ETTR - Lots of room for push in post

So I have been exposing to the right 1 or 2 thirds and generally things have been fine. I altered my PP technique a little recently and my first step (in LR3) is, with all the sliders at 0, find how much I can pull up exposure before clipping.....

I am finding I can push exposure between .75 and 1.25 in most cases before highlights clip even after ettr!

is this normal? should I try exposing a full stop to the right? thoughts?

My understanding of ETTR/HAMSTTR is that you expose to the right as far as possible until the highlights are blown or further if this is not an issue.

If you are serious about ETTR, set your camera Picture Style to Neutral and your Contrast and Saturation all the way back to -4. Then, you can set your Exposure to the right and not get any "false" clipping alerts in your camera. For example, you can set a bright white to as much as +3 EV without clipping, although many of us are "a bit" more catious, tending toward say +2 2/3 EV. This, though, will tend to show clipping if your camera Picture Style settings are at the defaults.

In your Raw processor, when things are "To The Right", some hightlights will be right at the point of clipping, in fact, in your default preview should show soe clipping, but you can pull the exposure back to "tame" the highlights and show detail. But the point would be that your shadow areas will have more exposure and therefore less noise. That is the point of ETTR/HAMSTTR -- to maximize the exposure even if the scene comes out "bright", because we can "tame" it in the Raw processor with no harm to the image as long as needed highlights are not actually clipped.

I've also have read in many places that 5DMkII raw files do NOT like to be "pushed" up a lot, the shadows will have noise and artifacts.
Whereas, the Leica M9 is the opposite, it's files doesn't like to be "pulled" back.

buggz wrote in post #13332619
I've also have read in many places that 5DMkII raw files do NOT like to be "pushed" up a lot, the shadows will have noise and artifacts.
Whereas, the Leica M9 is the opposite, it's files doesn't like to be "pulled" back.

RAW files do not have any personal tastes about being pushed or pulled. The only thing that may happen when exposure is increased is noise becoming more apparent (SNR remains the same though). Those cameras with higher DR will allow stronger exposure corrections up before noise appears, and I doubt M9's sensor has more DR than 5D2's.

Regarding exposure being pulled back, it's nonsense to make any distinction. Any RAW file can have its exposure reduced and will behave in the same way. After all it's a linear scaling down.

Any other distinction beyond considerations on the amount of noise is like when someone tries to classify the quality of digital noise produced by a digital sensor ("sensor A is more film-looking, sensor B is more chromatic noise, sensor C is more pleasant to look at"): pure myth.

Regards

buggz wrote in post #13332619
I've also have read in many places that 5DMkII raw files do NOT like to be "pushed" up a lot, the shadows will have noise and artifacts.
Whereas, the Leica M9 is the opposite, it's files doesn't like to be "pulled" back.

Just to add in, any digital camera will "show" more noise if you push the shadows a lot. It's nothing unique to the 5D2. And that goes for lower ISOs as well as higher, although the high ISOs have been "pushed" by the camera and so tend to show more noise right off the bat.

If you want to see this "in action", you can do a simple test with your camera:

Set the camera in Raw format, Manual Exposure, and turn off any in-camera processing (like HTP, Auto Tone, Noise Reduction). You could also set your Picture Style to Neutral if you are using DPP (and turn your Contrast all the way down).

Then, frame a scene and set the camera to ISO 1600 and set your exposure to a good exposure with no "clipped" shadows or highlights. For this exercise it doesn't have to be an "ETTR" exposure but the scene should be well exposed at ISO 1600.

Take the shot!

Now, we've all heard that a high ISO "causes" noise and that a low ISO pic will be "cleaner", so we'll put that to the test!

Set your ISO to 100, and don't change anything else. In other words, you will still get the same amount of light collected by the sensor. The only difference in your exposure/brightness will be the in-camera amplification of the signal -- the ISO amplification.

Take the shot!

Of course, the ISO 100 shot will be "horribly underexposed by four stops! Well, but we expect it to be "cleaner", right?

So the acid test: load both shots into your Raw converter and "compensate" for the low ISO 100 shot by boosting the Exposure by four stops, giving it the equivalent brightness of the ISO 1600 shot, and compare!

What exactly you see will vary with model/sensor technology, so rather than me predicting a particular output, I would expect you to see some noise in the ISO 1600 shot but not more than the ISO 100 shot!

Now if you do in fact see less noise in your "pushed" ISO 100 shot, please let us know, and post 100% crops of the two!

WhooHoo!

tonylong wrote in post #13331890
If you are serious about ETTR, set your camera Picture Style to Neutral and your Contrast and Saturation all the way back to -4. Then, you can set your Exposure to the right and not get any "false" clipping alerts in your camera. For example, you can set a bright white to as much as +3 EV without clipping, although many of us are "a bit" more catious, tending toward say +2 2/3 EV. This, though, will tend to show clipping if your camera Picture Style settings are at the defaults.

