I have a large series of aligned images of an object in motion (camera and background stationary) I want to overlay into a single image to show the object in each frame without any alpha blending.
I could do this in Photoshop by importing each image as a separate layer, and then unmasking the object in each layer, as noted in How do I combine multiple exposures for action shots? But with dozens of images per composite I'm wondering if there is a more efficient method to do this.
There is a simple method to automate the process using Photoshop. It could be recorded as an action or scripted.
Load all of the images into a stack and take the median of each pixel (there is a built in function to do this, might just be in Photoshop extended though). This should give you an image of just the background.
Load up the first image and paste the background image over the top as a new layer. Set the blending mode to difference. Copy the result, paste it to a new layer, desaturate (under the image menu) it then apply a threshold (image menu again) of about 5-10. This should give you a black and white mask of just the object in motion. Create a new layer mask and past the black and white image in. Delete the other layers. You now have a cut-out of the object.
Create a new document and paste in the background. Paste in the cut-out. Repeat the above steps to cut out each image and paste them in turn.
Flatten the result, and you have your montage. It would take some work but if the number of image was constant you could create a Photoshop action to do all this.
Yes - I've done this twice, successfully using two separate software tools. The key was to use software tools that have image registration and alignment capabilities. There are quite a few available.
The first tool I used, and probably the most flexible, was done using a Panoramic stitching tool, PTAssembler (http://www.tawbaware.com/ptasmblr.htm). Specifically, I set PTAssembler to create a multi layer Photoshop object as it's output. One of PTAssembler's plug-ins automatically aligned the images. To make it work well, shoot your subject with a narrow DOF so that PTAssembler selects only points on the subject to use for image registration and alignment. Otherwise you'll have to manually edit some of the automatically selected 'control points' used to register the image. You'll need to use the PTAssembler setting that accommodates for a moving camera location - called 'Camera Position Optimization'(CPO). You'll need to use CPO, since, relatively speaking, your camera can be considered to be moving in respect to the subject.
There are other Pano tools that may work. For example, you may be able to do this with Hugin, a popular open source Pano tool, but I haven't tried it.
The second software tool I used was one that was developed for 'super-resolution' - PhotoAcute. (http://www.photoacute.com) In this case, I used the older version (version 2.94) that allowed me to save intermediate aligned images. I then stacked those aligned images in Photoshop. This method allowed for more refined 'morphing' of the images to get all of them aligned.
On a side note, depending upon how many images you need to align, what OS restrictions you have, how quickly you can learn new software tools, your budget for the tools, and so forth, you may be better served by manually aligning them in Photoshop. From a level-of-effort perspective, I'd say, if you only have 3 to 4 hours worth of manually alignment, it may not be worth learning new software tools. If you have 20 plus hours of manual alignment, it is definitely worth the investment to learn one of the new tools.
Let us know your decision and please share your results.
V/R
B Shaw
I read an article by a snowboarding photographer. The answer was faster than software methods. A very potent and high cycling flash unit, assuming the background is too bright and too distant to be affected. Expose for the background adjusting for the total exposure time. Adjust flash power to brighten and freeze the subject. Keep the shutter open while the flash unit strobes away. Properly exposed background and multiple subjects all in one frame -in camera! time spent setting up, but not in post process. This was in an ad/article promoting for some strobes. (Not talking speedlites here, four digit price. They were shooting snowboarders.
