under or over exposed?

[photogenicimages]

Ok, obviously it's best to get exposure right the first time in camera, but if thats just not happening which is better- under or over exposing a pic? I'm not talking pictures so blown out that you lose all data, just a little on the bright side and a few highlights lost. I'm just not sure which is 'healthier' if you will for the file, to be bumped up in ps or down.

[Daniel Bates]

Underexposing and "pushing" in post increases noise. However, the shadows contain much more information than the highlights do, so you'll have (noisy) detail if you underexpose, but no detail if you over expose. I almost always expose to keep the highlights in gamut and then adjust as needed in PS.

[kenw]

^ my plan exactly.

[*Mike*]

The rule of thumb is to "shoot to the right" of the histogram. You want the data as far to the right (bright) as you can, without blowing out the highlights. Due to the linear gamma of the sensor, the most data rich part of the histogram is the brightest couple of stops.

So, if you're not going to due much in the way of post, then expose for how you want the shot to look. If you're uncertain, or want to have the most data to work with when you start doing manipulations, then shoot to the right - which may leave the image looking a little over-exposed.

Highlights contain more detail info than shadows...

[*Mike*]

Bruce Fraser's Real World Camera Raw with Adobe Photoshop CS2 has some great info on this kinda thing... and it is a relatively easy read.

[Daniel Bates]

Mike's right, the highlights do contain more detail than the shadows. I misspoke.

A blown highlight is unrecoverable, however, while shadows are more often salvageable.

[shutterdrone]

Originally Posted by Daniel Bates

Underexposing and "pushing" in post increases noise. However, the shadows contain much more information than the highlights do, so you'll have (noisy) detail if you underexpose, but no detail if you over expose. I almost always expose to keep the highlights in gamut and then adjust as needed in PS.

As I understand it, that statement is correct for film, but not correct for digital sensors.

In either - if you blow a highlight, you've blown it -- it's gone.

Film has greater latitude in the shadows, and with film it's generally advised as better to underexpose and push the film, than it is to overexpose and pull it. It has, as I understand it, generally logarithmic exposure scale. (I could be wrong on whether its logarithmic, but the end result works just fine in film.)

A digital sensor, on the other hand, has a linear exposure scale - each stop of brightness is 1/2 the possible values of the previous. In a 12-bit raw file, you have a max of 4,096 possible tonal values. Of those, 2048 are available in the 1st (brightest stop) and 128 in the fifth (roughly darkest stop). So, shadows contain much _less_ data than highlights coming from your digital sensor. It's been discussed quite a bit that if maximum detail is the desired end-result, then one does better to expose to the right -- adjusting exposure until the brightest values are just about to blow, but not quite blowing. At that point you may adjust down exposure and achieve greater detail.

As the linked article also notes, you want to do this with RAW, you cannot use this technique when shooting in jpeg mode.

!c

[photogenicimages]

Thanks guys! You linked to some great info and resources for us to look at, we really appreciate it.