Dan Margulis Applied Color Theory

Is it Best to Overexpose?

Date: Sun, 31 Aug 2003 15:45:29 -0000
   From: Henry Domke
Subject: Is it best to overexpose when shooting digital cameras?

I came across a very interesting idea and wanted to know your thoughts on it. The concept is that the ideal exposure for digital capture is to get your histogram as close to the right side as possible but not so close as to cause the over exposure indicator to flash.

It is explained in this article:
http: //luminous-landscape.com/tutorials/expose-right.shtml

What do you think?

Thanks,
Henry F. Domke

www.henrydomke.com
________________________________________________________________________

Date: Mon, 1 Sep 2003 12:28:31 EDT
   From: Joe Butts
Subject: Re: Is it best to overexpose when shooting digital cameras?

As suggested in the article; Run your own tests and see for yourself the end results.

One of the biggest problems today is that we have access to so much information via the web and you don't know the individuals credentials or qualifications regarding the information presented. The only way to confirm whether or not it works is to try it for yourself. Then, sometimes, you must also consider; What's the worst thing that can happen if I do try this? Trying some pushing of the exposure on your digital images as a test will cause no harm other than some lost time -- maybe. Then again, you might find that they are right. It all makes sense. But, note that they are only talking about shooting in RAW format. You might find this a bit hazardous in jpeg. Or, if you're shooting a social function where you're hoping to get your exposures close enough to use for final output, and you're working on that high end of the curve, it only takes a very slight overexposure to give you an image that is unusable or very difficult to correct and make look as good as you want.

But, try it! You just might like it. Besides, everyone's opinion here is just the same as I mentioned above. How do you know the quality of the opinion and feedback you're getting. Only your own personal tests will tell you for sure.

Joe Butts
Joe Butts Photography
505-388-2826
________________________________________________________________________

Date: Mon, 1 Sep 2003 01:10:53 -0500
   From: Jim Bean
Subject: Re: Is it best to overexpose when shooting digital cameras?

hello henry, it is not necessary to read the article, if you were in a controlled environment.. you could 'fish around' for the 'perfect' exposure appropriate for that particular image and encroach upon that point of no return-highlight...however, if you are using some form of exposure meter..and creating a handful of images containing a variety of tone ranges without the benefit of  'testing for each exposure'  it would be more practical to leave the capture a bit dense/dark/thin, capture raw and establish the "correct exposure" back at your desktop.   enjoy the day, jim
________________________________________________________________________

Date: Mon, 1 Sep 2003 08:31:19 -0700
   From: Lee Varis
Subject: Re: Is it best to overexpose when shooting digital cameras?

This has to do with minimizing noise in digital camera captures. Most camera manufacturers set the parameters for their raw file conversions to optimize for the best trade off between signal to noise ratios and ISO. The ideal ISO for most of the DSLR CCD/CMOS chips, as far as signal to noise goes, is much lower than the normal rating of 100 ISO that the manufacturer advertises. Most photographers feel better with a general purpose ISO of 100 rather than 25 so manufacturers accommodate them by boosting the gain on the analog signal of the chip.

This amplification of the signal increases the level of the noise. The fact of the matter is that the noise level on most chips is so low that amplifying the signal this way has little impact on the final image. However, low values will show the increased noise because the noise level has been amplified above the lowest signal/black threshold. Much of this noise can be, and is, post processed out in the raw file "developing" stage that renders a full color RGB from the high-bit "mosaic" grayscale raw file.

All of these things involve various design trade-offs to give the best overall image file under the widest range of lighting conditions with the lowest amount of processing artifacts. In the end, most of the professional quality DSLR cameras deliver files that out perform traditional film captures. Photographers, as a rule however, are never satisfied unless they can exercise their obsessive compulsive tendencies to truly maximize the quality of their digital files beyond what is actually visible in print (I know because I'm a card carrying member of this club)! So...

