Chris Jablonski's picture

Expose to the Right!

No, not a guide for exhibitionists!

For those wondering, here is an example of how "exposing to the right" (i.e. of the histogram) can give you plenty of leeway in post, and give a very different end result than what you see on the back of the camera, which can look flat and uninspiring.

Actually this image has plenty of leeway, so did not really need this as much as some, but illustrates this "overexposure" technique vividly. It can make all the difference when the scene lacks contrast. If you want to increase it in an image taken in such circumstances (and usually good processing involves some increase in contrast somewhere, especially in skies - which are much more unforgiving than this scene), you can rapidly get banding and other grunge, the risk being reduced if you ETTR.

I haven't exposed further "to the right" because those tiny white shells would max out to bleached white dots if I had, not that you can see at the resolution available here, but you could in an A2 print.

Log in or register to post comments


Brian Gayley's picture

Great tutorial. Thanks! Maybe for example sake could you show 2 sets of the same shot: 1 with the ETTR concept, and one ETTL to demonstrate the banding you mentioned? I have a bad habit of underexposing so this is great. Until 4 years ago I shot portrait, wedding and senior photography so had some income to buy all the toys. I have since trimmed the fat in my toy cabinet so am struggling with exposures not having a monitor calibrator. I have my screen WAY too bright most of the time and find myself ignoring the histogram (I know, shame shame).

Chris Jablonski's picture

Brian, Brian! I delete all my ETTLs! ;-) Well, I would if I ever took one...

I have been thinking of images that illustrate this issue. I'll try to find some suitable ones for another post.

Brian Gayley's picture

I know. I know. 255 has more data than 0:) Every time I see a halo I think "Oh yeah, right side dummy"

Chris Jablonski's picture

I'm not sure Brian, although I think the haloes are not related to this, but rather to what the software does to fiddle contrast, which involves exaggerating differences across boundaries. Or something...

Alan Brown's picture

Hey Chris, isn't there something up with your histograms? If you increase exposure the character/shape of the curve should be maintained but just slide to the right, correct?
The 2nd image almost looks like contrast/other manipulations added.

Personally i try to keep the histogram from bumping either end, and if the dynamic range is too great (and the composition worthwhile) will take 2 or more images to ensure I have al the data.

From a technical standpoint my understanding is that as long as you stay in-between to boundaries you have the data and it can be manipulated.

Intriguing stuff and a great discussion. I am very interested in the technical side and may be totally wrong on this so let me know your thoughts.

Brian Gayley's picture

That was my understanding also. If it's not clipped, it's there to use.

Chris Jablonski's picture

Apologies to all! I should've been clearer, Alan. The second image IS massively manipulated. I mainly wanted to show that the bland ETTR original can be a good place to start, despite appearances; the second image is just the way I wanted it, along with several variants I haven't posted.

I'll try to do what Brian mentioned, and show the limitations of ETTL as well, but couldn't find suitable files. I might have to generate some. Ruth's recent post of lilies in this Group does illustrate the limitations of ETTL.

Regarding your technical point, my understanding is different. If the image has a narrow dynamic range (i.e. occupies only a portion of the histogram), like Ruth's lilies, then ETTR can make a big difference and make it easier - or even just possible - to come up with a satisfactory edit.

The original file here was made in very flat light, of a fairly low-contrast subject, and still barely makes it in with no clipping at either end. Frequently, we have to accept some clipping, and then... it becomes confusing, because it's better to allow clipping at the dark end, "exposing to the left".

So the guideline is more like: "Expose to the right as far as possible while avoiding highlight clipping", which is what I did here. I deleted some exposures that showed no clipping on the "blinkies" but where those shells were maxed out, but falied to show up on the camera because they're so small.

Francisco B's picture

So are you saying there is more recovery if you expose as right as possible without hitting white? The common saying ive heard with exposure is that your recovery is generally better if you underexpose, not overexpose. I find that in aperture/shutter mode my camera is pretty darn good at exposing correctly most of the time.

Chris Jablonski's picture

To be honest, Francisco, I have followed the ETTR principle as you summarise it at the start, on the advice of others. While relatively new to digital, I heeded the advice of those with experience. However, Alan has raised doubts in my mind, as I've never really tested out ETTR vs ETTL. I hope to do so soon, and to post the results.

Popular advice can be wrong! And you've obviously heard different.

One can often bring out a lot of detail in what look like hopelessly dark shadows in RAW, while in general blown-out highlights are the death knell for an image, unless they're tiny points. Hence it's important not to overexpose. But in a low-contrast scene, where there is exposure latitude... I don't know. Yet.

Francisco B's picture

yeah it seems like there is no absolute answer so logically it probably depends on the scenario. I think recovery is so good on these new cameras that as long as ur in the ballpark it doesnt really matter that much.

I wonder why these new cameras aren't smart enough to do hdr style exposure (exposing for foreground and background) like the iphone does. I'm assuming you are somehow losing quality information doing both at once instead of bracketing, but i dont really know.