A Surprisingly Tricky Subject

Stitched panorama of stitched panorama at the enttrance to the Rinpung Dzong
Camera: Panasonic DMC-GX8 | Date: 13-11-2015 11:21 | Resolution: 1920 x 1440 | ISO: 1600 | Exp. bias: 0 EV | Exp. Time: 1/13s | Aperture: 5.6 | Focal Length: 12.0mm (~24.0mm) | Lens: LUMIX G VARIO 12-35/F2.8

This really shouldn’t difficult. The image above is from the entrance to the Rinpung Dzong, in Paro. It’s a series of pictures of Buddhist deities which have been painted on bits of cloth, glued to the wall, and joined at the seams. Yes, it’s a stitched panorama. So why was it so difficult to make a photographic stitched panorama of it?

I set the camera to appropriate manual settings (to make sure that exposure was constant), faced each panel in turn, and when no-one was in shot took a picture. I then developed the JPEGs with exactly the same settings in Capture One. There was good overlap between the images, and as by definition it’s a series of images with a visible seam it ought to be straightforward to stitch images back together.

There seem to be two main challenges. Firstly as far as I can see all automatic stitching software assumes that the camera is roughly static, whereas I was shooting in a long thin tunnel, and moved the camera to face each subject. This is a well-established shooting technique, but seems to have minimal software support. Second, despite the manual exposure the resultant images vary significantly in brightness, and it looks like the camera was doing some measure of adjustment for the tricky lighting.

In the end I "went manual", importing the pictures as four layers in PhotoShop Elements, hand tweaking their position and geometry using free transforms, and then using the technique of painting the layer masks to choose exactly which elements of each image are visible in the final shot. The result isn’t perfect, but probably "good enough". In the meantime I’ve bitten the bullet and invested in some new stitching software which is supposed to cater for "moving camera" combinations, and we’ll see if it can make a better attempt.

View featured image in Album

Leave a Reply

Your email address will not be published. Required fields are marked *