By deliberately over exposing just enough that the majority of the levels in the raw file are over the mid point, WITHOUT actually clipping high values to white, the file can be post processed down into normal range. Lowering the values in post processing tends to drop the noise remaining in the lowest values BELOW the signal black threshold so that most noise in the image is invisible. For the most part, this exercise in OCD is a lot of work for very, very little benefit as the amount of noise introduced by the typical line  screen (or inkjet diffusion dither) is far more than what is present in most digital camera captured image files – you can't see the difference in print!!!

Having said this, I'm sure there is someone out there who will come forward and declare up and down that processing raw camera files this way is really better and they can see the difference, etc... etc... Much like the flap over 16 bit vs 8 bit image manipulation in Photoshop! High exposure images are easier to post process into a good normal range without introducing extra noise than low exposure images but really, all you need to do is get a good normal exposure for your digital camera system – be happy, its way better than film anyway!

 regards,

Lee Varis
http://www.varis.com
888-964-0024
 ________________________________________________________________________

Date: Mon, 01 Sep 2003 14:08:20 -0400
   From: Terry Wyse
Subject: Re: Is it best to overexpose when shooting digital cameras?

on 9/1/03 11:31 AM, Lee Varis wrote:

By deliberately over exposing just enough that the majority of the
levels in the raw file are over the mid point, WITHOUT actually
clipping high values to white, the file can be post processed down into
normal range. Lowering the values in post processing tends to drop the
noise remaining in the lowest values BELOW the signal black threshold
so that most noise in the image is invisible. For the most part, this
exercise in OCD is a lot of work for very, very little benefit as the
amount of noise introduced by the typical line  screen (or inkjet
diffusion dither) is far more than what is present in most digital
camera captured image files - you can't see the difference in print!!!

So, what we're basically saying is "expose for the highlight, develop for
the shadow" instead of the other way 'round like we'd do for film (negs that
is!)?

Terry

___________________________
WyseConsul
Color Management Consulting
v 704.843.0858
__________________________________
________________________________________________________________________

Date: Mon, 1 Sep 2003 12:56:15 -0500
   From: Bob Smith
Subject: Re: Is it best to overexpose when shooting digital cameras?

On Monday, September 1, 2003, at 10:31  AM, Lee Varis wrote:

Having said this, I'm sure there is someone out there who will come
forward and declare up and down that processing raw camera files this
way is really better and they can see the difference, etc... etc...

I won't go quite that far but I will say that the basic premise works... at least it has with the various Kodak cameras that I use.  Where I apply this is to expose so that if there's any error, it tends to be on the high side rather than low.  I can pull back a slightly light image to yield a much better photo than pushing up a dark one.  Think of exposing for digital much like shooting negs versus transparency.  Expose for the shadows, process for the highlights.  This technique works much better on some cameras than others so test for yourself; and its only really applicable to a raw file workflow.

Its really not that much trouble to work this way.  Most LCD displays on cameras can be set to show highlight clipping and/or a histogram.  Just glance at the display to be sure that you're not clipping critical highlight detail.

Bob Smith
________________________________________________________________________

Date: Tue, 2 Sep 2003 10:01:00 -0700
   From: David Cardinal
Subject: RE: Is it best to overexpose when shooting digital cameras?

Lee--In addition to the extra work, there is one really big caveat to trying to do this. All current D-SLR sensors are linear in their response to light. This means that out of the typical 12-bit (4096 value) Raw data, the "top half" or 2048-4095 only represents a single "f-Stop" of light, or a doubling of light intensity.

That is very little room to work with (the lower half of the histogram can represent as many as 7-8 stops of light, depending on the cameras signal to noise ratio). In particular, since once you hit 4095 not only is there no forgiveness or "shoulder" like their often was with film, but saturation above that can wreak havoc with surrounding pixels (bloom).