In your Raw processor, when things are "To The Right", some hightlights will be right at the point of clipping, in fact, in your default preview should show soe clipping, but you can pull the exposure back to "tame" the highlights and show detail. But the point would be that your shadow areas will have more exposure and therefore less noise. That is the point of ETTR/HAMSTTR -- to maximize the exposure even if the scene comes out "bright", because we can "tame" it in the Raw processor with no harm to the image as long as needed highlights are not actually clipped.

Well I already knew "the point" of ETTR but I was under the impression people were using less than 1, rather than 2 or more! I will have to get much more aggressive! Thanks for this tip.

You guys have got me totally confused...if you have properly ETTR in camera, why would you then want/need to boost exposure in LR? If you can boost exposure in LR, you have not ETTR in camera.
Boosting it in LR does not increase the amount of available information, information is only acquired in camera, it does, however, increase image noise.

chauncey wrote in post #13336559
You guys have got me totally confused...if you have properly ETTR in camera, why would you then want/need to boost exposure in LR? If you can boost exposure in LR, you have not ETTR in camera.
Boosting it in LR does not increase the amount of available information, information is only acquired in camera, it does, however, increase image noise.

Which is exactly what Tony, GUI and cagenuts said; if full ETTR has been done it would be impossible to push development ("exposure") without clipping. However, if the tonal range of the scene is great, the image might benefit if that range is compressed by "pushing" the shadows with Fill Light - or the cost in the increase in visible noise might be greater than the benefit.

Higgs Boson wrote in post #13336444
Well I already knew "the point" of ETTR but I was under the impression people were using less than 1, rather than 2 or more! I will have to get much more aggressive! Thanks for this tip.

ETTR is placing the highlight exposure in the RAW capture as close to clipping as is possible without actually hitting the clipping wall. How much the increase is above the conventional exposure to middle grey can vary greatly - there is no simple rule. If the scene has a wide range of tones, like a black cat sitting next to a white cat, the auto exposure may already place the highlights at the extreme right and no further ETTR would be possible. OTOH, if the subject is a grey cat sitting on the grass, the ETTR could require as much as a three stop increase over the meter reading.

In reality however, most practitioners of ETTR do less than the full version because even if the Picture Style is Neutral and the processing parameters are set to the minimum, the histogram is still misleading because it is the histogram of a white balanced image. WB does far more to inflate the red and blue channels than any other normal processing. So they will increase exposure until they seem to see clipping and stop, although they could have gone even further.
Nevertheless, it is also important to note something that GUI and others have pointed out: The only real benefit of ETTR is the reduction of noise or at least the minimization of shadow noise. The practice of ETTR dates from 2003 and cameras have improved greatly since then. With current models in situations of adequate light, low or even intermediate ISOs and output requirements that don't need to be museum quality, the noise presence is so low that the value added by ETTR is much less today and may not be worth the bother. The decision is yours.

Just as a point of interest:

tzalman wrote in post #13337006
ETTR is placing the highlight exposure in the RAW capture as close to clipping as is possible without actually hitting the clipping wall.

From this thread: https://photography-on-the.net …/showthread.php​?t=1063694

PhotosGuy wrote in post #12713006
Hopefully this won't complicate things, but that is what I call "Shoot to the Right" & explain an easy (for me) method to assign exposure here:
Need an exposure crutch?

I've been under the impression that "Expose to the right" originally was something entirely different in concept.
Think about a dark blue flower against green grass without any sky in the image. You push the RAW exposure up & then dial it back to what would be normal in PP. So highlights might not come into the calculation unless you have some in the image & they're important to it.

I wish I could find the link where that was first posted, but it's something to think about.

The only issue I have with the technique is having to reevaluate the histogram if the scene/location changes.

Higgs Boson wrote in post #13337559
The only issue I have with the technique is having to reevaluate the histogram if the scene/location changes.

It's kinda the point though because highlights can change as the scene changes. Then again I could be misinterpreting you comment.

Higgs Boson wrote in post #13337559
The only issue I have with the technique is having to reevaluate the histogram if the scene/location changes.

In the future we may have auto ETTR. Now that DSLRs have Live View and the real time histogram it would not be that hard to implement. In the same way that the data flow from the sensor is read to produce the histogram, the brightest pixels can be identified and exposure adjusted to bring their output value to just short of 255.

tzalman wrote in post #13337707
In the future we may have auto ETTR. Now that DSLRs have Live View and the real time histogram it would not be that hard to implement. In the same way that the data flow from the sensor is read to produce the histogram, the brightest pixels can be identified and exposure adjusted to bring their output value to just short of 255.

But that would take all the fun out of it:p