Ad
Ad
Ad
Pages: [1] 2 3 4 »   Bottom of Page
Print
Author Topic: Color profiling while Exposing To The Right  (Read 12517 times)
erpman
Jr. Member
**
Offline Offline

Posts: 61


« on: September 20, 2012, 10:26:06 AM »
ReplyReply

Hi!

I just got a color checker passport that I plan to use for making location-specific DNG profiles. In the manual Im being urged to get the exposure right when making the reference shot. But this got me thinking; If I will be doing the actual shoot exposing to the right, which exposure will be correct for the reference shot of the color checker? Correctly according to the camera meter and then overexposing the shoot, or the same exposure for color checker that I plan to use during the shoot? (assuming of course that I dont blow out the highlights on the color checker with neither exposure).
Logged
Schewe
Sr. Member
****
Offline Offline

Posts: 5489


WWW
« Reply #1 on: September 20, 2012, 12:23:38 PM »
ReplyReply

No...if you are going to make DNG profiles your exposure must be correct because the profile software will be expecting the color patches to be within a nominal range. Over or under exposure will alter the accuracy of the DNG profile.
Logged
Scott Martin
Sr. Member
****
Offline Offline

Posts: 1312


WWW
« Reply #2 on: September 20, 2012, 04:41:56 PM »
ReplyReply

Well for starters, exposing to the right isn't "overexposing" and doesn't blow out anything...
Logged

mac_paolo
Sr. Member
****
Offline Offline

Posts: 423


« Reply #3 on: September 21, 2012, 12:54:23 AM »
ReplyReply

When shooting the CC24 I always look for a bright histogram.
Relying on the histogram itself watching out for blinkies, I'm pretty sure the final shot won't overexposed (JPG preview is very conservative!)
In the years I got some "Overexposed. Shoot again!" Messages when trying to maximize ETTR, so the point is: get a proper exposure right OOC and stick with it. Smiley
Logged
MarkM
Sr. Member
****
Offline Offline

Posts: 328



WWW
« Reply #4 on: September 21, 2012, 01:25:23 AM »
ReplyReply

No...if you are going to make DNG profiles your exposure must be correct because the profile software will be expecting the color patches to be within a nominal range. Over or under exposure will alter the accuracy of the DNG profile.

What is correct exposure for a raw file in this context? If a profile from a ETTR exposure is off, why would we expect a profile from a 'correct' exposure to be right if your workflow depends on ETTR?

In practice I don't think it matters that much because your ETTR exposure isn't going to be that far off when shooting the patches. How far can you really expose to the right before you blow out the white patch? Aren't we talking a half a stop or so for most cameras?
Logged

mac_paolo
Sr. Member
****
Offline Offline

Posts: 423


« Reply #5 on: September 21, 2012, 01:42:16 AM »
ReplyReply

What is correct exposure for a raw file in this context? If a profile from a ETTR exposure is off, why would we expect a profile from a 'correct' exposure to be right if your workflow depends on ETTR?

In practice I don't think it matters that much because your ETTR exposure isn't going to be that far off when shooting the patches. How far can you really expose to the right before you blow out the white patch? Aren't we talking a half a stop or so for most cameras?
I know that the DNG PE normalize the exposure when reading the patch. Given that the white patch is quite bright per se, you won't be able to go that far with ETTR, nor it is that useful.
The difference between Adobe Standard (or any other profile) and the CC24 profile is substantial. The difference between normal exposure and ETTR for the CC24 shot is barely nothing. Smiley
Logged
erpman
Jr. Member
**
Offline Offline

Posts: 61


« Reply #6 on: September 21, 2012, 06:13:10 AM »
ReplyReply

Yeah, I guess I wont get that far anyhow since the white patch will blow out pretty soon anywa. ETTR is for lower contrast subjects anyway.

