Pages: 1 [2]   Go Down

Author Topic: Setting for JPEG Camera display-5DIII, to get best histogram and RAW capture  (Read 147854 times)

SZRitter

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 384

not exactly - blinkies or zebra will show you where the clipping is exactly... histogram not... you certainly can make an educated guess, but the thing is that you have blinkies/zebra together with histogram vs just a histogram... a no brainer IMHO

I can agree to this, I just find blinkies/zebra annoying most of the time. But, it is preference. Although, I'm not sure I have had one that actually uses blinkies and zebras, usually it was blue and red, creating annoying blobs on screen. When I shot DSLR instead of mirrorless, not nearly as annoying.

ONLY when you know the correct ISO or how to compensate for ideal exposure (which is all ETTR is), that meter is as dumb about this process as your JPEG histogram is for raw data.

You know, I think we are saying the same thing, just dancing around how we get to it. My language may not be clear, and my thought processes are often hard for me to properly illiterate. Where you are stuck seems to be thinking that I expect a histogram to be 1:1 to the final image. Hopefully, I never said that. Through shooting with the same camera for a couple of years, I have developed an educated guess on the data the raw file contains. Just like your testing, where you took a lightmeter and ascertained a useable 1.47 stop difference between the meter and that particular camera, my experience has lead to being able to get an idea of how my histogram lays out on my camera vs. what it will get in Lightroom.

RE: Lightmeters - If I use my 558R, which has a spot meter, to meter the lightest and darkest parts,  assuming a sensor that gets about 12ev that is useable and the difference between the brightest and darkest spots are 12ev, should optimal exposure not be exactly halfway between? That is what I meant by using a lightmeter.

And a question on Lightroom, isn't the exposure adjustment basically adjusting Gamma? -4EV seems close to what I have seen for a 0 gamma from a RAW file, but I haven't done any testing or really researched the answer to figure out if this was coincidence or reality.
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20630
  • Andrew Rodney
    • http://www.digitaldog.net/

Where you are stuck seems to be thinking that I expect a histogram to be 1:1 to the final image.
No, where you are stuck seems to be thinking the tool is honestly defined for it's task. It isn't!
If I were to show you a Histogram of an image that is in ProPhoto RGB represented as sRGB, you're being lied to. If I tell you you're driving 55MPH and you're doing 66MPH, you'er being lied to. If you are driving 55MPH and the meter tells you 56MPH, not really an issue. The disconnect between a JPEG histogram in sRGB (maybe Adobe RGB (1998) ) and the raw data isn't 1MPH out of bounds! I asked you once, I'll ask you a 2nd time. Do you have a raw converter that will show you the actual raw Histogram? If not, I guess I'll try to spends some time making a screen shot for you of it and what the lie on the back of the camera tells me when I properly exposure for raw (NOT JPEG). It's not close. If you like being lied to, I'm OK with that and as I said, my only objection would be you attempting to pass this off to others as acceptable when we could (and can in some cases) get a Histogram that shows is what the data really is. That's not happening in your methodology. Please remember the old sayings about making assumptions and 'close enough for..." That's exactly where the 'use a JPEG Histogram for raw' falls flat.

I don't need to look at a lie and interpolate it into an 'educated guess'. I was exposing film years before anyone had an idea what an image Histogram was. I don't need to guess. Proper exposure is basic photography! It's not difficult to properly expose these differing media, been done for over 100 years.

You look at the LCD, zoom in and see the image is out of focus, it is out of focus! You view the LCD and see you cut off some's head, it's cut off! You view the Histogram, you're viewing the rendered JPEG the proprietary camera electronics produced which isn't the raw. It's as simple as that.

