Studio portrait ETTR with proper jpg display (darkened)

[PhotoCat]

Shot portraits with raw with ETTR. It is understood that the jpeg preview will be overexposed. Is it possible
to darken the jpg and jpg preview with ML automatically? e.g. if 1 stop over-exposed in raw, then apply 1 stop darkening
to the preview jpg. (this darkening level can be adjusted manually too, if need be, as the photographer knows how much overexpo there is in studio
with the incident light meter)
This will keep the model happy during the shoot and at the same time details in the dark clothing can be preserved too.
Moreover, camera jpg can be used directly without raw processing off camera. Thanks!

[Audionut]

Look at the image review settings in the pref menu.

You can adjust exposure.

[PhotoCat]

Tks Audionut. Using your method, it is indeed darkening the image but the color is not quite right.
(seems like the darkened preview has been desaturated)
Just wondering what darkening algorithm is used?
Would be nice to produce the color in LCD preview as if the raw file went into DPP and dial down in exposure with picture style applied.

BTW, I realize there is a different overexpo limit for each camera such that blown highlight can be recovered.
My 5D2 at ISO100 is about 1.3 stops. 

BTW, is it possible to add an exposure lock on the LCD image darkening?  e.g. always dial down the exposure by 1 stop
for all playback etc. This way, one can shoot and preview studio pics as usual - all color is preserved during LCD preview
and jpg generation and at the same time performing ETTR.

I guess may we call this "transparent ETTR" ?  You turn on "transparent ETTR" but you shoot and preview as usual except you
need a little more exposure, knowing that shadow noise will be improved.

Thanks a lot!

[PhotoCat]

Added benefit is that in-camera generated jpg will have correct exposure and all the histogram and false color
work as usual. No need to do further raw processing.

[a1ex]

A while ago I tried to darken the preview jpeg without success - the negative gain got burned in the CR2 as well.

The darkening curve is:

Code Select Expand


a = exposure - 1 (from linear units)
y = (1 - 1/(1+a*x)) / (1 - 1/(1+a)), x from 0 to 1


[PhotoCat]

Tks for the insight A1ex.  Looks like you have started from the raw file too to perform the darkening.
Not sure how DPP does it then (within limits).  Tks agn!

[a1ex]

Nope, darkening is done from jpeg (it's just an approximation).

[PhotoCat]

oic. Is it possible to start from raw file in camera then like DPP with a negative exposure bias and picture style applied? tks!

[Audionut]

BTW, I realize there is a different overexpo limit for each camera such that blown highlight can be recovered.
My 5D2 at ISO100 is about 1.3 stops. 

You have this a little wrong.  The Canon meter exposes for a midtone, and the difference between midtones and overexposure is greater then 1.3 stops.

Yes it is different for different cameras.

Point the camera at a neutral subject with very little dynamic range, say an evenly lit white wall, meter for 0 on the camera meter.  That is metered for a midtone, now increase exposure until clipping.  You should get close to 3.5 stops before overexposure. 

As scene dynamic range increases, you have less headroom above the metered midtone to increase exposure, sine this headroom is being used by the dynamic range of the scene.

Edit:  Also the headroom above the metered midtone stays consistent with varying ISO.  It's the shadows that get clipped.

http://www.rawdigger.com/houtouse/lightmeter-calibration

And re-read a1ex' post.  Attempts to change the jpg preview got burned into the raw file also.  That is not good. 

[PhotoCat]

Point the camera at a neutral subject with very little dynamic range, say an evenly lit white wall, meter for 0 on the camera meter.  That is metered for a midtone, now increase exposure until clipping.  You should get close to 3.5 stops before overexposure.

I think we are both correct Audionut

I use spot metering to point at the white wall and set the meter to +2. So beyond that I have tested the 5D2 to
be able to recover highlight for another 1.3 stops.  2+1.3=3.3 stops (very close to your 3.5 stop depending upon subjective opinions)

With Magic Lantern, for portrait shooting at least, I find that I can get very accurate in-camera jpgs which significantly reduce post-processing time
by using the following metering technique:

1) False color to ensure face expose
2) RGB spot meter to ensure correct skin tone

Although I shoot raw+jpg, I believe in getting it right inside the camera to reduce post-processing time. (raw is for insurance only)

So if ML can get a raw-based jpg file and jpg preview when shooting ETTR, further raw processing in computer will not be needed, which not only saves time,
but it also makes the model happy when viewing the camera LCD as we shoot    

In summary, I think a raw-based in-camera jpg preview/jpg generation will very much facilitate the use of ETTR technique.  Tks!

[Audionut]

Fair enough.  However, +2 is an arbitrary number that only really holds relevance to your exposure decisions, and it helps to be accurate and specific around these woods, especially if you want to feature request or bug report with a developer. 

a1ex is working on a pretty tricky preview for dual ISO, and I imagine that if he can code it to the point where he is happy with it, something like you describe might also be possible.  It is somewhat of a dupe request, as most people want to brighten their autoETTR shots, as they use it to expose for somewhat extreme highlights.

[PhotoCat]

+2 in spot metering is not arbitrary. It means highlight close to white but jpg not blown out yet. 
Thanks for the heads up and all the hard work!