shooting to the right/digital blending

[OpticalMedia]

I'm wonder if anyone can help put me in the spotlight. I am trying  to establish the most effiecient technique I can use to get the best image quality out of my digital SLR while trying to cover the max possible dynamic range. However after studing all of michaels pages dedicated to this topic i am still confused. I Have a few questions in a specific order :

1. When would I use the expose to the right theory?
2. If i'm going to use this theory, is it not better rather to take multiple exposures and then digitally blend them?
3. What is the diffence btw the two methods?
4. Can they be used together ?
 
i'm sure with answers that will follow i'l have more questions.
Thank you in advance

[Tim Gray]

Always keep your histogram as much to the right as possible withough clipping the highlights.  This gives you as much postprocessing rooom as possible to manage the shadows.

If going as far right as possible still results in clipped shadows then you move into a blending mode by increasing the exposoure (time not f stop) by a half to full stop and take a second shot.  Repeat until the shadows are no longer clipped.

ETTR gives you maximum latitude in processing a single shot but sometime the dynamic range can't be captured in a single shot, hence blending of 2 (or more) shots.  Sometimes if you didn't bracket a scene with a _bit_ too much dynamic range you can still get a reasonable image by double processing the same raw shot and blending.

[Peter McLennan]

1. When would I use the expose to the right theory?

On every exposure you make.  ETR mazimizes the Dynamic Range available from your imager.

2. If i'm going to use this theory, is it not better rather to take multiple exposures and then digitally blend them?

Not necessarily.  Some subjects cannot accommodate multiple exposures.  Sports, for example.   Any situation with subject motion between exposures can't easily utilize blending procedures.

3. What is the diffence btw the two methods?

ETR is a way of maximizing the dynamic range available from a single exposure to an imaging system.  Multiple exposure blending is a way of maximizing DR from a given photographic situation.  By intentionally over and under exposing separate shots and combining the results, you can extract more information from a high contrast scene.

4. Can they be used together ?

Yes and no.  Note that exposure blending requires the combination of both "under" and "over" exposed images.  These may or may not fall within the ETR principle

 
i'm sure with answers that will follow i'l have more questions.
Thank you in advance
[a href=\"index.php?act=findpost&pid=71424\"][{POST_SNAPBACK}][/a]

[sgwrx]

sometimes you'll find yourself shooting a scene where the histogram shows as a big hump in the middle, nothing much down left and nothing much up right - like a bell curve.  this is especially a time when you want to use ETTR.

sometimes there are scenes that contain one or two white objects like the hull of a sailboat.  if it's pretty much the only thing that's pure white, and everything else in the scene is more or less mid-tone or darker, you can still use ETTR even though there might be some blowout in the hull.  of course, if you want to retain detail like a small thing blue pinstripe, then it wouldn't be worth it.

[dwdallam]

I've seen plenty of images I've taken that are exposed to the left and they are superior to those I exposed to the right. This is because the exposure is the camera's best guess for the scene. It is many times fooled into over or underexposing and only your LCD can show you that. The histogram will not guide you because it will look like the exposure is dead on, and then be visually blown out on the LCD.

Also, if you expose to the right, you can lose color saturation that RAW processing will not recover. So this ideas in theory is correct because it helps the sensor capture more detail in the shadows, but in practice, it can play havoc with your images.

As an example, I shoot many dusk and night shots, and I almost never get a nice, accurate exposure reading from my 20D or  nor now my 5D unless I use something like center weighted, aim for the most average brightness of the scene, and have a gradual light to dark transition in the scene--something like taking a picture of a boat and it's mast about 10 minutes after sunset. But then if I expose to the right, I lose color saturation, although I may save detail in the shadows.

But then again, due to inaccurate meter readings, I may be exposing to the right and not even know it. That being the case, what I do is bracket many shots in these situations. After all, you are shooting digital now. It cost nothing to shoot brackets. It does take more time though, so choose brackets logically, not willy nilly, or you will end up with 100s of images to look at. That eats hours, and hours, and h. . . . .

[bjanes]

I've seen plenty of images I've taken that are exposed to the left and they are superior to those I exposed to the right. This is because the exposure is the camera's best guess for the scene. It is many times fooled into over or underexposing and only your LCD can show you that. The histogram will not guide you because it will look like the exposure is dead on, and then be visually blown out on the LCD.

Also, if you expose to the right, you can lose color saturation that RAW processing will not recover. So this ideas in theory is correct because it helps the sensor capture more detail in the shadows, but in practice, it can play havoc with your images.

ill end up with 100s of images to look at. That eats hours, and hours, and h. . . . .
[a href=\"index.php?act=findpost&pid=71456\"][{POST_SNAPBACK}][/a]

I think that the above cautions about color saturation are overblown. The sensors in most digital cameras are linear right up to the point of clipping. A few show some nonlinearity near the top. As long as you remain in the linear range and avoid clipping, there will be no loss of saturation with exposure to the right. When clipping occurs, there will be color shifts and loss of highlight detail. Often, it is the white highlights that blow first, and in this situation the green channel will blow first with daylight white balance, resulting in a magenta shift.

When exposing low key subjects to the right, the picture will appear too light, and a negative exposure compensation in the raw converter will be needed. Since digital capture is linear in all channels, no color shifts or changes in saturation will occur. However, the shadows in the image will have less noise because of the ETTR.

Bill

[John Sheehy]

If going as far right as possible still results in clipped shadows[a href=\"index.php?act=findpost&pid=71432\"][{POST_SNAPBACK}][/a]

Eeek!  That phrase "clipped shadows" sounds like nails on a blackboard to me.  Shadows don't clip in camera captures; only in Levels and Curves tools in post-processing.

In capture, shadows get obscurred in noise, they don't get clipped.  They're still there; just hard to see.

[larryb]

Always keep your histogram as much to the right as possible withough clipping the highlights.  This gives you as much postprocessing rooom as possible to manage the shadows.

If going as far right as possible still results in clipped shadows then you move into a blending mode by increasing the exposoure (time not f stop) by a half to full stop and take a second shot.  Repeat until the shadows are no longer clipped.

ETTR gives you maximum latitude in processing a single shot but sometime the dynamic range can't be captured in a single shot, hence blending of 2 (or more) shots.  Sometimes if you didn't bracket a scene with a _bit_ too much dynamic range you can still get a reasonable image by double processing the same raw shot and blending.
[a href=\"index.php?act=findpost&pid=71432\"][{POST_SNAPBACK}][/a]

[larryb]

[attachment=885:attachment][attachment=886:attachment]

Always keep your histogram as much to the right as possible withough clipping the highlights.  This gives you as much postprocessing rooom as possible to manage the shadows.

If going as far right as possible still results in clipped shadows then you move into a blending mode by increasing the exposoure (time not f stop) by a half to full stop and take a second shot.  Repeat until the shadows are no longer clipped.

ETTR gives you maximum latitude in processing a single shot but sometime the dynamic range can't be captured in a single shot, hence blending of 2 (or more) shots.  Sometimes if you didn't bracket a scene with a _bit_ too much dynamic range you can still get a reasonable image by double processing the same raw shot and blending.
[a href=\"index.php?act=findpost&pid=71432\"][{POST_SNAPBACK}][/a]

The "Understanding Digital Blending" tutorial has been the most helpful shortcut. I have been using it on single scanned negs, and even the attached digital camera photo. The process is brilliant. In many cases, especially with color negs, I don't need to make multiple exposures.

[dwdallam]

I think that the above cautions about color saturation are overblown. The sensors in most digital cameras are linear right up to the point of clipping. A few show some nonlinearity near the top. As long as you remain in the linear range and avoid clipping, there will be no loss of saturation with exposure to the right. When clipping occurs, there will be color shifts and loss of highlight detail. Often, it is the white highlights that blow first, and in this situation the green channel will blow first with daylight white balance, resulting in a magenta shift.

When exposing low key subjects to the right, the picture will appear too light, and a negative exposure compensation in the raw converter will be needed. Since digital capture is linear in all channels, no color shifts or changes in saturation will occur. However, the shadows in the image will have less noise because of the ETTR.

Bill
[a href=\"index.php?act=findpost&pid=71461\"][{POST_SNAPBACK}][/a]

That is inaccurate in my practical experience given my specific explanation. Your technical explanation seems true enough, but in practice, it is not that easy. First, your camera may not be giving you an accurate exposure. In my case shooting highly contrasted sunsets and night shots, sometimes exposing to the right would mean blowing the entire scene to white. Using the histogram, it looks like it's bunched up to the left side. When I get home, it's near perfect. The only way to get around this inaccurate in camera exposure is to use a light meter, measure the highlights and the darkest areas, and set your camera manually for the exposure inbetween those two measurements. And still, this may not produce the "best" exposure in high contrast situations. Even using the cameras spot meter for measuring distant areas of a scene sometimes does not work. What is true is that the more negative exposure you take, the less detail you will have in shadows, and more noise, as you point out.

If shadow details are more important than your mid tones and highlights, take that into consideration and expose as needed, and vice-versa. In many of my shots, the shadows will be black anyway, as in forground silhouettes against a brightly colored sunset, and the colors and highlights are critical. However, I have the advantage of taking left and right exposures because of the situation, so I do take advantage of that.

Also note that in some images, you don't want a human eye like contrast between your highlights and shadows. You want that film/digital contrast, from darker to lighter to pull the eye into the scene. I have a friend who blends all of his sunset shots and I've heard many say that they lack something that keeps them interested in the images, but they all say that the colors are very pretty and the details are all very nice. I think part of that is becsaue the entire image has very little contrast from the highlights to the shadows--everything is very detialed and visible. Best thing to do is bracket and see which ones appeal to you the most for the effect you want, and then composite if needed. Be aware of the technical aspects, that Bill points out,  are pretty much fact--you will lose detail in the shadows the more underexposed you get. And you will lose highlights and color if you over expose too much. I do not think there is anyway for a human operating a camera to get the "perfect" exposure by simply exposing to the right. But that is my opinion and I will not try to support that assertion.

[John Sheehy]

I do not think there is anyway for a human operating a camera to get the "perfect" exposure by simply exposing to the right. But that is my opinion and I will not try to support that assertion.
[a href=\"index.php?act=findpost&pid=78734\"][{POST_SNAPBACK}][/a]

Linear data benefits nothing from under-exposure, except to avoid clipping of highlights.  Any problem due to high, non-clipped/saturated exposure is the fault of the RAW converter, or the parameters used.

Unfortunately, popular converters like ACR provide no way for the user to simply scale the exposure before the rest of the conversion.  Loss of color saturation in highlights, other than from clipping, is from converter contrast reduction of what it assumes, incorrectly, to be specular highlights.

[dwdallam]

Linear data benefits nothing from under-exposure, except to avoid clipping of highlights.  Any problem due to high, non-clipped/saturated exposure is the fault of the RAW converter, or the parameters used.

Unfortunately, popular converters like ACR provide no way for the user to simply scale the exposure before the rest of the conversion.  Loss of color saturation in highlights, other than from clipping, is from converter contrast reduction of what it assumes, incorrectly, to be specular highlights.
[a href=\"index.php?act=findpost&pid=78783\"][{POST_SNAPBACK}][/a]

hehe. There's always more to the story than meets the eye. Thanks for that information.

[thierrylegros396]

Linear data benefits nothing from under-exposure, except to avoid clipping of highlights.  Any problem due to high, non-clipped/saturated exposure is the fault of the RAW converter, or the parameters used.

Unfortunately, popular converters like ACR provide no way for the user to simply scale the exposure before the rest of the conversion.  Loss of color saturation in highlights, other than from clipping, is from converter contrast reduction of what it assumes, incorrectly, to be specular highlights.
[a href=\"index.php?act=findpost&pid=78783\"][{POST_SNAPBACK}][/a]

OK, Your right, but how to easily recover saturation using ACR      

[dwdallam]

OK, Your right, but how to easily recover saturation using ACR     
[a href=\"index.php?act=findpost&pid=78876\"][{POST_SNAPBACK}][/a]

If the data did not record due to technological limitations, you can't. It's not there to begin with.