But lets consider the following example: Im shooting a grey stone that fills the entire image. The brightest tones in the stone are mid gray, so I can shift the whole exposure +2 without overexposure. Wont exposing the color checker according to meter be misleading with regard to the actual subject?
Logged
mac_paolo
Sr. Member
****
Offline Offline

Posts: 423


« Reply #7 on: September 21, 2012, 06:22:45 AM »
ReplyReply

But lets consider the following example: Im shooting a grey stone that fills the entire image. The brightest tones in the stone are mid gray, so I can shift the whole exposure +2 without overexposure. Wont exposing the color checker according to meter be misleading with regard to the actual subject?
Honestly I can't figure out what you're trying to explain or achieve, sorry Sad
Are we talking about the ETTR in general or how to take the CC24 shot for the DNG PE?  Huh
Logged
deejjjaaaa
Sr. Member
****
Offline Offline

Posts: 884



« Reply #8 on: September 21, 2012, 09:45:28 AM »
ReplyReply

What is correct exposure for a raw file in this context?

check this = http://www.rmimaging.com/information/ColorChecker_Passport_Technical_Report.pdf

i 'd spot meter my camera off neutral patch N5 (has reflectance = ~18-20% and camera's metering is calibrated for ~12-14%) and I will not fill the frame with the target even with aperture stopped down... you need to fill only the area that tests for your lens show very very free from vignetting at stopped down aperture and also use a good tele lens + good hood... filling only central 25% (horizontally) of the image @ F5.8-F11 shall be something suitable... that is regardless of what DNG PE does inside to compensate for difference in illumination... because GIGO.
Logged
RFPhotography
Guest
« Reply #9 on: September 21, 2012, 09:55:35 AM »
ReplyReply

People are missing Jeff's point that "the profile software will be expecting the patches to be within a normal range". 

If you alter exposure of the chart you're going to alter the RGB values of the patches.  If you ETTR by a stop or stop and a half you may throw the RGB values of the patches outside of the range that the profile creation software is expecting. 

Example, I shoot a CC Passport at +1.5 Exp Comp and the light blue patch, top row, third from the left reads 83%, 84%, 94% in LR.  Shooting the Passport at 0 Exp Comp the same patch reads 54%, 55%, 70%.  That's a big difference and may be outside of the range of what the Passport software is expecting in reading the values captured by the camera to create the profile. 
Logged
deejjjaaaa
Sr. Member
****
Offline Offline

Posts: 884



« Reply #10 on: September 21, 2012, 10:22:32 AM »
ReplyReply

People are missing Jeff's point that "the profile software will be expecting the patches to be within a normal range". 

If you alter exposure of the chart you're going to alter the RGB values of the patches.  If you ETTR by a stop or stop and a half you may throw the RGB values of the patches outside of the range that the profile creation software is expecting. 

Example, I shoot a CC Passport at +1.5 Exp Comp and the light blue patch, top row, third from the left reads 83%, 84%, 94% in LR.  Shooting the Passport at 0 Exp Comp the same patch reads 54%, 55%, 70%.  That's a big difference and may be outside of the range of what the Passport software is expecting in reading the values captured by the camera to create the profile. 

the question from the person who asked (as I understand it) is how do you exactly meter and what exactly DNG PE expects ideally (to avoid any internal compensation)... your exposure compensation, Bob, is based on what kind of metering exactly ? matrix ? center weight ? spot metering (as I suggested a post earlier) ?

PS: true, you can just simply bracket exposures no matter which method you use for metering and then just select a proper raw based on what ACR (or LR) shows w/ the most neutral settings for conversion for patches in RGB numbers in the particular color space used for such readout (in RPP, for example, for icc profile creation, you do this by checking camera's RGB numbers, those before any color transforms to any color space, for the most bright neutral patch - ACR/LR naturally do not allow to do this)

Logged
RFPhotography
Guest
« Reply #11 on: September 21, 2012, 11:38:21 AM »
ReplyReply

Well, the original question said nothing about DNG PE.  I made the assumption, perhaps incorrectly, that if the OP is using a CC Passport then the Passport software is also being used.  The original question did refer to 'the manual' so I think my assumption is solid.

