Posted by Paul Bojan on February 19, 2011 at 8:57 PM
Hello all. Before I ask my question I must give a thanks to all that have commented on my photos and had replied to my threads. This is a very helpful website. More than I actually expected.
Back to the important stuff. Many of you have commented on my photos about blow out whites, and I can see what you are saying. I often get fooled by my histogram and get overly cautious and go +1/3 or more exp compensation just trying to avoid the left shift in the histogram. Is it better to shoot a bit dark and add a bit of light in post processing or try to get it dead on in the camera. I find right now this is one of my more troubling areas. I have CS4 and light room both I am very new with. its a bit overwhelming right now with all their features and options.
Highlights can be difficult to control. There are exellent tutorials at Cambridge In Color. There are two separate tutorials on using the histogram.
Following them should help with your exposure issue. While Photoshop is a great tool for tweaking tonal values, etc., you have to start out with an image that well exposed. The fewer corrections done in Photoshop, the better, imho.
--P
Posted by Ron Bielefeld on February 20, 2011 at 9:13 AM
Overall, I agree with Preston. If anything, get the exposure dead-on in camera. You really can't go wrong with that being your goal. However, once you are confident you can get the exposure right in-camera/in the field, then you can do some different things.
In general, I do not like to underexpose an image; by mistake or on purpose. Dark means less data to a digital sensor and thus if you shoot dark and try to bring up the exposure in post-processing you will get noise added to your image as photoshop tries to make up data for your image in the darkest areas of the image.
However, if you shoot a bit too bright, to where the brightest areas are still Ok, that is, not blown, you are maximizing the data the sensor collects for any particular image and you can bring down those areas to the correct level without the worry of adding noise. Brighter equals more data to a digital sensor.
By shooting a bit bright you get more detail in the shadow parts of your image, say under wings of birds and the like. I like to shoot just a bit bright and then bring down the different areas (shadows, midtones, highlights) of the image different amounts with the goal of getting the best detail, yet natural look, in as much of the image as possible.
Oh, very important point, this only works if you shoot RAW. If you are shooting Jpeg you really need to get your exposure correct in-camera. Jpegs are compressed and you loose a lot of the data the camera sensor collects by shooting Jpeg. That is the huge advantage to shooting RAW. You have all the data the sensor collects to work with in post-processing.
I hope this helps you Paul.
Posted by Jim Bullard on February 20, 2011 at 10:39 AM
Hi Paul,
You didn't say whether you are shooting RAW or JPG. If you are shooting RAW you can recover some highlight detail because the sensor actually records a broader brightness range than the histogram shows which is a JPG equivalent. Shooting JPG you get a standardized conversion from RAW and you should try to keep the exposure as close as possible to the right side without running up the side whenever the brightness range exceeds the width of the histogram. If if you are going to lose either shadow detail or highlight detail it is better to lose shadow detail. Exception: When there are specular highlights in the image it is okay to lose them. Your exposure should be based on highlights where you want some degree of detail, Zone IX in zone system terms.
I have to disagree with Ron's statement "Brighter equals more data to a digital sensor". It is a common notion but not true. Images are made up of individual pixels. Every pixel is represented by exactly the same amount of data as every other pixel regardless of tone. For a full explanation see my blog at http://jims-ramblings.blogspot.com/2010/02/right-idea-wrong-reason.html.
Posted by RL Potts on February 21, 2011 at 1:31 AM
If you cram an exposure against the right side of the histogram, you can produce an image approaching HDR. If you expose an image on the left side of the histgram, you get an image that descends quickly into noise and loss of detail. For whatever reason, the 1024 bit rational seems to work out. So don't pay to much attention to anyone's ramblings (including mine).
Exposure is an art and sensors are not perfect. Chroma performance is not a straight line function, in other words, there is no rule of thumb that will give best performance in every situation. For myself, I do all exposure settings manually. No meter that I have used is very reliable. Check for yourself, You can get readings that range all over the place in any scene that is complexly lit. For my work, a properly exposed image is somewhat to very over-exposed until I adjust it. Important shots should be Raw. That's what I think. Also keep in mind that the histogram is a jpeg representation of a raw image.
Posted by David Bostedo on February 21, 2011 at 9:26 AM
To clarify what Jim linked to and RL mentioned, by Exposing to the Right, you are simply maximizing the signal-to-noise ratio of the sensor. (As you mentioned, Jim, this may show up in the final image as related to saturation, but the sensor itself is collecting single value monochrome data through a colored filter, so has no idea of the actual color anyway.)