In practice, this has meant that many negative film shooters new to digital first notice that their brand new expensive D-SLR is trashing their highlights if they're not careful. Wedding photographers in particular seem to have to work to adjust to the way digital exposures work. Those who previously shot with slide film are more used to this issue, so it doesn't seem to be as big an adjustment.

So, if you have a really controlled lighting situation with a low dynamic range scene, then I could see tinkering with over-exposure. But as a general technique I think the cure may be worse than the disease:-)

Nikon in particular had this problem with the original D1 (Sunny scenes often had blown highlights) so they tweaked the metering on the X & H a little. Of course, that does mean that sometimes people now want/need to compensate back up if they have a low contrast scene where highlights aren't an issue.

--David Cardinal
Pro Shooters LLC
http://www.nikondigital.org
________________________________________________________________________

Date: Tue, 2 Sep 2003 10:28:52 -0700
   From: Richard Chang
Subject: Is it best to overexpose when shooting digital cameras?

On Sunday, August 31, 2003, at 08:45  AM, hfdomke wrote:

I came across a very interesting idea and wanted to know your thoughts on it. The
concept is that the ideal exposure for digital capture is to get your histogram as close
to the right side as possible but not so close as to cause the over exposure indicator
What do you think?
 
This concept has as much merit as always following the incident dome on your light meter.  It cannot work in every situation.  Photographers are vested with an order to keep track of tonality/contrast; the tonality/contrast of the scene's lighting, tonality/contrast of the reflectances to be recorded, tonality/contrast of the file as it is exposed and adjusted, and the tonality/contrast capablility of the paper's rendering surface.

Using a histogram as a light meter involves no real accuracy for accurate placement of the highest important value.  A histogram can't tell you when the exposure is perfect.  It can tell you when the exposure's wrong.  As professional photographers, we can surely do better than a histogram's vague notion of what to do.

If the scene to be imaged is low in contrast, placing the histogram toward the right edge may not allow adequate range adjustment without blowing the highlight.  Yes, you can round the shoulder to avoid blowing the highlight, but that leaves local highlight contrast flat and lifeless.  The solution is to expose less and pull the highlight up to its proper place, thus ensuring adequate local contrast throughout the range of the file.  If this turns your digital capture (Nikon, Canon, Fuji) into a noise ridden nightmare, you may need a better sensor.  Lee Varis' comments previously posted, consider the manufacturing constraints of the lower end cameras whose price point dissallows 4,000:1 signal-to-noise sensors.

The same underexposing tactic can be used when shooting for the newsprint target.  If the highlight on the target is 10 or 12 percent, shooting a file whose highlight is 245 will cause problems.  That's because 245 is about 4 percent; pull 4 percent down to 10 or 12 and you'll have a mess.

The concept of exposing for highlight or shadow is actually: expose for threshold.  For a neg, threshold is the shadow (minimum density).  For transparency (a positive) and for printing (also a positive), threshold is the highlight (minimum density).

Digital photographers are authors of visual content whose message is communicated on the paper we render on.  That paper is critical because different papers are capable of rendering differing numbers of tones (density range).  Knowing the paper you'll print on will allow exposure of the highest important value, where it belongs.  That placement might be lower than the histogram indicated exposure because a subsequent range adjustment may be in order.  This is often called craftsmanship, this ordering of operation that results in professionally rendered images. Exposing appropriately means you might be giving content providers downstream, something better than merely workable.

Emoting your audience is the goal of a good photograph.  While the right edged histogram may be OK some of the time, it's clearly not the thing to do always.

Richard Chang
http://www.TransitionOfTone.com
________________________________________________________________________

Date: Tue, 2 Sep 2003 11:00:35 -0500
   From: Jim Bean
Subject: Re: Is it best to overexpose when shooting digital cameras?

The same underexposing tactic can be used when shooting for the newsprint
target.  If the highlight on the target is 10 or 12 percent, shooting a
file whose highlight is 245 will cause problems.  That's because 245 is
about 4 percent; pull 4 percent down to 10 or 12 and you'll have a mess.

hello richard, could you please expand on this comment.  regards, jim bean
________________________________________________________________________

Date: Tue, 2 Sep 2003 17:03:05 -0700
   From: Lee Varis
Subject: Re: Is it best to overexpose when shooting digital cameras?

David wrote:

So, if you have a really controlled lighting situation with a low dynamic
range scene, then I could see tinkering with over-exposure. But as a general
technique I think the cure may be worse than the disease:-)

Richard writes:

Exposing appropriately means you might be giving content providers
downstream, something better than merely workable.

and

Emoting your audience is the goal of a good photograph.  While the right
edged histogram may be OK some of the time, it's clearly not the thing to
do always.

Hear, hear...As I said earlier:

> all you need to do is get a good normal exposure for your
> digital camera system – be happy, its way better than film anyway!

A good exposure will be different for each subject - determining what that good exposure is requires a bit more than an automatic placement of the histogram. If it was that easy auto-exposure routines would always work and you all know how often auto-anything works well!

regards,

Lee Varis
\http://www.varis.com
888-964-0024
________________________________________________________________________

Date: Thu, 4 Sep 2003 14:22:47 -0700
   From: Richard Chang
Subject: Shooting for specific targets

>hello richard, could you please expand on this comment.  regards, jim bean

Digital capture targeted for a specific paper shouldn't be all that difficult.  Fundamental to notion of making a considered image, is the number of useable tones in your image file and the number of tones the target is capable of rendering.  The key to this discussion is density range.  When our picture file has the same density range that our target can render and when the file's highlight and shadow values match the paper's highlight and shadow values, we have an optimal starting point for the crafting of our file for output.

We can measure our output target with a reflection densitometer.  Choose DENS on your reflection densitomter and zero the densitometer on the non-image white of the paper.  Then measure the blackest ink density.  The value reported by the densitometer will be the density range of the paper. Coated sheetfed stock measures (in my experience) around 1.85.  Coated web is a little lower, Calendared web is down around 1.5 and newsprint can be as low as 1.2. If your capture software can be set to the target's measured density range, it's a lot easier to make pictures for the selected target by lighting for the target's highlight and filling or remapping for the target's shadow point.

The number of usable tones in a file is dictated by how many (we'll use 8 bits and 256 for illustrative purposes here) of the tones in the file are pure white and/or pure black.  Tones that describe pure white and pure black aren't available to describe picture detail, so they aren't useable for making shape.  A tone curve is the tool we use to force tonality to pure white and/or pure black.  Open a Photoshop curve dialog and click the double arrow at the bottom to define threshold (make it the shadow, bottom left).   Move the highlight from the top right corner, straight to the left, move the shadow from the bottom left corner, straight to the right. What you've done is changed the range.  How far you've moved the highlight and shadow dictates how many useable tones are left over in your file.   Is there an optimal number of tones for specific paper targets?  Yes.  Testing your capture by rendering on the paper will help you decide where to shoot.

Once you've decided how many tones to use, you'll want to shape the curve and light your scene specifically for the number of useable tones you've decided to include for your target.  Lighting in concert with your curve is critical and defines your skill as a content provider. If your target holds 7% as its highest separable value, you'll want to consider shooting/exposing your highlight right at 7%

Tethered shooting has typically involved the application of the curve prior to display of the just-captured image.  Shooting untethered in RAW or JPEG obfuscates the process because you don't see what's up until you get back to a computer and apply the curve.  The lower the density range (fewer useable tones) the more the exposure is critical.  Shooting the histogram to the right edge for low density range targets will typically lead to dissappointment in too much range and too little adjacent contrast, especially in the highs.

Remember to test thoroughly prior to making any changes in your workflow. Involve interested parties in your testing; as long as you initiate the testing, you'll be perceived as being on the ball.

Richard Chang
http://www.TransitionOfTone.com  

Adobe Photoshop training classes are taught in the US by Sterling Ledet & Associates, Inc.