I had been discussing the idea of highlight "limits" in digital photography with a member of our Canon photography forum Photography-On-The.Net (POTN). I was explaining to him how "bright" you could set an exposure in the camera without "technically" clipping those hightlights. I explained that if you are shooting in the Raw format, which retains a full amount of image "data", you can raise your exposure at least a "stop" (twice the amount of light) higher than you could if you were shooting a jpeg.

Well, I had tested this myself previously, but I decided to do a more "structured" test for proving the concept, posting here, and sharing with that forum.

So, for this test, I used my six-year-old Canon 5D "Classic". I chose this because I suspected that it might be more "sensitive" to highlights in its metering (giving maybe misleading warnings) although I am fully confident in its Raw file quality.

So, I opened a "New" document in Photoshop to give me a blank white "subject".

My "method" was to meter off of the white document and then shift a bit to photograph the entire computer screen so that various details and tones would be part of the picture. I brought the images into Lightroom, and then took a screen shot and opened that in Photoshop, where I trimmed it a bit down to the image and the Lghtroom Histogram and Basic Develop sliders, and the results are what you see here. Note that the line below the histogram includes R, G and B "percentage" values with my cursor in the "white space", showing how Lightroom is interpreting the tones as currently "rendered"

Also, the screen shot lets you see the Lightroom display of shooting info, which you can use to follow step-by-step!

Lastly, for my final two screen shots, I triggered Lightroom to show highlight "clipping" alerts for the final shot, both before and after my "tweaking" -- more on that later.

My first shot would be a "regular" exposure, as if the camera was in an "Auto" mode, so I filled my metering "circle" with white and set the exposure so that the meter "needle" was in the center of the meter "scale", which is what the camera would do in an auto or semi-auto mode, and I took the shot -- the first shot below. As you can see, there is "radio frequency interference", but that actually comes in handy later!

I'll include notes and observations for the different images.

Then, for the second shot I cut the shutter speed in half, raising the exposure by one "stop". Note the image is brighter, but the white is not "white".

For my third shot I went to the "normal" "safe" setting for white, what you would do if you were shooting a jpeg. This was done by cutting the shutter speed again in half, giving white a "2 stop" lift from the medium exposure -- four times as much light. AS you can see, the result is quite bright, and most people would stop there for all practical purposes.

But, the purpose of this test was to check the idea that the Raw file can retain detail when you push the exposure by one more stop, to "+3 EV", or eight times the amount of light for a "medium" exposure!

So, for the fourth shot I did just that -- I again halved the shutter speed and took the shot! As you will see in the fourth image, wow, way "overexposed"! In fact, my old 5D was "screaming" with highlight alerts, even though I had "tuned" it to be not-so-sensitive!

And, in Lightroom, I took a screen shot of Lightroom showing highlight alerts of the out-of-camera shot, so the next image shows that, and yeah, blazingly bad highlight clipping (white areas).

So then, the test. I pulled the Exposure back, less than 2 stops, and applied a bit of Highlight recovery and "lowering" and a bit of contrast, and so the next image is the result of that. Amazingly, not only do all the details of the desktop become clear, but in the white area, to show that in the Raw data details still are retained, you can even see the RF Interference, even though my adjustments were too "mild" to make them as intrusive!

And, in the next image, again telling Lightroom to show highlight alerts, there were none! In Lightroom, if you have "real" clipping, there would still be alerts even after pulling your exposure all the way back!

For those who don't understand that last part, in other words if you suspect that pulling back the exposure of a blown highlight would just turn it "gray" with no highlight alerts, I took one more shot! I bumped the exposure by one more stop, so +4 EV. So, the next image is the out-of-camera Lightroom default "rendering" -- a totally messed up overexposure!

Then, I totally cranked on that exposure in Lightroom. In the next image, you can see I pulled the Exposure back to the max, -4 stops, and then pulled up the Recovery slider and pulled back the Highlights slider, all to bring as much of that white as possible out of clipping. As you can see, "some" of the image got a little help:)! Still terrible though!

But then, the question: did the additional stop result in "real clipping"? Or, does the remainging white still have "something" that Lightroom can "work with?"

Well, one "check" is that RF Interference. In the last shot, the recovery actually brought out the interference! In this shot, none.

And then, note that the RGB values at the cursor are "99.9%, 99.9%, 99.9%". Now, don't ask me why they aren't 100,100,100 like in the original shot, but for all intents and purposes...

And then, in case any further proof was needed, I went ahead and activated the Lightroom highlight display and, as you can see, there is a lot of the "white space", showing clipping even after turning the Exposure and other sliders all the way.

So there you are, a "proof of concept" test, and it does work!

For practical purposes, I don't know how often I'd push things that far, but...think of a wedding, with a white bridal gown, and next to the bride is a groom, with a black tux. The "normal" approach could be to set the gown to +2EV and "hope" that you could lighten up the tux a bit at your computer. But, it's nice to know that we have with Raw a FULL STOP to go before totally destroying that white bridal gown!

'Nuff said!