Basically, in dark areas there is very little light hitting the sensor, and the noise that is always present in the circuitry (via inaccuracy in reading, "static" introduced by neighboring circuits, etc.) will be much more significant in relation to light hitting the sensor. So when you read out the values, and de-mosaic to determine the final pixel color (during RAW development, or when creating a JPG in camera - prior to that pixels don't really have a color) you get values jumping around based on noise.
The way to minimize this is to allow more light to hit the sensor - that is, increase your exposure. The noise is pretty constant, but raising the signal by getting more light will make the noise matter less. That's why exposing to the right works.
The "number of levels available", or "number of bits available", was a misleading explanation put forward first on Luminous Landscape and since corrected (although the original article is still there I think).
Discussion
5 repliesBack to the important stuff. Many of you have commented on my photos about blow out whites, and I can see what you are saying. I often get fooled by my histogram and get overly cautious and go +1/3 or more exp compensation just trying to avoid the left shift in the histogram. Is it better to shoot a bit dark and add a bit of light in post processing or try to get it dead on in the camera. I find right now this is one of my more troubling areas. I have CS4 and light room both I am very new with. its a bit overwhelming right now with all their features and options.
Highlights can be difficult to control. There are exellent tutorials at Cambridge In Color. There are two separate tutorials on using the histogram.
Following them should help with your exposure issue. While Photoshop is a great tool for tweaking tonal values, etc., you have to start out with an image that well exposed. The fewer corrections done in Photoshop, the better, imho.
--P
In general, I do not like to underexpose an image; by mistake or on purpose. Dark means less data to a digital sensor and thus if you shoot dark and try to bring up the exposure in post-processing you will get noise added to your image as photoshop tries to make up data for your image in the darkest areas of the image.
However, if you shoot a bit too bright, to where the brightest areas are still Ok, that is, not blown, you are maximizing the data the sensor collects for any particular image and you can bring down those areas to the correct level without the worry of adding noise. Brighter equals more data to a digital sensor.
By shooting a bit bright you get more detail in the shadow parts of your image, say under wings of birds and the like. I like to shoot just a bit bright and then bring down the different areas (shadows, midtones, highlights) of the image different amounts with the goal of getting the best detail, yet natural look, in as much of the image as possible.
Oh, very important point, this only works if you shoot RAW. If you are shooting Jpeg you really need to get your exposure correct in-camera. Jpegs are compressed and you loose a lot of the data the camera sensor collects by shooting Jpeg. That is the huge advantage to shooting RAW. You have all the data the sensor collects to work with in post-processing.
I hope this helps you Paul.
You didn't say whether you are shooting RAW or JPG. If you are shooting RAW you can recover some highlight detail because the sensor actually records a broader brightness range than the histogram shows which is a JPG equivalent. Shooting JPG you get a standardized conversion from RAW and you should try to keep the exposure as close as possible to the right side without running up the side whenever the brightness range exceeds the width of the histogram. If if you are going to lose either shadow detail or highlight detail it is better to lose shadow detail. Exception: When there are specular highlights in the image it is okay to lose them. Your exposure should be based on highlights where you want some degree of detail, Zone IX in zone system terms.
I have to disagree with Ron's statement "Brighter equals more data to a digital sensor". It is a common notion but not true. Images are made up of individual pixels. Every pixel is represented by exactly the same amount of data as every other pixel regardless of tone. For a full explanation see my blog at http://jims-ramblings.blogspot.com/2010/02/right-idea-wrong-reason.html.
Exposure is an art and sensors are not perfect. Chroma performance is not a straight line function, in other words, there is no rule of thumb that will give best performance in every situation. For myself, I do all exposure settings manually. No meter that I have used is very reliable. Check for yourself, You can get readings that range all over the place in any scene that is complexly lit. For my work, a properly exposed image is somewhat to very over-exposed until I adjust it. Important shots should be Raw. That's what I think. Also keep in mind that the histogram is a jpeg representation of a raw image.
Basically, in dark areas there is very little light hitting the sensor, and the noise that is always present in the circuitry (via inaccuracy in reading, "static" introduced by neighboring circuits, etc.) will be much more significant in relation to light hitting the sensor. So when you read out the values, and de-mosaic to determine the final pixel color (during RAW development, or when creating a JPG in camera - prior to that pixels don't really have a color) you get values jumping around based on noise.
The way to minimize this is to allow more light to hit the sensor - that is, increase your exposure. The noise is pretty constant, but raising the signal by getting more light will make the noise matter less. That's why exposing to the right works.
The "number of levels available", or "number of bits available", was a misleading explanation put forward first on Luminous Landscape and since corrected (although the original article is still there I think).