Optimize In Camera Jpeg White Balance RGB Histogram For Better ETTR.

[Tim Lookingbill]

This may be old news to some folks here but I just found a way to make the jpeg histogram more useful for ETTR. I tested this out just now just to make sure I could affect my Pentax K200D's individual RGB jpeg histogram display rendered according to white balance setting and exposure.

Just discovered my used 2008 Pentax camera offers an adjustable Custom White balance to make it green, blue, purple, amber etc.

So I set the Custom WB at neutral 5000K/+6 Tint as read in ACR's As Shot and took a shot of my LED display showing a light gray field. As would be expected the incamera individual RGB histograms are all the same shape where the non-clipped highlights line up perfectly between each other. Took another shot of my LED display opening one third stop which had all three highlights slammed and spiked to the right.

Adjusted the Custom WB by moving the color temp color graph map to ALL GREEN and took the shot again using the same exposure that clipped the highlights with the previous neutral WB exposure. Guess what! The RGB histogram highlights weren't slammed and spiked to right. In fact they were nice ant hill shapes with no spikes.

Thought I'ld share this with those who have cameras with an adjustable Custom White balance and individual RGB histograms that reflect the affects of changing exposure and white balance.

Has anyone tried this?

This also could work by setting a custom WB using a color target that would make the green channel even more biased and then go back into the WB color adjust graph map and make it even more green which would give more predictable exposure latitude for ETTR. ACR As Shot WB does show the effects of distorting the in camera WB this way.

I did a search on this subject to make sure this wasn't already covered and came up with nothing.

[BAB]

I think you would need RAW histogram by channel view to achieve proper ETTR? Then the camera is not cooking the file output. Your test may only be correct for a given situation of EV.

[digitaldog]

I think you would need RAW histogram by channel view to achieve proper ETTR? Then the camera is not cooking the file output. Your test may only be correct for a given situation of EV.

You are correct about a raw Histogram. No need for JPEG Histogram hacks either. At least for photographers who had ever shot transparency film without a Histogram anywhere near the camera. Far from difficult and photography 101 (optimal exposure). Get something like RawDigger to run exposure tests to actually understand how your sensors optimally expose the raw data, understand how light meters actually operate, ignore the LCD lies about data you are NOT capturing. ETTR is an old and poor term that needs to go away. The correct term is correct exposure; something most photographers have achieved for well over 100 years.
Has anyone tried this? 

[DP]

I did a search on this subject to make sure this wasn't already covered and came up with nothing.

search for UniWB... it was covered miriad times

[TonyW]

This may be old news to some folks here but I just found a way to make the jpeg histogram more useful for ETTR. I tested this out just now just to make sure I could affect my Pentax K200D's individual RGB jpeg histogram display rendered according to white balance setting and exposure.

Just discovered my used 2008 Pentax camera offers an adjustable Custom White balance to make it green, blue, purple, amber etc.

So I set the Custom WB at neutral 5000K/+6 Tint as read in ACR's As Shot and took a shot of my LED display showing a light gray field. As would be expected the incamera individual RGB histograms are all the same shape where the non-clipped highlights line up perfectly between each other. Took another shot of my LED display opening one third stop which had all three highlights slammed and spiked to the right.

Adjusted the Custom WB by moving the color temp color graph map to ALL GREEN and took the shot again using the same exposure that clipped the highlights with the previous neutral WB exposure. Guess what! The RGB histogram highlights weren't slammed and spiked to right. In fact they were nice ant hill shapes with no spikes.

Thought I'ld share this with those who have cameras with an adjustable Custom White balance and individual RGB histograms that reflect the affects of changing exposure and white balance.

Has anyone tried this?

This also could work by setting a custom WB using a color target that would make the green channel even more biased and then go back into the WB color adjust graph map and make it even more green which would give more predictable exposure latitude for ETTR. ACR As Shot WB does show the effects of distorting the in camera WB this way.

I did a search on this subject to make sure this wasn't already covered and came up with nothing.

Guillermo Lujik who does post here occasionally wrote (and devised?) UniWB.  The thrust of your thread seems to be similar?
http://www.guillermoluijk.com/tutorial/uniwb/index_en.htm

I believe the main problem with trying to achieve ETTR in camera for JPEG is apart from the fact that a "correctly" exposed raw will likely be an "incorrectly" exposed JPEG is that the camera metering system may let you down - unless you are able to actually meter a highlight ROI that you must hold detail and then set exposure accordingly. 