There was no mention of metering pattern.  I used Matrix but it didn't matter because the target was shot taped to a solid background that was close to neutral in brightness.  If I were shooting it in a different environment I might use a different metering pattern and/or I might fill the viewfinder with the neutral grey patch.  I'd do what I needed to do to get a 'proper' exposure.  That wasn't the point of the original question, I don't believe.  The question was about whether or not the CC Passport target should be ETTR'd or not.  The answer to that, as Jeff pointed out, is no.
Logged
deejjjaaaa
Sr. Member
****
Offline Offline

Posts: 884



« Reply #12 on: September 21, 2012, 12:26:47 PM »
ReplyReply

The question was about whether or not the CC Passport target should be ETTR'd or not.  The answer to that, as Jeff pointed out, is no.
whether or not it should be ETTR'd depends on camera and metering... you might have such light and such metering that you have to put some positive exposure adjustment on camera (or change the metering)... but this adjustment certainly shall not bring the most bright patch (w/ the most reflectance) to the very edge of clipping (in raw channels)... so basically the answer shall be - you expose the target so that such and such patch(es) shall be within such and such range (of measurements in your tool, like ACR, using such and such parameters in that tool), whether as a result the shot is w/o exposure compensation on camera or with exposure compensation on camera is not important...

or for example if your shots are then done w/ exposure compensation to achieve some ETTR (for raw), then why you do not want the same for the target shot used to generate profile that will be used to convert those raw files ?...


Logged
RFPhotography
Guest
« Reply #13 on: September 21, 2012, 01:35:08 PM »
ReplyReply

No.  What you're describing is not ETTR.  What you're describing is making and adjustment for the meter reading to 'correct' for the bias of the meter in certain situations.  Two entirely different things.  If I meter a Passport sitting on a snowbank I'm still going to have to add 1.5 to 2 stops of exposure.  That is not ETTR.  That's adjusting for the bias of the metering system to generate a 'proper' exposure.

Why you don't want to ETTR the target goes back to Jeff's original point; which you seem to be missing:  The profiling software expects values for the patches to be within certain a certain range.  If the exposure puts the patch values outside that range the profile won't be accurate.
Logged
erpman
Jr. Member
**
Offline Offline

Posts: 61


« Reply #14 on: September 21, 2012, 01:58:33 PM »
ReplyReply

Quote
No.  What you're describing is not ETTR.  What you're describing is making and adjustment for the meter reading to 'correct' for the bias of the meter in certain situations.  Two entirely different things.  If I meter a Passport sitting on a snowbank I'm still going to have to add 1.5 to 2 stops of exposure.  That is not ETTR.  That's adjusting for the bias of the metering system to generate a 'proper' exposure.

Why you don't want to ETTR the target goes back to Jeff's original point; which you seem to be missing:  The profiling software expects values for the patches to be within certain a certain range.  If the exposure puts the patch values outside that range the profile won't be accurate.

This makes sense. Since the passport covers most of the range it wont really be possible to really ETTR it anyway, so we will be forced to a normal exposure. But as you point out with the snow example, adjusting for meter bias is another thing. In that case we would risk underexposing and delivering to the software a patch that was out of range the in the other direction. Spot metering the middle gray patch for the reference shot, and then adjusting for the subject according to histogram/blinking highlights afterwards seems logical in most situations.

Well it certainly makes more sense now, thanks all for the input.
Logged
deejjjaaaa
Sr. Member
****
Offline Offline

Posts: 884



« Reply #15 on: September 21, 2012, 02:05:11 PM »
ReplyReply

No.  What you're describing is not ETTR.  What you're describing is making and adjustment for the meter reading to 'correct' for the bias of the meter in certain situations.  Two entirely different things.  If I meter a Passport sitting on a snowbank I'm still going to have to add 1.5 to 2 stops of exposure.  That is not ETTR.  That's adjusting for the bias of the metering system to generate a 'proper' exposure.

that is what I am trying to get - what is proper exposure by numbers or by exact technique for metering...

Why you don't want to ETTR the target goes back to Jeff's original point; which you seem to be missing:  The profiling software expects values for the patches to be within certain a certain range. 

and what is that exactly (numbers) for DNG PE and for XRite's software ?
Logged
deejjjaaaa
Sr. Member
****
Offline Offline

Posts: 884



« Reply #16 on: September 21, 2012, 02:10:46 PM »
ReplyReply

This makes sense. Since the passport covers most of the range it wont really be possible to really ETTR it anyway

what if your metering underexposes vs what it shall be from the profile creation standpoint ?


