On our recent Antarctica workshop we ended up making many panoramic images based on turning our cameras to vertical orientation and shooting a single row of overlapping images. These ranged from simple 3-frame sequences in search of higher resolution, to gigapixel monstrosities taken with medium-format cameras. Usually photographers were mainly looking for a wider view.
When on land we had the flexibility to set up tripods and experiment with multi-row panoramas, but many more single-row panos were made, especially when working hand-held from the moving ship.
I do most of my own stitching work using Photoshop CS5 (usually via Lightroom’s Merge to Panorama feature). I do sometimes use other programs such as PtGui, and some of the workshop participants were using this and other software during the expedition. But in general I’ve found that the stitching in CS5 (and CS4 also) is just so convenient and hassle-free that it’s my first choice.
Salomon Glacier, South Georgia (A2_055680)
I have 3 guidelines I use when taking panorama sequences. You may think you’ve heard them before, but bear with me and read them all:
Modern stitching software is remarkably good at blending the component frames together, but I’ve found that by continuing to use these rules it’s easier to achieve better quality (and with less work):
- Shoot in Manual. Each frame should have the same exposure: the same ISO, shutter speed, and aperture.
Each should also have the same white balance, which leads on to:
- Shoot in RAW. This allows you to apply the appropriate DNG profile and/or white balance values to all photos in the sequence after the fact.
Those are fairly well-known guidelines for shooting panoramas, but there’s one basic one which many people seem to not know:
- When shooting from a moving ship, start at the stern and progress the sequence towards the bow. Don’t simply always work left-to-right: sometimes right-to-left is much better.
The background to this guideline is based on two reasons. Firstly, most of these panoramas will have very little important foreground detail other than water and waves. But features such as small icebergs, birds or kayakers floating on the water, or just wave-tops all fill in important details, and the image is going to look strange if the same kayaker appears multiple times. By moving with the motion of the ship this is much less likely.
But even when repeated details are as innocuous as wave crests, there’s another important reason to avoid having them appear in multiple frames: it can seriously confuse panorama stitching software.
One of the convenient features of Photoshop CS4/CS5’s auto-align function is that you don’t have to tell it where each frame sits in the sequence. You can feed it a mass of frames and it will assemble the panorama by itself. You’d much prefer it simply lines up the mountains in your sequence without deciding that the frames need to be distorted to make the waves line up also!
Danco Coast, Antarctic Peninsula
EOS 7D, ISO 400, 70-200mm/4 L IS (A2_055681)
Of course the same guidelines will apply if shooting a panorama from land but that includes moving elements such as a river: shoot the sequence against the flow.
When shooting a pano I use the above guidelines, and end up with a sequence of frames. Some people prefix the sequence with something like a shot of their hand to act as a reminder, although I generally don’t do this. Once I have the images loaded up in Lightroom my general workflow is:
Identify the sequence. I have two labels (colours) I use for this: “Start of composite” (green) and “Member of composite” (yellow). During my initial editing passes through the photos I simply tag the images, and in a later phase I can pick them out, using Library filters and/or a Smart Collection (I have a “Composites” Smart Collection for this).
Sometimes when shooting a sequence I have to restart the sequence (e.g. due to having to change the focal length). So I sometimes end up identifying new “Start of composite” images within the sequence. But in the end I have multiple series of obvious sequences.
Once I’ve identified and selected all the images within a sequence, I hop over to Lightroom’s Develop module. With Autosync turned on (by flipping the switch next to the “Sync” button at the bottom of the right Develop panel) then whenever I make a change to an image, the same change is applied to all the other selected images. I make all the adjustments required, starting with the white balance, Exposure, Blacks, and Brightness. Often that’s about it, as I have tuned the defaults for my cameras (including having Lens Profiles selected which automatically remove distortions, chromatic aberration, and vignetting).
I’ll switch between Survey (N) and Develop (D) modes, selecting the brightest image to tune Exposure, the darkest to tune Blacks, etc. If I find dust spots that need removing I’ll turn Autosync off and on as required to fix them in all images yet avoid duplicating rocks.
Having developed the images, I invoke Edit In > Merge to Panorama. After selecting Cylindrical mode, I leave Photoshop to grind through the images and do its job.
Once it’s finished and I’m happy with the result, I’ll flatten/merge all the layers together, straighten then crop the final result and save it back to a TIFF file for Lightroom to process. Of course, for those images with more than 65,000 px on a side (unusual for me at the moment) I need to save them to PSB (Photoshop’s Large Document Format) and save a smaller version as a TIFF for “normal” software to handle. Incidentally, the TIFF file automatically inherits the metadata of the last image in the sequence, so by default it will be included in the Lightroom filters for composites.
near Cape Saunders, South Georgia (A2_055679)
Those 9 EOS 5DmkII images (each 5616 x 3744 pixels: 21 Mp) were quickly merged into this 12938 x 5581 (72 megapixel) image without any distortion hassles.
If you haven’t tried shooting multi-frame panoramas yet, don’t be afraid to give it a try. And if you’re on a moving ship, just remember: shoot in the direction of your travel.