Open Fridge HDR Exercise

.

I don't know what other men do when their wife is having a "girls night out".  I photographed the fridge - with the door wide open.  It worked best this way because she  doesn't understand me.  The next day her friends would hear - "You'll never guess what I caught Allen doing last night!  He turned out all the lights in the house, got out his camera and opened the fridge up wide open.  Then he stood in the light from the fridge and exposed it over and over - at least 36 times.   When I asked what he was going to do with 36 photos of the fridge he mumbled something about HDR.   ROTFL - I'm not sure what HDR means but the R is for Refrigerator."

By definition HDR means "High Dynamic Range".   HDR utilizes a series of photos taken at different exposures and reassembles them into a photograph that has more detail than any one of the individual photos.  

In the photo above, I photographed my refrigerator with an open door.  I made it more challenging by turning off every other light in the house.  Without using HDR, I would have a choice with detail in the highlights or details in the shadows but not both.  As you can see, HDR was able to capture shadow and highlight details in the same image.

Canon's Series 1 cameras can be set to take a bracketed set of 7 images.  The difference between exposures can be as little as 1/3 stop to well over a stop.  Canon deliberately cripples their other SLRs by limiting them to a bracketed set of 3 exposures.  One of the reasons for this test to see how much a smaller set of bracketed images will degrade the results.

I took a lot of photos - every 1/3 stop for 12 stops.  There were a total of 36 images.  I wanted to see what difference the number of exposures would make in the results.  

The first thing that I learned was that Photomatrix could not handle 36 exposures - at least not on my PC.  It failed with 18 exposures as well  I assembled HDR images that were 3 exposures (1 stop) up to 10 (3 1/3 stops) exposures apart.  Photomatix Pro has several default settings - I made images with the 'Photographic' and 'Painterly' settings.  

The image above was made with the Photographic setting.  The image below was made with the Painterly setting.  If you compare the two, you will see that the Painterly setting has more details in both the highlights and shadows.  Keep in mind that I used default settings.  Photomatix Pro has many other adjustments I could have used to improve my results.

It is possible to simulate HDR using a single exposure.  I processed each of the 36 exposures and chose the one with the best results.    In the image below, the top exposure is a highlight detail of the best of the pseudo-HDR images.  The bottom image is highlights using HDR with multiple exposures. 

The image below shows a shadow area in both images.  The pseudo-HDR image is on top and has much more grain and little detail.

When I evaluated the results I expected the images with exposures 1 top apart to be better than images that were 1 2/3,  2 2/3 and 3 1/3 stops apart.  I was shocked at how similar they looked.  In the 'Photographic' examples below, the top left image is bracketed at 3 1/3  stops and the image on the top right is bracketed at 1 stop.  The bottom left is 1 2/3 stops and the bottom right is 2 1/3 stops.

The image below used Photomatix Pro 'Painterly' setting.   As before, the top left image is bracketed at 3 1/3  stops and the image on the top right is bracketed at 1 stop.  The bottom left is 1 2/3 stops and the bottom right is 2 1/3 stops.  Once again, the results are very similar.

There were some problems as the dynamic range increased between exposures.  Compare the apples in the top image (5 stops apart) with those in the bottom image (2 stops apart)

The Bottom Line

I learned that for this subject, I needed an exposure that captured detail in the highlight and an exposure that captured details in the shadows.  The number of exposures in-between wasn't as important as I thought.  I certainly didn't need more than 7.   If I'm very careful, three may be enough much of the time.

I also learned the pseudo-HDR can be tried in desperation but will not deliver much improvement.

Finally, I demonstrated that HDR was able to produce satisfactory results in spite of the extremely difficult available light.  The only way that I could have solved this problem without HDR would have been to expose for the highlight details and use artificial lights to reduce the contrast in the shadows.