Cannot speak for your Pentax but... Auto metering may be problematical in as much as you really do not know exactly what areas are being measured and what 'clever' algorithms being applied by the manufacturer (Nikon Matrix).  Centre-weighted is also an issue as it is not always evident how much weighting given to the overall measuring area.  Spot metering should be best but again need to know the exact area of measurement i.e. size and position in relation to what you see through the viewfinder or in LV.  In addition, you need to establish a meter calibration point to find out how much you need to allow for exposure correction once you have measured your most important highlight ROI e.g. +2, +2.5, +3 EV's to ETTR

Other than that relying on the camera screen LCD and RGB and composite histograms will give you an 'acceptable' capture quite/very often, and if you have the time to test exposures (not the decisive moment of course because you will miss it  ) by manual or auto bracketing then you may achieve a degree of satisfaction for ETTR JPEG by seeing the histogram stacked to the right. 

[Tim Lookingbill]

search for UniWB... it was covered miriad times

I've heard of UniWB several years back in this forum. I'm not trying to win any converts for true ETTR with this post.

I'm simply relaying a procedure I stumbled upon after toying around with Custom WB that allows adjusting hue to force the jpeg green channel forward ahead of the red and blue. I find it to be a good enough demonstration of how the hue of WB can affect exposure parameters close to full saturation when shooting Raw for cameras that don't have an in camera Raw histogram.

And to be clear the magenta tinted test shot that severely clipped the histogram on all channels was recoverable but there were linearity issues where after I white balanced in ACR showed slightly darker tones around 200RGB were yellowish and those lower were neutral. This is in PV2010 reducing the Exposure slider where doing this with the Brightness slider compressed the detail that forced the yellow to stay neutral at the sacrifice of detail.

That's the bit of detail left out of ETTR discussions is what exposing this close to saturation does to linearity according to the PV version and what highlight recovery slider one uses. PV2012 just moved this preserve linearity without compression feature over to another slider.

I might have over exposed the magenta tinted test image 1/3 more stop where it would extremely clip in the jpeg histogram but then recovery in ACR or LR might just compress the highlight detail which doesn't look good shooting brightly lit white bilious clouds.

Mine is an old camera with less dynamic range capabilities so newer cameras would most likely allow more recoverable detail with exposures closer to saturation.

But what is the quality of what is recovered. Preserve highlight color detail and linearity or accept compression to be the only solution.

[digitaldog]

WB has no effect on raw data let alone exposure. Only the amount of light striking the sensor: shutter and aperture.

[DP]

Guillermo Lujik who does post here occasionally wrote (and devised?) UniWB. 

UniWB predates Guillermo Lujik (but he contributed to popularization)

[Tim Lookingbill]

Something else I discovered shooting wide dynamic range scenes that force an exposure of  1/250/s, f/16, ISO 200 on a sunny day to preserve the highlights of houses where I got a lot of noise in the shadow areas of houses that wound up being more midrange instead of deep shadows meaning there was more bounced fill light in the sides of houses not getting direct sunlight on such a bright sunny day.

But around close to dusk and in the shade I exposed at 1/80's, f/8, ISO 800 and there's NO NOISE in the shadows of a macro shot of some old dead leaf laying on green clover. Explain that!

My Pentax came with a CCD sensor (not CMOS) which doesn't do so well with heat and the ISO 800 shot at dusk was a bit cooler around that time.

[digitaldog]

The true way to know if highlights are 'preserved' due to exposure is to examine the raw Histogram.
The true way to know if there is no noise is first to bracket exposure and examine the noise instead of assuming none exists in any.
Explained but accepted? 
I'm not sure why a few (amateur/hobbyists) have to make something as fundamental as idealized exposure for any media (neg, transparency, raw, JPEG) more difficult than it has to be.
Next we'll hear how difficult and complicated it is to focus (prior auto focus lens). 

[DP]

1/250/s, f/16, ISO 200
1/80's, f/8, ISO 800

1/80@f8 is a lot more exposure (incoming light) than 1/250 @ f16 - so for deep shadows in both shots it might not be a difference that 1/250 & f16 was a sunny day and 1/80 & f8 was not... you might preserve the details in the important highlights on a sunny day, but severely underexposed your shadows vs that shot during the dusk... spot meter shadows next time to see if they are exposed equally between two scenarios.

plus a good chance that analog gain in off sensor ADC in your Pentax @ ISO800 is more beneficial than @ ISO200 -> amplifying analog signal before ADC up to a certain limit is good - result looks like lesser readout related noise

ambient temperature might be a factor too

Pentax K100D probably has the same 6mp CCD & off sensor ADC = http://www.photonstophotos.net/Charts/PDR_Shadow.htm#Pentax%20K100D ... see that going above nominal ISO200 is beneficial for deep shadows (if you can't saturate the sensor with exposure for whatever reason)

[Tim Lookingbill]

1/80@f8 is a lot more exposure (incoming light) than 1/250 @ f16 - so for deep shadows in both shots it might not be a difference that 1/250 & f16 was a sunny day and 1/80 & f8 was not... you might preserve the details in the important highlights on a sunny day, but severely underexposed your shadows vs that shot during the dusk... spot meter shadows next time to see if they are exposed equally between two scenarios.