, so we will be forced to a normal exposure. But as you point out with the snow example, adjusting for meter bias is another thing.

ETTR in local (this forum) sense is when you provide the maximum exposure yet do not have any raw channel clipping in your shot... if that is a defintion you can ETTR any target.

In that case we would risk underexposing and delivering to the software a patch that was out of range the in the other direction.

and what is the range ?

Spot metering the middle gray patch for the reference shot, and then adjusting for the subject according to histogram/blinking highlights afterwards seems logical in most situations.

too much depends on camera and light... unless you bracket (immediately or w/ shot review - does not matter) - but then the question is - when you review the raw file in ACR/LR - what are the numbers that you compare some patches with ?

Logged
RFPhotography
Guest
« Reply #17 on: September 21, 2012, 03:44:23 PM »
ReplyReply

Quote
that is what I am trying to get - what is proper exposure by numbers or by exact technique for metering...

I think you're complicating things simply for the sake of complication.  If someone is competent with a meter they will understand how meters work and be able to expose the target properly.

Quote
and what is that exactly (numbers) for DNG PE and for XRite's software ?

I don't know that that information is available for public use.  But given that XRite take care about exposing the target properly and being very careful not to clip any channels it would make sense that it's something just either side of a proper metered exposure (don't bother with the 'what metering method' bravo sierra).

Quote
what if your metering underexposes vs what it shall be from the profile creation standpoint ? ETTR in local (this forum) sense is when you provide the maximum exposure yet do not have any raw channel clipping in your shot... if that is a defintion you can ETTR any target.

Then just as with the snowbank example you're compensating for a bias in the meter.  You're not purposefully increasing exposure from a 'proper' exposure.  And it's not true that any target can be ETTR'd.  If the target contains patches that cover the spectrum from black to white then you can't ETTR the target otherwise you'll clip channels at the top end. 

Quote
too much depends on camera and light... unless you bracket (immediately or w/ shot review - does not matter)

No, too much doesn't depend on the camera and light.  The lighting under which a target should be captured is even light.  Not dappled light.  Not halfway in dark and halfway in sun.  The lighting on the target has to be consistent.  Doing this there should really be no need to bracket. 

Logged
deejjjaaaa
Sr. Member
****
Offline Offline

Posts: 884



« Reply #18 on: September 22, 2012, 10:07:04 AM »
ReplyReply

I think you're complicating things simply for the sake of complication.  If someone is competent with a meter they will understand how meters work and be able to expose the target properly.

you are trying to avoid answering the question - what is the proper exposure of the target from the software (that makes a profile) standpoint... I understand that you do not know the answer, only developers know... neither does Schewe... people like Eric Chan know... hence this is an attempt to summon such people.

Then just as with the snowbank example you're compensating for a bias in the meter.  You're not purposefully increasing exposure from a 'proper' exposure.  And it's not true that any target can be ETTR'd.  If the target contains patches that cover the spectrum from black to white then you can't ETTR the target otherwise you'll clip channels at the top end. 

that exactly depends on the definition of ETTR that you are using... if ETTR is about the end result (raw channels histogram) - then every single raw file can be ETTR'd regardless of the target  Wink ... just try to distance yourself from thinking about dialing a positive exposure compensation... you might be in a situation when (based on your camera, light, target, metering) you need actually to dial in a negative exposure compensation to get an ETTR'd raw file... so no matter what you do with camera controls, a raw file is ETTR'd when you have raw channels historgams shifted to the the point of clipping (but no clipping) in any combination of all raw channels (be it 1, 2, 3, etc channels) - hence every shot can be "ETTR'd"

Logged
RFPhotography
Guest
« Reply #19 on: September 22, 2012, 10:44:45 AM »
ReplyReply

You're playing a game of semantics. 
Logged
Pages: [1] 2 3 4 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad