Gray Card, Why Do You Mock Me?!?

There are three variables every camera must determine when taking a picture. In fully automatic mode, the camera chooses all three. You, as the photographer, also have the option of controlling one, two, or all three of these values, too. These three variables are the shutter speed, the lens aperture, and the light sensitivity (aka ISO) of the sensor.

Being the control freak that I am, I quickly made the change from shooting in fully automatic to fully manual. My rationale was that if I let the camera make the choices and it chooses poorly, then there is little I can learn from to get better the next time. If *I* make all of the choices and take a poor picture, then at least I can learn from my mistakes and strive to do better next time.

There are hard and fast rules one must understand when setting these variables yourself, but fortunately they are simple ones, too. Every time you double the shutter speed, it doubles the amount of light that reaches the sensor at the back of the camera. Similarly, every time you double the ISO value, the sensitivity also doubles. And, yes, you guessed it, every time you go up one f/stop (the aperture is measured in f/stops), this also doubles the amount of light. So you can see, by creatively doubling one variable and halving another, you can effectively control what kind of picture you will have.

Somewhere along the line, someone with way too much time on their hands decided to analyze thousands (if not probably more) of photographs to determine what the average light reflectivity was in each photo. After what I'm sure was a mind-numbing effort, it was determined that the average light reflectivity was 18%. This meant that of all the light that struck the subjects in the photographs, only 18% of that light was reflected off of them and into the camera. Thus, the light meters built into all cameras from that point forward were set up to expect that every exposure would contain 18% reflected light.

In a perfect world, this would be great. But, as I'm sure everyone out there knows, we do not live in a perfect world. If I were to send back all of the food that didn't meter a perfect 18%, I would be a very hungry food blogger. Fortunately, a relatively easy solution exists to help out with this problem. Most (if not all) camera stores carry what are called "Gray Cards" which is a large cardboard (usually) rectangle that ... wait for it ... reflects exactly 18% of the light that strikes it. The theory goes that if you position the card where your photographic subject will be, fill the LCD screen with the gray card by zooming in, and adjust the three variables until your exposure meter reads "0", you should have a proper exposure. Now the only step left is to remove the gray card and snap the photo at the preset exposure levels and et voila! A perfectly exposed picture.

Hmmm ... don't believe the hype. At lunch today I decided to try this out before my first course arrived at my table. I leaned the gray card against my glass of water and after setting the ISO to 100 and the aperture to f/2.8, I adjusted the shutter speed until the exposure meter marker sat exactly at zero. According to the exposure meter, the correct shutter speed was two full seconds. Okay, a little long given the light levels where I was sitting, but hey, this is SCIENCE(tm)! It is accurate and repeatable! Additionally, the live histogram on the LCD panel had a spike right in the dead center, just as it should.

When my soup arrived, I removed the gray card and placed the cup of soup exactly where my water had been on the table. Here was the shot my camera took:


ACK!! Clearly this looked overexposed on the LCD panel and when I checked the histrogram in picture review mode, my suspicions were easily confirmed. The graph was heavily bunched up on the right side of the graph and spiked up to the top.

Scratching my head, I decided to start reducing the shutter speed until I got a good histogram reading (where it was evenly spread from dark to light) and took another shot, this time with the shutter speed being at four-tenths of a second:


Hey, look a that! It's actually a cup of soup, not the light from the nuclear bomb that just detonated outside the window! I would expect that at the first exposure setting, there might be a tick up or down on the actual shutter speed, but at ONE-FIFTH the original setting ... either I'm doing something horribly wrong or all this gray card metering before taking a "proper" exposure is just bullshit. Strong words, I know.

Honestly, the entire reason for going through this step makes sense to my science addled brain, but with results like this? Well, I clearly hope that I have just made a mistake in the process i used to utilize the gray card.

If anyone out there has any ideas of why this failed so spectacularly, I am open to any and all suggestions. For now, I'm going to continue to base my exposure settings on the live histogram on the back of my camera. While it isn't foolproof, it also has produced the most consistent results to date.

The Curious Case Of The Blown Out Chicken

So after all that work to educate myself on using the live histogram on the LCD panel on the back of my G12, apparently there are exceptions to every rule. If you read my previous post, you will know that the G12 offers three different tools to help you decide whether or not the photograph you are about to take is properly exposed. First would be the LCD panel itself. While using the panel to compose the picture works beautifully, I've discovered that the image that appears on the panel looks way brighter than when I look at it on the computer in GIMP. Panel out.

The second method to judge the correctness of the exposure that I use in Manual mode is the exposure value indicator. This is a scale that runs from -2 stops to +2 stops and has a little marker that moves up and down and you change the exposure options. Until I discovered the power of the histogram, I would routinely adjust the shutter speed until the little marker landed on zero. I assumed this meant I had a properly exposed photo. In good light? Yes. In poor light? Absolutely not.

Through the suggestion of some on-line friends, I discovered the third, and best option, the histogram. Essentially a graph of all light in the photograph from dark to light, I noted that as I lengthened the shutter speed to let in more light, I would systematically drive the graph to the right. I would essentially do this until I had a nice even spread across the entire range. Suddenly, my low light photos started looking really good. I still had to adjust them in post processing, but only a little bit instead of huge swaths of lightening.

So imagine my surprise when I took a picture of my Thai Market Salad at Market Gourmet at Montrose (check out the food blog if you want more info) and when I got around to actually examining the photo on my netbook, I discovered something a little bit disconcerting:


This is generally a good photo. What surprised me was how blown out some of the sides of the chicken cubes looked. Okay, let's check the histogram:


A nice distribution to be sure, but I haven't even pushed the graph all the way to the right. I tried applying a High Pass filter to sharpen up the photo thinking that might bring out some of the detail and while it did improve the photograph overall, the detail still remained hidden. I even tried darkening the photo a bit, but that was a bust, too.

In an effort to understand, I returned the very next day (today, in fact) and decided to not only order the exact same salad again, but fortunately, the exact same table next to the window was available from yesterday's lunch. In fact, after setting the ISO to 100 and the f/stop to 8.0, the shutter speed for the first exposure was set to 1/3 of a second, just like the exposure I took the day before. Of course, the salad looked slightly different, but that's okay. I took four sets of exposures, driven at first by the histogram and then gradually backed down in order to hopefully expose more of the detail with an underexposed photograph.

The first set of exposures, I pushed the histogram. And it showed in the resulting pictures. They were too exposed. The second set were better:


Here was the histogram for this picture:


A nice distribution. Maybe pushing it a bit on the right hand side. The photo reveals definite loss of detail on the cut chicken pieces.

Fine. Dial down the shutter speed slightly. Here was what resulted:


More detail on the chicken, slightly darker exposure. Here's the histogram:


Now we're just barely hitting the right edge of the histogram. But things look better. Note that the exposure value indicator was just above the "0" mark for this one. Of the three, I think this one was the best compromise between driving the exposure using the histogram and what I saw on the screen.

Just for extra options, I shortened the shutter speed one more notch and took another photo:


This one came in right at "0" at the exposure value indicator, which is where I used to shoot at. I would've considered this a "proper exposure" about two months ago, but now I know to push it a bit on the G12. Here is the histogram for this photo:


Looking at the histogram, it looks fairly underexposed (you should know that I'm adjusting the shutter speed 1/3 of a stop at a time, so there isn't a dramatic difference in the resulting photographs). While there is the most amount of detail in the chicken, the rest of the photograph looks too dark.

In the end, I would probably end up going with the second photograph that I took today as the "hero" shot. Even though the histogram indicated that I could've pushed it farther to the right, real world examination of the photograph dictated that it was a good balance between proper exposure and overexposure.

For the purposes of disclosure, the first day I shot simply with a UV filter. The second day, thinking that polarized light might be an issue, I used both my UV filter and my polarized light filter, which did help the images out, in fact. All images, on both days, were shot on a tripod with an ISO of 100, f/stop of 8.0, and varying lengths of shutter speed. Each image was "sharpened" using a high pass filter with a value of 10 ... just to keep it all equal.

I guess the moral to the story is that rules are great, until they need to be broken. A rule that seems to be particularly valuable is to take your exposures at multiple shutter speeds. The brightest one (based on the histogram) might not be the best one.

Using The Histogram To Get Great Photos

Having a camera that gives you total control over the final exposure can be the best and worst thing possible. Obviously having total control allows you to control every single aspect of lighting, framing, and ultimately how the exposure will turn out. The downside is that without the proper training and education on how to use all of the tools at your disposal, pictures, especially like those I take all the time of food in poorly lit restaurants, can come out underexposed and flat.

When shooting in manual mode on my G12, there are two different tools at your disposal for judging whether the picture will be properly exposed. As soon as the camera is set in manual mode, a graduated vertical meter appears on the right hand side of the LCD display. The numbers range from -2 to +2 in 1/3 increments. As you adjust the ISO, shutter speed, and aperture, an indicator moves up and down the scale indicating the exposure value (or level) that the camera "sees."

In a well lit scenario, when the indicator hits the "0" mark, I have found that the resulting image is nicely exposed. However, as has been pointed out to me now by several people (and an on-line photography forum), in poorly lit scenarios, my pictures would come out woefully underexposed. The fact that the LCD display on my camera tends to display the pictures brighter than they appear on my laptop also adds to this illusion that I've captured a good photo.

My first reaction was to simply start adjusting the cameras settings so that I was shooting at "+1" instead of "0". However, this STILL resulted in me having to artificially lighten the photos in GIMP. Frustrated, I turned to the other tool at my disposal, the histrogram. A histogram is a graph whose x-axis represents all colors of light from black to white. The y-axis represents the intensity of that color in the photograph. While the LCD display on my camera displays both the exposure level and the histogram, until just a few days ago, I had only been looking at the histogram as a novelty. No more.

One of the pieces of advice I received when submitting my photographs to the Internet forums was to lighten the photo by pushing the histogram as far to the right as I could. While I didn't initially understand what this meant, it wasn't until I got into a situation where I could practice using the histogram that I finally understood what this meant.

I had lunch in the bar area of a local Fairlawn restaurant recently and I purposely asked to be sat in the corner booth, arguably the darkest spot in the entire restaurant.

As my server dropped off my glass of water, I got my camera prepared to take the picture:


Setting the ISO to 100 and the aperture to f/2.8, I continued to increase the shutter speed until the histogram on the back of my camera looked like this:


Notice how the graph is even an flat the entire way across? With the picture underexposed, the graph tends to bunch and peak on the left hand side. The same would be true for an overexposed photograph, just in the opposite direction.

Next up? Three different shots of my lunch, a fried fish sandwich with chips. Picture #1:


In the histogram, you'll notice peaks at both the left side and the right side:


The peaks are due to the white plate and the dark background. But the more important thing to notice is that the graph spans the entire histogram. It wasn't until I got home and started really examining the above photograph that I really began to believe that this was finally the technique that could set me free.

Picture #2 of my lunch:


And it's histogram:


Again, you see a slight bump on the right due to the intensity of white in the picture above, but with the color distributed across the entire graph, this picture looks properly exposed to me. There are also a few overexposed reflections toward the back right of the photo, which I'm sure also contributed to that small spike.

And, finally, picture #3:


And it's histogram:


I actually took all of these photographs using several different shutter speeds just so I could compare the histograms when I got home and could use my computer to do so. In almost every case, the picture where the graph had been pushed as far right as possible was the one that ultimately looked the best.

Will using the histogram always result in a perfectly exposed photograph? It seems to be looking that way, but I've also learned that taking an extra photograph or two at a notch or two below what I think is optimal is also a good idea, too. Sometimes a more well-lit photograph can begin to lose detail and a less well-lit one can show off the richness of the colors a little better.

I have purposely limited myself to no-flash photography of food. Given this limitation, I have to do everything possible to make sure I still get the best photograph possible given my camera's limitations. The histogram seems to be a powerful ally in accomplishing that goal. If you have access to a live histogram on your camera and you don't already use it, I would encourage to learn how. It can mean the difference between flat, dull looking photographs and ones that really pop.

I should mention that none of the photographs I posted in this entry were altered from their original state with the exception of resizing and cropping them to be Internet friendly.