Please tell me, outside this camera LCD Histogram lie, where else you are shown that plotting of data that isn't representing the data as it really is. This Histogram as a tool is excellent for when the camera is set for JEPG and a kludge and lie when shooting raw. The answer isn't to continue to accept the lie, the answer is to demand the camera manufactures provide the truth about the data and stop accepting that the lie is 'close enough'.
« Last Edit: February 26, 2015, 10:52:31 am by digitaldog »
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

AlterEgo

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1995

No, where you are stuck seems to be thinking the tool is honestly defined for it's task. It isn't!
If I were to show you a Histogram of an image that is in ProPhoto RGB represented as sRGB, you're being lied to. If I tell you you're driving 55MPH and you're doing 66MPH, you'er being lied to. If you are driving 55MPH and the meter tells you 56MPH, not really an issue. The disconnect between a JPEG histogram in sRGB (maybe Adobe RGB (1998) ) and the raw data isn't 1MPH out of bounds! I asked you once, I'll ask you a 2nd time.

Do you have a raw converter that will show you the actual raw Histogram?

with my camera (Sony A7 now) I do compare specifically tuned OOC JPG histogram (UniWB, certain camJPG parameters to make if very flat) vs RawDigger histogram (as I use that to do the tuning) - and I get clipping indication with no worse than 1/3 EV precision... so it is 1mph - but OOC JPG is not very usable as an image, yes... but then I do not care about colors in EVF/LV... if you are talking about "usable" (as a finished picture) OOC JPG then true that its histogram might be far away from the raw data.
Logged

ThomasR99

  • Newbie
  • *
  • Offline Offline
  • Posts: 44

Wow...been a few days since I stopped by here, didn't realize the discussion had advanced so far!

I accept that the JPEG histogram is a 'lie', like the speedo that says 55 when you're doing 65.  My real question then is how do I make it as 'little' of a lie as possible?  W/the speedo I'd ask someone w/a calibrated unit to drive exactly 55, see where my needle points to and make a mark there.  Regardless of what the numbers on the backplate of the dial say, when the needle hits that spot I'm at 55.

So...what in camera settings can I use to get me as close as possible in a JPEG raw histogram to what the RAW h'gram would look like?

I haven't really taken a chance to 'calibrate' the camera, that is take a series of images at differing exposures and therefore different levels of 'blinkies' showing and then try to recover highlights in Aperture.  Might make an interesting exercise some day.

Thomas.
Logged

jferrari

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 484

I accept that the JPEG histogram is a 'lie', like the speedo that says 55 when you're doing 65.  My real question then is how do I make it as 'little' of a lie as possible?  W/the speedo I'd ask someone w/a calibrated unit to drive exactly 55, see where my needle points to and make a mark there.  Regardless of what the numbers on the backplate of the dial say, when the needle hits that spot I'm at 55.

Why would you even care about your speed if there were no police? With your 5DIII it is very difficult to get anything blown-out using internal metering. As long as you are shooting RAW everything can be adjusted in post. It's a different story if you are shooting JPEG and If you're only shooting JPEG's with your 5DIII you need to send it to me! I'll even pay shipping!      - Jim
Logged
Nothing changes until something changes.

Tony Jay

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2965

Why would you even care about your speed if there were no police? With your 5DIII it is very difficult to get anything blown-out using internal metering. As long as you are shooting RAW everything can be adjusted in post. It's a different story if you are shooting JPEG and If you're only shooting JPEG's with your 5DIII you need to send it to me! I'll even pay shipping!      - Jim
That makes no sense if one is shooting raw and attempting ETTR.
The camera will tell you that you are over-exposing and the subsequent in-camera histogram will also tell you that you are over-exposed and have blown the highlights.
With the 5D mark III there is at least 1.5 stops of highlight headroom on a raw exposure beyond where an in-camera histogram will tell you that the highlights are gone.
Noise issues with this camera make ETTR a much bigger deal than with any of the late-model cameras using Sony sensors.

As for the OP's question about how to get the best estimation of the raw histogram in-camera the answer I provided very early one - set the picture style to faithful. It won't be accurate for all the reasons very eloquently explained by Andrew Rodney but IMHO it is the best of a bad scenario. Roll on the raw in-camera histogram!

