A Panorama Example

Last year I produced a first attempt at an instructional video. Admittedly, the video is pretty amateurish, shot with an iPhone on the fly. Still, the video has been seen by 25,056 as of this writing, and has generated more than a fair share of comments and criticism (mostly because I blah-blah-blah too much!). But I think the eyeballs are attracted because of the video’s content, namely how to shoot panoramas.

In making the video I was so distracted I actually missed taking a few shots in the sequence, which spells death for a pano, since you cannot just clone in large sections of images. Well, I guess you can, but it would never look right. When I discovered the error of my ways (or means), I was plenty upset. But I was saved by the fact that I had taken another pano for practice before I made the video. One of the things I teach in my workshops is to try to get at least two panos of the same scene to cover yourself in the case of screwups.

I thought I’d try to accomplish two things with this blog. First, show what a pano looks like taken from the scene of the instructional video, although about 100 feet back from the exact spot of the video. Second, to give you an idea of how I post-processed the pano.


First some data. The pano consists of 78 images taken in 3 rows of 26 images each. They were taken with a D800 at 36MP per image. You can imagine how long it took to assemble the image, even though I have a screaming fast pro desktop! Second, I am in the midst of several deadlines, so I did not devote a whole lot of time to post-processing, as you’ll see. The majority of the time was spent waiting for files to process and to stitch them together. Third, I would never have been able to do this without the help of my assistant, Bob, who graciously agreed that we needed to get this done since people have been requesting this for a year!!

This is image as soon as it was done being processed and stitched together. It is right out of the camera and saved as a jpg. You will notice several things wrong about it right off the bat. The software (PTGUI) did not do a perfect job stitching the horizon on the right. Next, as I took the series, the clouds shifted, so that there is a dark patch in the stitched image about 1/4 of the way onto the scene from the left. One can’t always compensate for that when shooting a big pano. You have to pray you can fix it in post.

In this second iteration, I fixed the horizon in Lightroom by cloning three sections to smooth it out. I did not obsess over it for this demo, but it is fairly good. If this were a client image, I would obsess… a lot.

In this image I painted over the dark patch of grass with a brush, lightening the area with the Exposure slider by .2 and .3.

I also did a very slight tone curve adjustment in the BASIC panel to increase contrast, and added a sliver of Clarity.

For this “final” image I added some contrast to the sky between the two rock piles, to bring out the clouds a bit. Don’t forget that you really can’t use polarizing filters when doing panos, so I knew from the start that I would have some sky adjustments to make in post.

… And a Few Tidbits

A few more quick points. When shooting a pano you will often encounter issues with clouds and sky. Clouds move, sometimes very quickly, but these can be patched during post. Water also poses challenges, but can often be smoothed in post. Most important, if you take too long, as I did in the above pano, exposure can change dramatically, leaving you with significant post work, if you can save the image at all.

One thing I recommend is that you shoot your first panos in small jpegs, so that post-processing is dramatically shortened. Once your technique is etched into muscle memory, you can set up your shoot and proceed through it quickly and smoothly in RAW.

That’s about it for this demo. I promise in future instructional videos to be more succinct. But, in any event, I really do appreciate your critiques and the lively comments and discussions that result.

Happy shooting!

blog comments powered by Disqus