Lori Carey Photography

Monday, December 30, 2013

Experimenting with Dstretch to enhance Alabama Hills petroglyph images

Alabama Hills petroglyphs enhanced using Dstretch

A few years ago while doing research for a trip to the Alabama Hills at the base of the Sierra Mountains, I came across a hand-drawn map that showed the locations of petroglyphs I had never before seen mentioned. One of the locations was along my planned route so I made a point to stop and search for them. We scrambled to the top of the rock formation and began searching, and searching, and searching. We didn't find anything so we began working our way back down and around the formation, searching as we went along. We still hadn't found anything when we were joined by a couple guys who were happy to join in the search. I was about to give up when they convinced me to climb back to the top with them. After another 20 minutes of searching my husband finally yelled out "Found it!" and sure enough, barely visible now that the sun was sinking and the light was hitting at angle we found the first dragonfly and spiral. Motivated by the finding we continued our search and eventually spotted three dragonflies on three separate rocks, each with a spiral or concentric circle. They were so old and faded that it was extremely difficult to get decent photos of them. They hardly showed in the photos at all, which wasn't surprising considering how difficult they are to see in real life. So the files were left sitting on my hard drive.

A few weeks ago a blog post by fellow desert explorer Daren Sefcik reminded me that several years ago I had obtained a copy of Dstretch, a specialized software for enhancing photos of pictographs. I never used the program much because it is intended for photos of pictographs, not petroglyphs, and I haven't found many pictographs. It does a decorrelation stretch (by applying Karhunen-Loeve transform, according to the website, mathematical computations that are beyond my knowledge level) to the colors of the digital image. It creates false colors to make the pictographs stand out from the background, making them easier to isolate. While it has a purpose for scientific research and documentation I struggled with the wild and crazy colors for my purposes. But Daren inspired me with his post showing how he uses layer masking to get rid of most of the wild colors and I decided to see what it could do with my photos of the Alabama Hills petroglyphs.

Alabama Hills petroglyphs enhanced using Dstretch

My results were a bit mixed. Dstretch worked better on some images than on others. The problem with using Dstretch on petroglyphs (rather than pictographs) is that there are no painted colors to enhance; petroglyphs are carved/etched/scratched into the rock so at best we see only lighter areas of rock. Relatively young petroglyphs carved into varnished rock are easier to see, but the petroglyphs I found at Alabama Hills were carved into unvarnished granite and were very faded.

Dstretch isn't an easy program to use and I think if I understood the advanced capabilities better I might be able to achieve better results. I am also using an old version of the program (I think I obtained it some time between 2006 and 2009) and I believe that there is a newer version with more advanced options.

Alabama Hills petroglyphs enhanced using Dstretch

In all but one case it definitely made the petroglyph easier to see. All of the examples in this post show the original photo, the photo after using Dstretch, and the final image using layer masking to remove most of the crazy colors from everything except the petroglyph.

Alabama Hills petroglyphs enhanced using Dstretch

Daren mentions on his blog that using layer masks this way means the images are subjective, subject to my interpretation of them. In fact I simplified his method of masking by simple stacking the Dstretched version on top of the original, adding a layer mask and manually masking out everything except what I interpreted to be the petroglyph. I think it's a workable solution to being able to visually share the petroglyphs, but it is by no means scientifically accurate.

The one image I had the most difficulty with was one of the dragonflies (for some reason I only photographed two of them!). This one was in bright sun and the guys tried to create enough shade with their hands and bodies so I could photograph it. The bright spots on the edges are the areas they couldn't block. This section of rock was the most exposed to the elements and you can see how much damage has been caused by erosion. Although we were all certain this was another dragonfly I had a very difficult time trying to isolate the petroglyph using Dstretch because there wasn't enough color variation between the carved petroglyph and the eroded rock surface. This was the best result I was able to achieve and as you can see I wasn't even sure what areas to mask out and what to leave in.

Alabama Hills petroglyphs enhanced using Dstretch

I also had a tougher time with this one. I believe we may have thought this was a third dragonfly, but Dstretch brought out what appears to be part of the engraving that we didn't originally notice with the naked eye. I can't say with any certainty that is an accurate depiction.

Even though Dstretch is not intended to use on photos of petroglyphs my results were good enough for my purposes that I will probably give Dstretch a try the next time I photograph old and faint petroglyphs. If you're interested in learning more about using Dstretch I recommend the tutorials on Daren's site as a great starting point.

No comments:

Post a Comment