My first attempt at the Brenizer method: lessons learned
During the two hours spent in the Geo W. Reed, I took the opportunity to try the Brenizer method for the first time with a friend and photography enthusiast as subject. The Brenizer method is the name given to the use of panorama stitching with a subject in the foreground made from individual pictures usually captured with a medium telephoto prime lens. Ryan Brenizer, a well-known wedding photographer, developed this innovative application for an existing technology to create wide scenes with very shallow depth of field. The most common tool for this technique is a 85mm f/1.4 lens which I had hauled around just for that exercise.
I had tested the stitching software on one occasion and it worked flawlessly. However, I had stopped down the same lens so the whole picture was in focus. Additionally, the picture was taken outdoors and without anything or anybody in the foreground.
For this exercise, I had a volunteer to stand in so I opened up the aperture to f/1.4 and included a door frame in the immediate foreground for extra challenge. I took 85 images and the stitching software ended up needing 73. The final size of the stitched panorama is 384 megapixels before cropping. I just stitched the pictures from the JPEG files and the overall white balance does not seem too bad. If I wanted to make sure that every individual picture making up the panorama share the same white balance settings, I can either use custom white balance in the camera (shooting a white card for example) or processed all files from RAW.
To capture all the pictures, the auto-focus (AF) and auto-exposure (AE) features must be separated from the shutter release button. Once the AF and AE are locked onto the subject, I start shooting in a concentric pattern starting from the subject outward. The pictures must be overlapping for the edge detection technique to work.
So here’s the resulting panorama:
Overall, the process is quite fast but a few flaws become readily apparent: there is some ghosting in the out of focus area and the edge of the far door is not aligned properly. Also, the near door frame is heavily curved as is the edge just behind the subject.
Based on this experience, I would avoid including visual elements nearer than the subject due to the heavy distortion induced by the camera angle change needed to capture images that close. Secondly, I would stand further from the subject as I think it would make straight edges more linear. As for ghosting and the incorrect edge, I guess all I can do is to experiment with the software as it has multiple edge detection and blending options.
Next time, for an indoor panorama, I might be tempted to use a flash as well to see if it is possible to obtain a seamless result with additional lighting.
Here is a picture taken with the 85mm at f/1.4 as a single capture for comparison. It is taken from further back than the panorama so it demonstrates the vast difference in field of view between the two images.
All in all, I think it is a very ingenious technique that can be most useful as it offers a way to obtain images not previously possible. I will keep on experimenting with the technique as it seems easy enough to get into but it still requires some refinement from behind the viewfinder.