DPChallenge: A Digital Photography Contest You are not logged in. (log in or register
 

DPChallenge Forums >> Photography Discussion >> Expose to the right? Why?
Pages:  
Showing posts 1 - 18 of 18, (reverse)
AuthorThread
07/31/2006 03:55:41 PM · #1
Someone asked on another forum how the various members there deal with exposure when working with digital cameras. Here's my response. I thought it might be of use to others here on DPC. As the images are not hosted on DPC I don't think I'm able to embed them here but I've included links.

I shoot in RAW which give some latitude however it's worth understanding how the data is recorded.

In the real scene there may well be 7, 8 or more stops worth of light from the darkest details to the brightest. But a digital sensor can record only about 5. So you are necessarily going to lose some of the detail in the real scene (unless you use a tripod and take more than one exposure of the same scene but let's ignore that option for now).

So let's look at the five stops the camera sensor can record.

IF I have this right (and please someone correct me if I have as this is what I've understood after quite a bit of study but doesn't mean I have it right):

Let's say, for sake of illustration - I know the figure is too big, that the file size of this one shot is 31 MEG. Because of the linear way in which the sensor works fully 16 MEG of that data is used in recording the brightest stop. Then 8 MEG is used recording the next brightest stop. 4 MEG records the middle stop. 2 MEG records the second darkest and 1 MEG the darkest stop of data.

So... firstly, I want to make the MOST of that brightest stop. I don't want to underexpose so much that my entire picture data is squished into the darker four stops. Whilst I can, in RAW conversion, stretch that data across five stops worth all I'm doing is stretching 17 MEG of data instead of having 31 MEG of data in the first place.

I therefore use exposure compensation to expose to the right of the histogram ensuring I'm really making use of the stop with most data recorded.

HOWEVER, I do NOT want to overexpose. Blown highlights are, to my mind, very obvious and not visually appealing. Whereas if one loses detail in the darker shadows it's just a case of having more shadows than one might but perfectly visually acceptable.

So I need to ensure that my exposure is as far to the right of the histogram as it can be WITHOUT being cut off at the right.

That said, if it's a TINY TINY bit cut off at the right I can recover a little bit of the "lost" information in RAW conversion. So there is a little little leeway.

I.e. I want this:
//static.flickr.com/64/203121315_48e5e4fbe4_o.jpg

rather than this:
//static.flickr.com/59/203121316_567a7f0713_o.jpg

(original images on //www.flickr.com/photos/kavey/203121316/)

Having exposed to capture as much data as possible I can then spread it out in terms of contrast/ final brightness etc. by using the various tools I have in conversion software.
07/31/2006 04:16:55 PM · #2
Nice overview Kavey!
I just wanted to say that I also am a big proponent of "expose to the right." It's different than film, where the objective was normally to expose to obtain as near to the final result as possible.
I do want to weigh in on DSLR dynamic range. While I see the 5 stops figure posted very often, most current-model DSLRs are capable of >= 8 stops of DR. The 20D, for instance, weighs in at 8 stops, and a little more can be eked out by making use of the DR above the JPEG "blown highlight" indication. Accuracy of recovery of these highlights is not always 100%, and depends greatly on the RAW converter. Canon's converter, for example (DPP) pretty much tosses all color information in these highlights, but retains luminosity.
07/31/2006 04:24:57 PM · #3
Yes, that's true... some DSLRs do indeed have much higher dynamic range but the general rule applies. The vast majority of data is given over to the brightest stop with very little used to record the darkest. So it really makes sense to expose to the right without overexposing.

I use ACR and can recover some highlights but I try NOT to expose to rely on that regardless.

Message edited by author 2006-07-31 16:26:42.
07/31/2006 04:37:42 PM · #4
Originally posted by Kavey:

I use ACR and can recover some highlights but I try NOT to expose to rely on that regardless.


I also use ACR, and find that it seems to do the best job of recovery out of those converters I've tried.
07/31/2006 04:39:12 PM · #5
That's good to know as I've not had much experience of the others, not enough to make a sound judgement on them.
07/31/2006 04:48:21 PM · #6
It's like playing blackjack
07/31/2006 04:51:26 PM · #7
Originally posted by kirbic:

Nice overview Kavey!
I just wanted to say that I also am a big proponent of "expose to the right." It's different than film, where the objective was normally to expose to obtain as near to the final result as possible.
I do want to weigh in on DSLR dynamic range. While I see the 5 stops figure posted very often, most current-model DSLRs are capable of >= 8 stops of DR. The 20D, for instance, weighs in at 8 stops, and a little more can be eked out by making use of the DR above the JPEG "blown highlight" indication. Accuracy of recovery of these highlights is not always 100%, and depends greatly on the RAW converter. Canon's converter, for example (DPP) pretty much tosses all color information in these highlights, but retains luminosity.

Question for Kirbic....What raw converter do you think does the best job with hilites?
07/31/2006 04:53:33 PM · #8
Originally posted by cloudsme:

Originally posted by kirbic:

Nice overview Kavey!
I just wanted to say that I also am a big proponent of "expose to the right." It's different than film, where the objective was normally to expose to obtain as near to the final result as possible.
I do want to weigh in on DSLR dynamic range. While I see the 5 stops figure posted very often, most current-model DSLRs are capable of >= 8 stops of DR. The 20D, for instance, weighs in at 8 stops, and a little more can be eked out by making use of the DR above the JPEG "blown highlight" indication. Accuracy of recovery of these highlights is not always 100%, and depends greatly on the RAW converter. Canon's converter, for example (DPP) pretty much tosses all color information in these highlights, but retains luminosity.

Question for Kirbic....What raw converter do you think does the best job with hilites?


I know you asked Kirbic but Bibble's "highlight recovery" tool works great!
07/31/2006 04:56:05 PM · #9
Originally posted by cloudsme:

It's like playing blackjack


Quite accurate, especially since the in-camera histogram is not always the best indicator (speaking of cameras that do not have an RGB histogram).

Per your other question, see my post above. ACR seems to do the best job of those convereters I've used. I have not used the latest versions of a few converters. The one's I'm not familiar enough with to really comment are Bibble, Silkypix, and to some extent C1 (it's been a while since I played with it).
07/31/2006 05:03:28 PM · #10
I too am a big fan of ETTR shooting.
I could never figure out why film always had so much detail in the midtone to shadow areas without blowing out the sky in landscape shoots.
One of our local photographers, (Obie Oberholzer) in particular shot landscapes in the Karoo in the 80's with medium format, featuring the sun in the sky, and oodles of detail in the rocky land around him.
I eventually learnt to close up my aperture (thanks to the kindly folks here at DPC) and that helped bring light sources like the sun back into my pictures (instead of skulking away from them...).
Then, on learning why to expose to the right, I incorporated that into my shooting workflow, and now retrieve far more information than I previously had to be satisfied with, particluarly in the shadow detail areas.
I also suffer from far less noise now than I previously used to.
Can't help with the converters question though.
I've yet to do proper studies on that.

Anyone?
07/31/2006 05:08:26 PM · #11
Can someone please explain what should be done to get most of the details on the right hand side of the histogram. Like what options to use?

Thanks.
07/31/2006 05:13:20 PM · #12
Are we talking about +1/3 to +1 to the right of center?

I feel like Yogi Berra...I'm 80% confused but I know 50% of what you're talking about!
07/31/2006 05:14:12 PM · #13
Originally posted by MicShortland:

Can someone please explain what should be done to get most of the details on the right hand side of the histogram. Like what options to use?

Thanks.


Once you've exposed to get the highlights just short of being clipped, you now are faced with an image that, upon loading on the 'pooter, may look "overexposed" compared to your vision for the final shot. How you adjust depends on how you shot the pic.
- If you shot in JPEG mode, then use curves to reduce exposure and tailor the contrast (and possibly color balance) to suit your vision
- If you shot in RAW, do some negative exposure compensation during conversion, and tweak the white balance to suit.

The second scenario makes for a much less severely edited JPEG, which will, in the end, yield smoother tonal transitions and less chance of posterization.
07/31/2006 05:15:10 PM · #14
Mic, since most DSLRs do not offer a live histogram the way I do it is to first take a shot using the camera's automatic exposure settings according the camera meter. Then I view the histogram of the shot I've taken. Then I use my exposure compensation feature to adjust the exposure up or down from what the meter suggests until I get what I want in my histogram.

Of course, you don't need to do this with every shot since you'll usually be taking several in the same conditions and can work out how much to over or under expose at the start and stick to it (checking the results now and then to ensure it's still accurate) for all the rest of the shots in those conditions.

Message edited by author 2006-07-31 17:16:31.
07/31/2006 05:15:20 PM · #15
A very Good overview of Exposing Right on The Luminous Landscape.
07/31/2006 05:16:16 PM · #16
This was fascinating to read, so thanks!

Does the "most data in the highest stop" work for DSLRs only or also for P&S?

Also, I assume that when you say "stop" you really mean "brightness zone"? Since there's only one f-stop setting for the entire photo?

Finally, is it that the most data goes to the brightest zone in each photo? So it depends on your photo, with the most data going to a midrange brightness in a dark shot and a brilliantly bright one in a light shot? Or it's the most data goes to the rightmost zone in the histogram, regardless of what's in a specific photo?
07/31/2006 05:27:27 PM · #17
Originally posted by levyj413:

This was fascinating to read, so thanks!

Does the "most data in the highest stop" work for DSLRs only or also for P&S?


Yes, works for P&S as well. It works this way because all digital sensor sued in today's cameras are linear devices, but our eyes are logarithmic.

Originally posted by levyj413:

Also, I assume that when you say "stop" you really mean "brightness zone"? Since there's only one f-stop setting for the entire photo?


Yes. There is one aperture setting and one "stop" of difference is a halving (or doubling) of light. That's what is being referred to. The upper half of the linear scale is the first stop, the first half of what's left is the next stop, etcetera.

Originally posted by levyj413:

Finally, is it that the most data goes to the brightest zone in each photo? So it depends on your photo, with the most data going to a midrange brightness in a dark shot and a brilliantly bright one in a light shot? Or it's the most data goes to the rightmost zone in the histogram, regardless of what's in a specific photo?


Data always is scaled to the whole histogram. So if there are no values in the upper stop of the histogram, you've lost half your data depth. Instead of working with 256 levels per channel, you're living in the bottom 128.
07/31/2006 05:31:57 PM · #18
Ah. Thanks especially for pointing out the effect of the linear scale vs. logarithmic measurement. Hadn't occurred to me that the first stop is the entire right half of the histogram!
Pages:  
Current Server Time: 05/25/2025 05:35:25 AM

Please log in or register to post to the forums.


Home - Challenges - Community - League - Photos - Cameras - Lenses - Learn - Help - Terms of Use - Privacy - Top ^
DPChallenge, and website content and design, Copyright © 2001-2025 Challenging Technologies, LLC.
All digital photo copyrights belong to the photographers and may not be used without permission.
Current Server Time: 05/25/2025 05:35:25 AM EDT.