Envisioning the VR World of the Future

I've put together some different ideas about things I'd like to see in the VR world.

First, affordable full-frame digital SLRs with about 20 megapixel image sensors will speed up post-production VR processing by providing equivalent or better than film performance with no scanning required. Of course to process those giant RAW files, and stitch them into something resembling a panorama, we'll need 5-10 gigahertz PCs with a few gigs of RAM.

Second, a few terabyte shoebox archive of every photo I've ever shot. Every panorama too. That's how much storage I need today to hold all my panoramas and the images they were built from.

Third, HDR imaging should become the norm, maintaining a full 48 bit image throughout the panoramic workflow.

Fourth, the entire panoramic workflow needs to honor color profiles. Why bother tweaking the color of images at scan-time or when converting from RAW image format when a stitching tool ignores the profile and produces a panoramic image with no color profile?

Fifth, all platforms (on which people would want to view VRs will be able to playback panoramas out of the box. Photographers should not have to spend time creating twenty different formats of a single panorama just because this player works here but not there, etc. It should be as easy and universal as JPEG viewing is today.

Sixth, recording of sound snippets along with images. The resulting soundtrack would be incorporated into the VR if desired.

Seventh, GPS location metadata automatically embedded into the VR workflow.

Eighth, automation of the post-production process is critical for wider adoption. Professionals have the time and motivation to scan film or tweak RAW images, monitor stitching, wait for the stitching to finish, manually override bad stitches, adjust final the panoramic image in Photoshop, build a movie file with associated hotspots or DRM, and build a webpage around the final product, but the average person is not going to put up that much time and effort to build one scene. It's got to be as easy and not more than a few times longer as dealing with regular single shot photography.

Ninth, panheads the size of today's manual click-stop models having motors embedded to control x y and z movement to enable automatic capture of a spherical pano.

Tenth, all the photo management tools (iPhoto, Photoshop Album) should have plugins that enable automatic stitching of panoramic sequences.

Eleventh, panheads for shooting spheres will have level tolerances one tenth the level accepted today. My understanding is that one degree of wobble is acceptable today. But have you tried to shoot a cubic with every image rotated by a different one degree of wobble? I think panhead's manufactured tolerance should have no more than 0.1 degree of wobble.

Twelvth, all stitching software should come with a database of all known lenses, and like many software packages which check for updates over the Internet, should download new lens metadata automatically from the stitcher maker. Why should all photographers spend big chunks of their valuable time calibrating all their lenses in every new stitching program? Think of the collective waste of energy compared to the manufacturer doing it once.

Stitching of cubic or spherical panoramas could be made much easier with advanced software. Imagine dragging a set of images into a stitcher and have the software figure out the arrangement of the shots and lens type, then building a complete VR, unattended.

But these process improvements are easy to extrapolate from current trends. Other developments are perhaps less obvious:

A lot of photolabs have a checkbox option for delivering your photos on CDROM, why not add a checkbox for automatic building of panoramas?

A "laserpod": Integrating a laser pointer with a camera or lens's nodal point, so that when you want to shoot a panorama, the laser light is activated -- and illuminates the center of rotation point on the ground. In the uber-version, the laser is a rangefinder which gives you continual feedback on what elevation to hold the camera steady at. There might be less need for a bubble level as well (if the beam was guaranteed to always point perpendicular to the central barrel of the lens).

I envision the day when a good digital camera senses you are shooting a panorama and provides audio or in-viewfinder feedback to help shoot it level and with the right amount of overlap; then stitches the images in-camera so that when you get home, the panoramic image is merely one of many images you have to transfer to your multi-terabyte archive.

Cellphones will be able to capture and playback VRs--why? well it could be a cool thing since many people carry a cellphone with them at all times but usually not a camera.

Stereo and time-based VRs would be much easier to shoot and build.

I would like to see the capture process take less time than shooting a cylinder, not more. If you use the same lens for cylinders as for cubes, it takes about 2-3 times longer to shoot a cube. For my purposes, that's too long. So everyone's going to start using fisheyes to shoot cubics, which is fine but the problems with processing fisheyes need to be streamlined.

So there you have it, a few ideas for how our small world of VR can get bigger and better.

Last update: 1/6/06; 2:41:54 PM.