The weather is perfect, the lighting is just right and you’ve just finished shooting a 3 frame panorama of an interesting scene. You’re anxious to get home and assemble the images. So you grab a cup of coffee, load up your images and proceed with the panorama assembly. Then the reality hits you that you didn’t shoot the panorama in manual and the auto-assembled image looks unusable. What can be done to save this panorama? There are probably many programs out there that do a better job with assembling panoramas than Photoshop. But I use Photoshop and have saved many panoramas in the way I’m about to describe.
I made a similar mistake on a panorama of an interesting shot while in Norway. Could I re-shoot it? Maybe. But many compositions are once in a lifetime shots that can never be duplicated. I processed my images and was disappointed with the result. Below is the result of how Photoshop’s auto panorama routine handled my files of unequal exposure. I’ll show how to get a better result than this using a manual technique. Stemming from my astrophotography, I learned how to do manual panorama assemblies which can sometimes salvage shots that can’t be assembled properly by Photoshop.
Let’s get started. Fire up Photoshop and load up your pano images in separate layers. You can do this manually by opening up each image and doing a Select All and the Copy/Paste. But I like to use the script Load Files Into Stack.
When you’re done you should end up with something looking like this, with your pano images in separate layers (lower RHS).
The images then need to be aligned for the panorama. This too can be done manually but using the Auto-Align Layers is generally the easiest and fastest option. It generally does a very good job. Highlight all the layers in the Pano and click Edit then Auto-Align Layers. I usually just choose the Auto option and let the computer do its work. On my antique mobile workstation and with my 5DII images and Photoshop CS5, this takes a while. When it’s done, you’ll probably end up with something that looks worse than what Photoshop assembled. But be patient.
Having the image in layers like this gives us considerable flexibility to manually blend the layers to produce a usable image. You first might make some curves or levels adjustments, to try to match the brightness of the various layers. It won’t be perfect, but the closer the better. You can also arrange the layers changing which layer is on top. This helps to find the best overlap to aid in the manual blending process.
The magic occurs when the various layers are masked to manually blend the image. The secret here is to use the image that is covering the majority of the scene and manipulate the masks so that it blends the various elements of the image, letting through the images below. Add a layer mask to this layer and then invert the mask (so it’s black). Choose a medium sized paint brush and paint the mask in white to reveal the areas of the image below that you want to see. I start with masking the layer that has the most features that most need hiding or blending. It takes a little practice to see what needs to be hidden and what needs to be revealed and which layer is best on top. But try several arrangements and choose the best result. Below is the result of my layer swap.
Here’s the same image that’s partially masked using a small paint brush with soft edges.
Continue to paint the various features to hide and reveal the areas of the image that provides the best blending. I use a smaller brush with soft edges in a jagged path in areas of finer features. I also like to paint little features so that they lie entirely in one or the other frame. On this image, the crane hook is a perfect example. It split the frame. But whenever possible these important features should lie on one frame. So try to blend the image accordingly. After a little work and experimentation, the image should begin to come together. Here’s what part of my mask looks like.
Depending on your image, you’ll need to duplicate this process on several layers to encompass the entire scene. Sometimes I’ll also do a little blurring of the mask, to help blend the masking even more. It’s not always needed but can sometimes be helpful. If you mask too far and hit the edge of the image below you can step backwards or paint it over again with a black mask color to hide it again. If you’re not familiar with Masking in Photoshop, I’d encourage you to do a little research. There are many masking techniques that can be used and help with manually blending the panorama.
When you’re happy with the image, you can flatten the layers and proceed with the rest of your processing needs. The image I used in this tutorial is a custom white balanced IR image. I generally convert these to B&W, add a little contrast, touch-up and complete the image. Here’s the final result of this panorama.
If your panoramas have a larger exposure difference between frames, you’ll need to do more work on the front of the process. When I forget to shoot my pano’s in manual mode, the resulting exposure difference is usually pretty small and this process works well. This is certainly not a catch-all process. But I can typically generate better results with my manual method vs. what Photoshop does with the automated assembly.
I’ve also use this manual blending technique on a very large 8 frame panorama of the Orion Nebula (over 63MP). The automated results were nowhere close to what I wanted. So I had to manually assemble and blend this image. So the next time you have some panorama images that you thought might not be usable, try a manual panorama assembly and see if you can recover the image into something usable. Happy shooting (and processing).