Tony Jay
Logged

SZRitter

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 384

I would also propose you do a bit of reading on gamma and sensors. The 2.2 gamma of a rendered jpeg is made to effectively give your system gamma of 1. Also note that as you approach the high and low end, your sensor will have an additional gamma that will lead to inconsistencies. Thus, not trying to cram everything into the top portion of the sensor's can lead to easier and more accurate colors.

http://www.cambridgeincolour.com/tutorials/gamma-correction.htm

And for giggles, an article that basically tells you that there is no 100% proven exposure technique to work with 100% of scenes.

http://www.cambridgeincolour.com/tutorials/digital-exposure-techniques.htm
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20630
  • Andrew Rodney
    • http://www.digitaldog.net/

And for giggles, an article that basically tells you that there is no 100% proven exposure technique to work with 100% of scenes.
Probably true but most pro's and serious photographers strive to get as close to that goal as their knowledge of the science of photography allows. And then there's "Close Enough" for the others...

A soft proof on the most expensive and sophisticated display products on the planet will never match a reflective print 100%. That doesn't stop some of us from attempting the highest degree match we can achieve.

Actually the article is wrong if you consider bracketing....
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

AlterEgo

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1995

My real question then is how do I make it as 'little' of a lie as possible? 

first of all you start by getting rid of WB = UniWB (raw channel multipliers as close to 1.0 as possible)
Logged

Jim Kasson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2370
    • The Last Word

first of all you start by getting rid of WB = UniWB (raw channel multipliers as close to 1.0 as possible)

Right. And here's a way -- several ways, actually -- to do that:

http://blog.kasson.com/?page_id=2387

Jim

SZRitter

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 384

Probably true but most pro's and serious photographers strive to get as close to that goal as their knowledge of the science of photography allows. And then there's "Close Enough" for the others...

A soft proof on the most expensive and sophisticated display products on the planet will never match a reflective print 100%. That doesn't stop some of us from attempting the highest degree match we can achieve.

Actually the article is wrong if you consider bracketing....

Most pros I know or have heard, strive to get the shot that works best for their needs. This often takes into consideration total workflow, and usually the image you have to mess with the least out of camera ends up being the best shot. But to be fair, I don't know too many pros.

Bracketing isn't always an option. I used to shoot a lot of events and sports, and bracketing was almost never an option if you wanted to get the right shot.
Logged

bernie west

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 323
    • Wild Photo Australia

Given this is in the Beginner's Questions forum, it is probably a bit much to expect the OP is a tranny sharp shooter.  In that case, the histogram is going to be of some use to him.  Understanding that it is lying and why it is lying is part of that process.  I never mastered film exposure (didn't do a lot of photography before the digital revolution), and I find the histogram useful in a broad sense.  I'm never going to rely on it for high accuracy, but it's good to get a rough idea of where you exposure is falling.  I recently changed from my old 5D to a Nikon, so I am yet to get a really good handle on the how the jpeg histogram might relate to the underlying raw data, but I'm still in the ballpark from my previous experience with the 5D.  With the 5D, I got pretty good at guessing raw ETTR from the jpg histogram.  It was even more challenging for a while, as I didn't realise my on-camera histogram could display each colour channel - D'oh!  So I had to make an allowance for what type of light I was shooting in in regards to what channel was likely to be more clipped than the others.  Also understanding how an image goes from raw to a jpeg, as I mentioned earlier, allowed me to understand which channels are more likely to clip in the jpg histogram.  Of course, none of this was an accurate science, but any information you can add into the process helps, IMO.

Regarding getting your jpg histogram to tell the smallest lie possible, in regards to ETTR, you can use the uni white balance technique that Guillermo van Luik developed here on LL a number of years ago.  I played around with it for a while on my 5D and found that all that pixel pushing/peeping was just too much for me to bother with.  And as others have said, I too don't like how highlights come out that have been captured just sort of sensor saturation.  These days I prefer to leave a little head room in the raw (by usually just over exposing the jpeg) so that the highlights look a little more natural. 
Logged

kirkt

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 604

Wow...been a few days since I stopped by here, didn't realize the discussion had advanced so far!

I accept that the JPEG histogram is a 'lie', like the speedo that says 55 when you're doing 65.  My real question then is how do I make it as 'little' of a lie as possible?  ...

Thomas.

