for the sensor, why stray when it's all metadata anyway?

I also recommend getting the Auto-WB removed from the buttons by reassigning your buttons. Auto-WB, whether pushed accidentally (intentionally done on every set-up is even worse), is a pain. I have seen that screw up more post workflows and cause more panic than any other camera button or setting. And most DPs coming from an HD perspective are always in a habit of setting their white balance, which is good, but on RED it's not needed. And is a real problem with various workflows where changing the clip values or shot metadata is not an option without transcoding.
 
Thanks so much for the feedback, hugely helpful. I'm just glad this was only just a test shoot and nothing important.

I guess my only question in all of this now would be, why don't the Histogram and False Color meters ALWAYS read the RAW data?. It seem to me that exposure of what's actually being captured by the camera, i.e. RAW data, is the bottom line thing you want to get right all the time in a real world shooting situation, not some misleading version of that data. (...and why isn't this all in big print in the manual? )

I guess I don't understand what the value is of knowing the exposure distribution of a Rec709 or RedSpace version of the RAW data is.

Thanks,
 
Paul Didier Mogg said:
Thanks so much for the feedback, hugely helpful. I'm just glad this was only just a test shoot and nothing important.

I guess my only question in all of this now would be, why don't the Histogram and False Color meters ALWAYS read the RAW data?. It seem to me that exposure of what's actually being captured by the camera, i.e. RAW data, is the bottom line thing you want to get right all the time in a real world shooting situation, not some misleading version of that data. (...and why isn't this all in big print in the manual? )

I guess I don't understand what the value is of knowing the exposure distribution of a Rec709 or RedSpace version of the RAW data is.

Thanks,
Click to expand...

Absolutely right. Doesn´t make any sense to me too...

I´m prepping for a documentary shoot in British Columbia right now and I´d love to have something to rely on...gonna do tests in the woods here today...
 
Well, I can see where you are coming from, but at some point you are going to move out of raw and into a REC709 or Redspace. So if you have ETTR with Raw, you will be clipped and have to bring down the exposure. Also, bear in mind that using raw for clipping point is okay, but what about other areas of exposure? If you use redspace you will get a good idea of where 18% grey is etc.
 
Jeff Kilgroe said:
R

REC709 is the devil unless you're doing the native post route through Adobe CS4, then it's just necessary evil.
Click to expand...

Jeff, we just started to check CS4 for RED workflow. Can you explain that ?


Thanks,

Frank
 
Very informative thread, thanks all. I'm left with one dumb question, though. I like the raw view, I generally use it to make my exposure decisions, being confident that "what the sensor sees" is the bottom line. Now I'm confused, and this thread doesn't quite answer my simple dumb question. Suppose I must work in rec 709 color space. Since my exposure was set according to the raw curves + ETTR rule, my highlights will display clipped, and so will be my color channels. However, nothing has gone wrong at the sensor, plenty of space... Well, shall I be happy with the space I'm left with in raw ? I mean, shall I be able to get back some detail in my highlights in post when I go the rec 709 route ? Any simple answer for the layman ?
 
Watch the rising colour bars in between the histogram and RGB "traffic lights". Those are exclusively RAW so if you don't see the top red colour on that you have room to play.

HTH

Paul
 
I believe the histogram is accurate enough to get a good exposure in RAW view (see Graeme's post at the top of this thread):

http://205.234.135.241/forum/showthread.php?p=349313#post349313

False colour however is definitely NOT accurate in RAW view (I'm not sure how far it's out, maybe a stop, stop and a half...basically it's just not reliable enough in RAW view) - see Stuart's post here:

http://205.234.135.241/forum/showthread.php?t=21635

The new colour bar (with the rainbow colours) is also another RAW exposure meter (I think I'm right in saying this is quite handy because if you're viewing in, say, Rec709 the rainbow colour bar is simultaneously showing you what is going on in the RAW world...more or less).
 
Jones said:
Jeff, we just started to check CS4 for RED workflow. Can you explain that ?
Click to expand...

The Adobe CS4 plug-in assumes REC709 colorspace at this time. Hopefully with the next update we'll have more color spaces and eventually support for other LUTs / curves defined elsewhere.
 
You must log in or register to reply here.
Back
Top