LR4: Unexpected results from -5 EXP, +lots white

[dreed]

I've been playing around with Lightroom adjustments a bit and found something quite unexpected.

Using an image that has been ETTR'd (pic0), you can see what auto-tone from Lr gives (pic1).

For the sake of just messing around, I moved the exposure to -5 (and -100 highlights) and then compensated for that by moving the white balance and bringing up the shadows.. The result is a very vivid image that almost looks like it has too much saturation except that I've dialed in no saturation. Mostly, I get a result that I quite like from a scene that I didn't really expect much from.

This doesn't "work" with every image, just some but when it does, the result is quite eye grabbing (for me anyway.)

Can anyone venture a guess as to why the image responds in the way that it does? (I'd like to understand what's going on here.)

[John Hue]

Okay I'm not an expert but... : Highlights -100 / Shadows +100 seems to be a quite popular starting point for ETTR images like sunsets and contre-jour, it compresses the tonal range. I don't think there is much more to tell about what's happening here ? It just seems like your also needed to bring the exposure down since your Exposure To The Right was really on the far right of the sensor's capabilities.

[dreed]

The -100 highlights, +100 shadow are really just fine tuning- attached is what happens without them (very similar outcome to pic2 from my first post.)

The impressive outcome is the very clear separation of brightening the light colours independent of the darker ones, separating out the subject from the background.

[dreed]

I tried to achieve a similar result in C1 but I don't know where the "White point" control is hidden (or what it is called) but I'm still using C1 v6.

Is there really nobody else that has "developed" pictures in this way before?
Or anyone that can explain why the results are as they are?

[Glenn NK]

Somewhat OT -and I may be guessing a bit:

I'm starting to wonder if the reason that there is so much anti-ETTR is because of the pale washed out images on the camera LCD, then pull the image up in LR/CS, and again see a pale washed out image and conclude that the colours are faded.

This is a natural inclination, particularly with printed images - if printed images look like an ETTR image, then they WOULD be washed out and the detail is gone.

BUT - and this is a big but - provided that none of the three channels have been blown out, the ETTR image will have the most detail and therefor the most colour.

It's so easy to be fooled by an LCD (either a camera or monitor).

When the first Nikon D800's came out, many had a greenish tinge on the LCD, but the images were just fine.  I seriously think too many judge their images by the LCD - big mistake.

BTW, that shot of the castle really turned out well compared with the undeveloped image.

What you found doesn't really surprise me.

Glenn

PS - I'm quite open to rebuttals and/or correction.

[dreed]

Ok, I was hoping one of the LR gurus here could explain why the image turns out the way it does but alas nothing Does that mean that it is unexpected for others too?

In further playing, the interesting combination is how low the exposure can go vs how high the white point is (as if they were two opposite buckets on scales that need to be balanced.)

[jpegman]

Using digital photography optimally implies that one want the most data for the processing enginee (ACR) to have inputted. This means pushing the brightest areas of the scene as far to the right as possible without burnout, and letting the processing engine "pull the exposure back". Pushing it to the maximum without overexposing can be a bit tricky as the Histogram on the camera LCD shows the "blinkies" well before the image is actually overexposed which is what Dreed found out.

In reality, if one could push the exposure of the highest highlight to the edge of overexposure (98-99%) there is a lot more information in the shadows (highlights were maxed out in exposure) that the processing software can work on which results in more detail in deep shadows and even can show a sharpening improvement as well as noise reduction over a less optimum exposure.

There really is no buckets to balance - you expose the highlight bucket as high as you can and your cameras dynamic range will place the shadow bucket as high as it can.

Because of this exposure to the edge, your jpeg LCD image (and of course your ingestion preview in LR since it is also based on the camera generated jpg) will look pale and washed out. Just know that when you reduce the exposure in LR's develop module you will have a better, more detailed, lower noise image.

It works, try it.  Even in all dark subjects like the proverbial "black cat in a coal bin" pushing the brightest area to  the right (i.e. maybe overexpose rather than the traditonally thinking under-exposure of a dark subject  will give a "thin" washed out image BUT ALL THE DATA your camera can deliver has been preserved for post processing.

[Tim Lookingbill]

There is a human perception issue that may be the cause of why you think Pic2 looks so eye grabbing. It's called adaptation of the human optical system.

Your eyes adapt to the light, faded appearance of the ETTR version which is the first thing that appears when opened in LR. You have no other reference to compare to what it should look like, but your emotional response, separate from optical stimuli, has created a sort of pre-visualization in the brain based on the emotional response of similar shots seen in the past.

You first hit Auto Adjust and it improves. You still have no visual reference to compare against on how to make it better, so you apply a severe Exposure slider adjust that makes the mid-range textured foliage go dark, acting as a surround to frame and accentuate the contrast of the lighter, sharper and flatter geometric shapes of the building.

Only now from what I'm seeing the image looks too dark where all that tree shadow detail is plunged in black which takes away the benefits of shooting ETTR by throwing away all that data it was suppose to capture.

I prefer the look of Pic1 because it's the most balanced with regard to showing all the detail in the image.

[jpegman]

I think the issue with Pic1 and Pic2 is that the initial exposure on Pic0 was pushed way beyond 255 on all 3 channels over a large part of the image to the extent that a -5ER was necessary to try and bring back this drastically overexposed image. 

If Dreed really just pushed the exposure to the 99% edge (max ETTR), he probably would have only required more like a -2 to -3ER to bring the image back to "normal" which would have opened up the shadows when the mid range tones were brought back into control.

The tough part is to get the ETTR just to the right edge without going over in the field and then the rest of the image will be easily processed and controlled with ACR.

The idea and theory is sound, but, in digital land one must always respect how far one pushes ETTR without blowing out elements one wants to show some semblance of data (analogous to Adams Zone 9 in the film days).

Push the exposure to just kiss the max ETTR from the area one wants to have a touch of detail and the shadows will fall in with as much detail as your sensor is capable of delivering.

[Schewe]

Can anyone venture a guess as to why the image responds in the way that it does? (I'd like to understand what's going on here.)

The reason that the image responded the way it did is because of the image adaptive control set of the recent ACR/LR controls. Nothing about what you posted is surprising nor unexpected...doing ETTR and then severely reducing exposure is doing exactly what I would expect–redistributing levels from the top of the histo to lower levels. This is as expected...

Note however, from your posting, the LR "Auto" looks a lot better than your -5 exposure. Not sure what you think you got from the 3rd posted image as it look way too dark...

The whole thing about the new controls (PV 2013) is that the tone controls offer a LOT of variability and you can tone the image in many different ways...ETTR is obviously an optimal method of capturing (as long as you don't clip). What you do with it is up to you.

If I were t adjust the original image, I would be pretty close to what LR's Auto gave you (with maybe a few tweaks). But all you've posted are screenshots so I won't bother to tweak them...

[dreed]

Thanks all for the replies, especially those from Tim - I didn't for a second imagine that my reaction was quite so ... predicated by the inputs.

If anyone would like to play, I've attached a 1024x768 16bit TIFF of the raw (pic0) using ProPhotoRGB (downsized to fit 4MB)

[Tim Lookingbill]

Wished there was a way to get consistent foliage detail in similar captures.(see attached)

Doesn't matter what version of ACR/LR I'm using. Clarity slider isn't this refined so I have to resort to notched point curve tweaks to bring out flat looking foliage shadow detail.

Wonder if it has something to do with sensor response and/or lens flare.

[davidedric]

This could be a time when the Nik plug ins give the local control you are looking for?

[dreed]

Regarding detail, it was dark and raining when I took the shot.