plus a good chance that analog gain in off sensor ADC in your Pentax @ ISO800 is more beneficial than @ ISO200 -> amplifying analog signal before ADC up to a certain limit is good - result looks like lesser readout related noise

ambient temperature might be a factor too

Pentax K100D probably has the same 6mp CCD & off sensor ADC = http://www.photonstophotos.net/Charts/PDR_Shadow.htm#Pentax%20K100D ... see that going above nominal ISO200 is beneficial for deep shadows (if you can't saturate the sensor with exposure for whatever reason)

Some very convincing points I hadn't considered.

Most of my shots at ISO 800 regardless of the exposure always gave me lots of noise.

The macro shot at dusk was not only dim lighting but the majority of the scene was of dark objects like the deep green clover which usually requires I set exposure to let in more light, but even then I should've gotten much more noise. ACR defaults has the bulk of the data occupying the left 1/3 of the histogram meaning I underexposed a somewhat dark scene.

But I think I'm going to go with heat as the cause because I just remembered shooting a 5 sec. long exposure to capture a snow scene at night at my apt. lit only by a couple of porch lights from nearby tenants. I set it at ISO 200, the lowest with the K100D and there was very little noise in the darkest shadows but it was very cold out. ETTR worked the best shooting this way with a tripod which I used.

[digitaldog]

And my ISO 800 images show LESS noise than ISO 100 in this case because ISO doesn't have anything to do with exposure; the amount of light striking the sensor. It does have something to do with dumb meters and not too intelligent photographers who don't understand how to expose:

[smahn]

And my ISO 800 images show LESS noise than ISO 100 in this case because ISO doesn't have anything to do with exposure; the amount of light striking the sensor. It does have something to do with dumb meters and not too intelligent photographers who don't understand how to expose:

Since it's the same amount of light in both cases, why do they differ?

[Frans Waterlander]

And my ISO 800 images show LESS noise than ISO 100 in this case because ISO doesn't have anything to do with exposure; the amount of light striking the sensor. It does have something to do with dumb meters and not too intelligent photographers who don't understand how to expose:

There you go again, ad hominem attack. Maybe you should explain the reasons why your 800 ISO shot has less noise, instead of attacking those you don't agree with.

[faberryman]

Why do people make something which is simple and straightforward like exposure and make it unnecessarily complicated?

[digitaldog]

There you go again, ad hominem attack.

Nope, simply facts backed up with an example.

Maybe you should explain the reasons why your 800 ISO shot has less noise, instead of attacking those you don't agree with.

Explain the obvious. To you again?

[digitaldog]

Since it's the same amount of light in both cases, why do they differ?

Best explained in this post from way back in 2011 and the diagram:
https://forum.luminous-landscape.com/index.php?topic=56906.msg466484#msg466484

[stamper]

Why do people make something which is simple and straightforward like exposure and make it unnecessarily complicated?

Exposure simple?

[Garnick]

There you go again, ad hominem attack. Maybe you should explain the reasons why your 800 ISO shot has less noise, instead of attacking those you don't agree with.

Think back to film days.  If the subject seemed to call for more light hitting the film we had three alternatives.  When possible and feasible, we could perhaps literally add more light, as in a studio situation or something of the sort, where adding light and balancing the colour with filters was one alternative perhaps.  Of course the next approach would a combination of aperture and or shutter speed adjustments. The final approach was to use a higher ISO film, manufactured to be more sensitive to light.  The way I understand it, sensitivity to light is the most obvious reason for increasing the ISO, whether for film or for a digital camera sensor.  Of course the downside of using a higher ISO film was the grain factor, which was eventually overcome somewhat in some of the colour and B&W emulsions.  In the digital photography world, what we used to call "grain" is now referred to as "noise", or the S/N Ratio.  Now after all of this rambling I believe the phrase I used previously (sensitive to light) tells the tale as far as the apparent reduction of "noise" as we increase the ISO setting.  Even though the intensity of the light source does not change and nor does the shutter speed or aperture, as the ISO is increased the sensor's sensitivity to light increases, thus the reduction of noise, especially in the shadows and other darker areas of the image.  Again, as you at least approach the "Optimum Exposure Level" you will start to detect less noise in those areas normally most affected by a lack of exposure. 

I imagine Andrew might jump in and perhaps deflate my theory to some extent at least.  However, it's the way I see the situation and understand what's actually happening when I raise the ISO setting.  It does not increase the amount of light striking the sensor.  The sensor is then set to gobble up more of it, and for the most part that's a good thing.  I'm also certain that Andrew would offer a much more understandable way of putting it, which I look forward to.

Gary