In the previous two posts, I covered shooting and tone mapping the 36 exposures that make up my now not-so-recently posted landscape. In this post, I’ll breakdown the stitching (the easy part), fixing, and final cleanup, including moving some cows and the final color correction pass in Lightroom. I’d like to note that I didn’t actually take screenshots while I was making the original image, so I’ve been basically recreating the image a second time just for these posts. Any minor discrepancies will be due to that.
Once I have all the HDR patches ready, I send them over to Photoshop for stitching. For panoramas that I intend on being displayed as a single flat image (i.e. not an interactive QTVR or anything), I’ve found Photoshop’s stitching tools to be quite capable… assuming things are shot correctly. From Lightroom, I select the 12 HDR images in question and right click > Edit > Merge to Panorama in Photoshop. This launches Photoshop and presents me with this screen:
In this case, I go ahead and set it to auto, and turn on Vignette Removal and Geometric Distortion Correction. This begins a rather time-consuming process requiring a lot of waiting. At this point, I’m on to beer number two, Founder’s Devil Dancer; a 12% Triple IPA that is hands down my second favorite IPA ever, right behind Dogfish Head’s 120-minute IPA 😉
Above is what Photoshop gave me. Not too shabby. (Do note that the first view of this image usually has the seams visible. This is just a graphical side-effect. If you zoom in, you won’t see these seams.) Before I get to cropping, I’m going to do some further distortion correction using Adaptive Wide Angle. I can’t use it across multiple layers though, so I’ll go ahead and flatten the image. This will turn my background (currently transparent) white, but as I’ll be cropping and doing all kinds of repairing, this isn’t an issue for me.
For safety, I’ll duplicate the now flattened layer and do the distortion on this new copy in case I want to go back to the original. Opening Adaptive Wide Angle on this new layer, I tell it that I’m correcting for a panorama. This will make certain assumptions about how the image is being distorted. The first and obvious thing here will be to correct any distortion on the horizon. I’ll go ahead and draw a line across the horizon, and once I complete this, it’ll straighten things out. I can rotate the line and, holding shift, snap it to be perfectly horizontal. I made just a couple other adjustments, which were drawing a couple lines on the power lines to remove some of the curvature in the sky—might as well get some use out of these before painting them out. Hitting Ok will apply this to the image, which again may take a little while. *Sips beer*
Okay, I’m not sure how to explain a lot of this in text with static screenshots. It’s at this point I regret not just recording all this as I worked. It’d be much more useful that way. Some of these screenshots were recreated for the blog, others were just shots of various layers in the actual PSD.
The first thing I would do here would be to crop off some of the excess empty spots, but I want to save those cows so I’ll get them on their own layer first. A simple marquee selection and Cmd + J (Ctrl + J on PC) to copy the selection to a new layer does the trick. Since they’re going on top of a hill at a different angle, a bit of rotation and warping is necessary. Then I simply blend them in with a layer mask and soft black or white brushes to paint portions of the layer away or bring bits back where necessary. It takes a careful eye, but most of the grass blends pretty easily. I keep the stuff immediately around the cows to help them feel like they belong there.
Now I can crop the image since the cows are safe on the hillside. I didn’t crop with any particular size or dimensions in mind. It was purely based on how much extra space was around the image after the stitching process.
I use the magic wand selector to grab the white areas and use Edit > Fill, set to Content Aware, as a preliminary step.
Fixing that Sky
Ultimately the problem with this sky is a failure on my part during shooting. I was interested in the cows so I ended up cheating things down lower and and apparently failed to get enough of the sky on the left. Not to worry though, since I was shooting so much that day. This is where that extra exposure comes in that I mentioned before. I found a shot that seemed like it’d work, tone-mapped it, and then opened it in Photoshop so I could drop it in. First, though, I did some broad paint work just to get a clean sky to work over.
Some of the other things I cleaned up were things like buildings on the horizon I found distracting and taking the dodge tool to the particularly dark areas of the clouds. Below are a before and after showing the the cleanup that was done.
Back to Lightroom
The final look of this photo would be determined in Lightroom. Luckily, all adjustments are non-destructive, and there’s a history that can be cycled through at any time. I created a timelapse (in 4k) below of each adjustment being applied. The main thing I want to point out is how I handle the highlights. This is the key to a natural-looking HDR in my opinion. The purpose of shooting HDR, for me, is an attempt to capture what I saw with my own eyes. And let’s be honest: despite our ability to see much more dynamic range than our cameras can capture, the sun and brightly light clouds are too bright for us to to look at. So why should and HDR photo have perfect detail in these areas?
So one of the early adjustments I do is to basically crank the highlights all the way up, allowing the sun to appear night and bright, as it should. I’m still getting more detail in these areas than I would with a standard exposure, but they glow in a very natural way. The rest of the photo has a surreal, life-like range of tones that you would be able to see, while the sun is just as it should be.
The rest of the adjustments are very fine tweaking of the blue of the sky, the color of the sunlight shining on the treetops, and the overall warmth of the image. As you can see from the screenshot of the history and the timelapse, I push things back and forth until I find what I’m looking for. It’s a very experimental stage for me, and the part where I finally feel like I’m finding the photograph, whereas nearly every step thus far has been a technical process of putting the pieces together.
And here’s the final image! You can click here to see it in its full-res glory (it’s about 45 MB). I hope this series of posts was interesting for at least some people, and I extend my apologies to anyone who was waiting for part 3. Life just got in the way, and I’m a terrible procrastinator (or would I be a good one?). Next time, I’ll definitely just opt for a video of the process 😉