[thierrylegros396]

Sorry for the previous post, the word "loss" is missing which change all the meaning.

Linear data benefits nothing from under-exposure, except to avoid clipping of highlights.  Any problem due to high, non-clipped/saturated exposure is the fault of the RAW converter, or the parameters used.

Unfortunately, popular converters like ACR provide no way for the user to simply scale the exposure before the rest of the conversion.  Loss of color saturation in highlights, other than from clipping, is from converter contrast reduction of what it assumes, incorrectly, to be specular highlights.
[a href=\"index.php?act=findpost&pid=78783\"][{POST_SNAPBACK}][/a]

OK, but how to easily recover loss of saturation in the highlights without affecting saturation in the "mid and dark areas" with ACR      

[John Sheehy]

Sorry for the previous post, the word "loss" is missing which change all the meaning.
OK, but how to easily recover loss of saturation in the highlights without affecting saturation in the "mid and dark areas" with ACR     
[a href=\"index.php?act=findpost&pid=79042\"][{POST_SNAPBACK}][/a]

My point is that there isn't an easy way.  You have to play with the exposure, brightness, contrast and shadow sliders until you get what you want.

[thierrylegros396]

My point is that there isn't an easy way.  You have to play with the exposure, brightness, contrast and shadow sliders until you get what you want.
[a href=\"index.php?act=findpost&pid=79072\"][{POST_SNAPBACK}][/a]

Yes, that's what I do.
But it can take a lot of time !

So, when I have a "shallow bell" histogram, I prefer not to expose to the right, but expose more in the center.

Results are often better !

The same problem can occur when you have to play with curves in order to decrease local contrast. You loose color saturation in that area.

And if you increase color saturation, you obtain too much color saturated dark areas !

[Ray]

I'm surprised no-one has mentioned auto exposure bracketing. If you really want to get the maximum dynamic range from a single exposure you are probably either going to have to mess around with an external light meter, or use the camera's histogram and highlight warning to assess the accuracy of the exposure, and then retake the shot if there's any doubt.

The histogram is based on a jpeg conversion and the jpeg conversion and the appearance of the histogram is influenced by the in-camera contrast settings. The RAW image, however, is not influenced by such settings.

I find the best approach is to do one's best to get an ETTR exposure but auto bracket +/- 2/3rds of a stop or so, or plus 1/3rd and minus 1 stop, whatever, as an insurance policy. If you are using 4gb flash cards there should be no problem.

I think the default setting is 'normal' exposure first and it's probably best to keep it that way because the normal exposure should be right most of the time. When trying to capture the moment within a very small time frame, the other 2 shots might not be ideal even though one is better exposed. On the other hand, it's not always possible to anticipate the best moment and there's still a possibility of getting both the perfect moment and the perfect exposure when using auto bracketing.

I can't see there's anything to lose by adopting the practice, except storage space.

[Tim Gray]

I'm surprised no-one has mentioned auto exposure bracketing.
...
I can't see there's anything to lose by adopting the practice, except storage space.
[a href=\"index.php?act=findpost&pid=79197\"][{POST_SNAPBACK}][/a]

I'm a big fan of bracketing, but more often than not at least for marginal DR problems I find that the "well exposed" shot of the 3 gives me all the latitude I need, notwithstanding some flashing on the display or an "iffy" histogram.

Although hard disk storage isn't an issue, CF storage is, as is processing time - which includes downloading the redundant files, building the thumbnails and actually executing some kind of mental judgement process against the files all of which chews up "real" time.

IMHO the "best" solution would be a histogram/display that accurately reflects the DR captured by the RAW file.

[henk]

Perhaps a bit "analog" but I still use my Cokin graduated neutral density filters!.  when the dynamic range is to big for my 5D. IMHO its much better to create a 100%image on fore hand than to fidle your way in Photoshop with Digital blending.
Henk