DPS 7-07: The Future of Imaging--Camera 2.0, Beyond HDR & more--Event Update & Product News

  Welcome to DPS 7-07. We hope you had a great summer. We sure did, highlighted by our July trips to Alaska for Brown Bear photography. Next year's trips are already mostly full so if you'd like to join us please sign up soon. Closer to home we've also announced our two Texas bird photo safaris for April and we still have one opening on our African "trip of a lifetime" photo safari in November. In the meantime I had a great time speaking at a symposium on High Dynamic Range (HDR) imaging at Stanford which has given me lots of new energy to re-capture many local scenes using new techniques.

Featured Article: The Future of Imaging--Stanford HDR Imaging Symposium  

 When I agreed to speak at Stanford's symposium on High Dynamic Range (HDR) imaging I thought it would be a small collection of like-minded folks ferreting out the best way to tease detail out of hostile scenes. It was that indeed but the event became much larger in size and scope and was really a look into the future of digital imaging that will affect all of us.

 

Camera 2.0:

Framing the discussion was Professor Mark Levoy's talk about his Camera 2.0 project--an attempt to build a radically new and open camera platform which will allow amazing new features and functions to be be developed, prototyped and delivered by smart folks all over the world. In just the first few months the graduate students working on an early version of their "Frankencamera" built a Viewfinder Alignment application which allows the automatic capture and assembly of panorama images on the fly--with instant feedback through the viewfinder of how each capture relates to the earlier ones and showing the area of the constructed image. Stay tuned for lots more breakthrough applications as the "Camera 2.0" platform hopes to do for photography innovation what open-source software has done for the computer industry.

 

The variety of prototype applications developed by just one class of Stanford students given a version of the new camera platform to experiment with on a cellphone camera was inspiring. Levoy is applying for an NSF grant to be able to seed lots of the improved units to other students and researchers throughout the country. The variety of interesting applications should be mind boggling. Whether this platform and its innovations find their way into the chronically closed systems of the Japanese camera companies is the big question. Let's hope so for all our sakes.

 

The Future of High Dynamic Range (HDR):

Closer to home for users of today's cameras there was a series of talks on the best way to display HDR images. Instead of having to compress their tonal range to fit on traditional displays, new HDR displays will allow the full range of light to dark to be shown right on screen. Today they are all pricey and troubleprone, but event sponsor Dolby Labs of audio fame is already preparing HDR monitors and TVs based on the Brightside technology they acquired.

 

Leopard Cub on Log
I used HDR techniques on a single capture of this famous young leopard cub to get the best rendering of a tricky scene. Too bad it wouldn't hold still for a 'real' HDR series of captures! I'm looking forward to more opportunities like this on our November Botswana safari & on our next one in May, 2011.

 

Of course that won't help with printing, since paper will be harder to re-engineer. So for that the art and science of tone-mapping is an essential element. There was no silver bullet on that score, but plenty of incremental improvements and ideas for innovation. One issue for photographers has been doing tone mapping with a combination of the natural but dull look of compression operators with the vibrant but unfortunately un-natural look of detail enhancement tools. Several researchers were pitching systems which separate the lighting of the image from the subject. Certainly if you can do that then you can factor out the uneven lighting and get a nicely rendered image of the subject. But until those systems are in real products it is hard to know from sample images how well they really work.

 

PhotomatixPro vendor HDRsoft was there showing their research on ghosting elimination. They hint that future versions of their product will intelligently determine what type of ghosting is happening in a series of images in order to more accurately eliminate it.

 

Beyond Today's Photography:

Some of the talks included stunning visual demonstrations, most noticeably one on Image-based Lighting from Paul Debevec of USC where a mirrored ball was photographed to measure the ambient light from all directions in one scene which could then be mapped onto computer generated images making them seem eerily lifelike. This technique was used to help make the computer generated head in Benjamin Button appear realistic. In fairness I'm not sure how that's going to help my personal photography, but it was sure awesome to watch.

 

Many of the other talks focused on how we begin to measure the quality and accuracy of HDR images and how they can be rendered onto regular "low dynamic range" displays. Some good food for thought although mostly still in the research stage. One software tool that is free and available for download is Photonaut, an image editor with an open plug-in architecture including a tone mapper. Several other companies were there with very cool looking HDR solutions but they were looking to license them to camera and software vendors for inclusion in other product, and were not offering them to photographers directly.

 

The Limits of Cameras

I spoke as part of the session on HDR Image Capture. While I focused on the needs of high-end (read D-SLR) photographers there were plenty of different perspectives. The issues I identified as being crucial to solve for more progress were HDR scenes with multiple white balances, HDR in scenes with subject motion, the difficulty of making HDR scenes look natural when using advance tone mapping, and the obvious need for an "HDR Bracket" function which would automatically capture a full set of bracketed exposures of a scene based on highlights and shadows. Some of the Adobe Photoshop team and the HDRSoft (Photomatix Pro) team were in attendance and optimistic that they'd be able to help with the first three problems.

 

The session also included a compelling talk by John McCann, who worked with Edwin Land and Ansel Adams among others, on the history and limitations of HDR and how high-dynamic range scenes were perceived by the viewer. He presented a compelling case that Ansel's use of his zone system really made him one of the first HDR photographers. Other panelists stressed the other extreme of the market--cellphones--and we all debated whether HDR techniques could actually help improve cellphone imaging or users would be better off with less, larger pixels on their phone cameras.