Try using the Magic Lantern firmware for the 5DIII.  In Live View it provides a real-time raw histogram, with raw clipping indication in EV for the raw R, G and B channels.  If you choose to set up a UniWB, you can specify in-camera WB multipliers explicitly (RGBG = 1 1 1 1), instead of having to load a specific magenta raw file onto a memory card to specify as a custom white balance.  ML will also perform automatic ETTR if you enable it, among many other features and implementations that make a 5DIII a very versatile raw shooting tool, instead of a JPEG camera with raw file capability.

Kirk

 
Logged

Rainer SLP

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 727
    • RS-Fotografia

The best advise for a beginner is to tape over the back screen and ever worry about. ...

Interesting statement and yes that has from my point of view a lot of truth in it.

In film days there was no way to know how the image would come out (you knew it 2-3 days later) and that with the time made you have more and more experienced.

Nowadays I catch myself peeking after every shot on the display screen, in order to later find out that the images are better or worse as seen in the display so I try not to lurk at the display anymore and rely on my experience from film days ...
Logged
Thanks and regards Rainer
 I am here for

robertvine

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 56
    • http://www.robertvinephotography.com.au

Try using the Magic Lantern firmware for the 5DIII.  In Live View it provides a real-time raw histogram, with raw clipping indication in EV for the raw R, G and B channels.  If you choose to set up a UniWB, you can specify in-camera WB multipliers explicitly (RGBG = 1 1 1 1), instead of having to load a specific magenta raw file onto a memory card to specify as a custom white balance.  ML will also perform automatic ETTR if you enable it, among many other features and implementations that make a 5DIII a very versatile raw shooting tool, instead of a JPEG camera with raw file capability.

Kirk

 

I've been shooting with ML on my 6D for a couple of week now and the RAW histogram is really good. Has anyone noticed a difference in the histogram you see in live view and the histogram on the image review?
Logged
Robert Vine
Darwin, Australia Canon 6D +

Isaac

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3123

Nowadays I catch myself peeking after every shot on the display screen, in order to later find out that the images are better or worse as seen in the display so I try not to lurk at the display anymore and rely on my experience from film days ...

Don't throw the baby out with the bathwater.

Check the display for obvious problems that will only be fixable by taking another photo.

Logged

bjanes

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3387

Each camera is different but by and large, just ignore the histogram on the camera when shooting raw. It's mostly a lie.

Personally, I make frequent use of the histogram in my work, and ignoring it would be a mistake, IMHO. It is not a complete lie, but only the partial truth. One should perform tests to relate the appearance of the camera histogram to the status of the raw file. Here are results of such testing for the  Nikon D800e using a neutral picture control.

In this shot, the luminance histogram is just short of clipping.



The green channels of raw histogram are slightly more than 0.5 EV below clipping. This is a good ETTR exposure that allows 0.5 EV of highlight headroom. With the excellent performance of this sensor, it is not necessary to perform extreme ETTR.



Giving 0.3 EV additional exposure produces highlight clipping of the camera histogram.



The green channels of the raw histogram are about 0.3 EV short of clipping.



The take home point is that an ETTR exposure just short of clipping on the camera histogram will give a decent raw histogram, and slight clipping on the camera histogram will give an even better raw histogram, but with some danger of blown highlights. If one needs to evaluate the blue and red channel, UniWB can be used, but this is usually not worth the trouble unless one is dealing with saturated reds and blues where white balance will clip the camera RGB histogram with intact red or blue raw channels.

Regards,

Bill
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20630
  • Andrew Rodney
    • http://www.digitaldog.net/

It is not a complete lie, but only the partial truth
Gotta say, if I ever need a good lawyer Bill, you're my man  ;D
All kidding aside, yes, it isn't a complete lie. But why should we be told half truths? When will we get a true raw histogram without a hack?
And there's still a huge world out there that believes that what they see on the LCD histogram is factual, not a half lie. They believe if they shoot for that lie, their raws (and JPEGs) are being optimally handled which of course isn't the case.
As a guy who captured a lot of images over many decades on film, the need for a histogram isn't essential to me, it would be far more useful if it told me the facts, not a half truth.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".
Pages: 1 [2]   Go Up