This image represents my second successful attempt to photograph the full arch of the Milky Way (out of three attempts-the first attempt at this same location was a complete failure. What a difference four years of experience makes!). In this post, I want to share how I optimized capture of the raw images by applying knowledge of my camera's ISO invariance and how the raw images were processed using Sequator to improve the final image quality.
ISO invariance is a buzz-phrase in the digital camera world these days, but I think most people have not yet grasped how to make use of ISO invariance in practical situations. To break it down simply, increasing ISO reduces the ability of the camera sensor to capture highlight data. For astrophotography, we can preserve highlight detail by using a lower ISO then boosting the overall exposure afterwards in post-processing. Based on my previous testing of my camera, I found that ISO 1600 seems to be a good option for maintaining dynamic range (i.e., highlight detail) while still providing a decent in-camera exposure (not too wildly underexposed). Without this understanding, previously I would have been shooting this scene at ISO 6400 thinking I was getting a better exposure at the expense of increasing noise a little.
If you want to learn more about how to improve your astrophotography by taking advantage of ISO invariance, Justin Majeczky has an excellent video, and Hal Mitzenmacher has a nice post about his experiment.
On this early June evening in southern Colorado, dark night started at 10:12 pm. The moon was a waxing crescent, 38 percent illuminated, and at the time the raw images were captured, was about 30 degrees above the horizon directly behind me. It took about 11 minutes to capture the entire sequence of 32 images, so by the time I finished, the moon had dropped about 4 degrees closer to the horizon (and a little while later it disappeared behind the range of 13,000 ft mountains that I was standing on.) The setting moon was in a great position for lighting the foreground and landscape in my composition, but it was a little bit too bright, so the Milky Way did not appear as prominent in the sky. On the other hand, I was able to capture foreground and sky in a single exposure; the foreground had to be darkened significantly in post-processing. (Side note: you have to make the most of the conditions. I initially planned to go out a couple of nights later with a friend because the moon would have set by the time dark night arrived. But the forecast called for clouds and storms for the next 7 days, and indeed, this was the only clear night that week. Also, the shot had to be taken right after astronomical twilight before the Milky Way arch rose too high in the sky.)
I hiked up to the ridge about two and a half hours before dark, partly to get some sunset and twilight photos, but mostly to avoid hiking through the forest in the dark (I figured I could muster the courage to hike back to the car in the dark since it was the only way to get home.) I had plenty of time after sunset to get the pano rig set up, figure out how many frames I needed, and determine how much to rotate the camera between shots. I had decided beforehand that I wanted to try stacking the images to see how the image quality could be improved.
As soon as it was dark enough, I captured a couple of 8 or 9 frame pano sets to make sure I had a good set of images to stitch together in case the stacking didn't work. For the stacked set, I captured 4 exposures for each frame. The wind was howling at my back, so I had to use my body to block the wind from the camera. Even so, on the last set of frames the camera was almost blown over and I had to grab a couple of extra exposures after I got all three legs of the tripod back on the ground. Then I quickly packed up and headed down off the ridge before I got blown off of there myself.
Sequator is a free application for stacking astrophotography images-it's essentlially the Windows alternative to Starry Landscape Stacker. Sequator is fairly easy to use and offers some powerful features for separating the foreground from the sky while stacking. The website offers detailed instructions and provides some tutorials; however, the author of the program is a Taiwanese software engineer so some of the explanations seem to have lost something in translation. Since this was my first experience with Sequator, I experimented with some of the options and wanted to share what I found to be the best settings.
The overall process include four main steps: pre-processing the raw files, stacking multiple exposures to create individual panorama frames, panorama stitching, and final post-processing.
I made the following basic adjustments to the raw files in Lightroom before stacking the images in Sequator:
You can see in the comparison below that the sky in unadjusted raw file is too dark. The foreground landscape on the other hand is just about right for a night photo and had to be darkened again in a later step. After adjusting the first raw file, the same settings were applied to all 32 exposures, then the images were exported as TIFFs.
Sequator does support stacking on raw files, eliminating the need to export the pre-processed raw files to TIFFs. This would probably be the best option for properly exposed images (not intentionally underexposed by setting a low ISO in camera) or for single-frame, non-panorama images. In this case, I wanted to brighten the exposures on the raw files, not the stacked outputs, and I wanted to apply lens distortion corrections for pano stitching, so I exported all of the exposures as TIFFs before stacking.
The screenshot below shows the Sequator settings used to process the individual pano frames for this image. The inputs are the four exposures to be stacked for each frame. For the base image (used as the reference for stacking the sky and foreground), I selected the last exposure as the base image for the first frame and the first exposure as the base image for the last frame to try to minimize movement between the first and last pano frames. I'm not sure this is important or that it even works, but that's how I did it since I was stacking for a panorama and the Milky Way moved significantly over the 11 minutes that it took me to capture all of the exposures. In my haste to pack up and get off the ridge and out of the wind, I forgot to capture a dark frame for noise reduction, so I didn't specify a noise image. For the outputs, I just named the files 1.tif, 2.tif, ..., 8.tif for later stitching.
For Composition, I selected "Align Stars" with the "Freeze ground" option so that Sequator would separately stack the sky (moving) and foreground (not moving). If you separately exposed for the sky and foreground, you could select a different option. For the Freeze ground option, you must mask the sky area. This does not have to be exact (the giant brush is the only option), but it is important not to select any of the ground in the mask. This step required maybe 30 seconds per frame.
Since this was my first experiment using Sequator, I wanted to see how some of the options affected the output stacked image. In particular, I was curious about the effects of High Dynamic Range and Enhance Star Light. I processed the same set of four exposures three times with both options off, HDR on, and both HDR and Enhance Star Light on. I didn't see a huge difference in the results, so I decided to process my images with HDR on and Enhance Star Light off.
I did not use "Auto brightness" or "Reduce light pollution" since the images were part of a panorama and those settings would likely be different when applied to different images. The other important setting is the output color space. The default here is sRGB, but I changed it to AdobeRGB since it is a larger color space. Unfortunately, ProPhotoRGB is not an option.
With the selected options, Sequator took about 14 seconds per frame to stack 4 exposures. After I had decided on which options to use, it only took about 20 minutes to process all eight panorama frames (32 exposures in 4 stacks).
The image below shows the original raw file compared to the stacked image. Although the Sequator output is darker, it has almost no noise in the sky and foreground, and the stars are just as bright. Compared to one of the single exposure panoramas that I captured earlier that night, the stacked image looks much better.
Lightroom had no issues stitching the eight frames of stacked exposures processed by Sequator. The stitched panorama is 70 megapixels, and the final cropped image is almost 45 megapixels.
I processed the stitched pano in Lightroom, focusing on trying to enhance the Milky Way and darkening the foreground. As you can see in the uncropped pano, the shadow of my tripod and camera are quite visible, so I took the final image into Photoshop to remove the shadow.
Overall, I'm very impressed with the capabilities of Sequator-I think it will be an essential part of my astrophotography workflow whenever I have multiple exposures to stack. I am also happy with the quality of the raw files produced by the a6500, especially compared to the night images I was getting from my a6000 previously. Unfortunately, I am disappointed with the final image. The Milky Way is not as prominent as I would have liked, probably because of the brightness of the moon even though it provided great light for the foreground. And the West Spanish Peak seems too diminished even though it dominates the view from that ridge.