One of the fun things about Microsoft life is hanging about on the various discussion channels for technology. Someone asked a great question with regard to taking a giant image of an entire crowd and having good enough resolution to see individual faces.
Bill Crow from the Seadragon team had the following to say: (BTW, check out the Seadragon site for sharing high-resolution imagery and of course Photosynth.net for 3D constructions and ICE for panaromic imaging solutions if you have similar needs – they may well fit the bill.)
While gigapixel images may be technically possible with a single exposure, it’s not very practical. It would require huge and expensive lenses to deliver that resolving power.
There is a project that captures gigapixel images on film using large format. While they produce gigapixel files, it’s questionable as to whether he’s resolving that level of detail.
Seitz makes a scanning camera that can take a 160 megapixel image, but this requires a window of time to scan and capture the image. Better Light has similar technology packaged as a digital back for large format cameras.
Modern medium format cameras can capture up to 60 megapixels in a single frame. Again, this requires some very expensive lenses even for that level of resolving power. Manufacturers included Phase One, Leaf, Hasselblad, Mamiya and others.
Everything listed above is insanely expensive gear.
The far more practical way for shooting gigapixel images is an automated pan/tilt head and a conventional camera. It can even be done, with some patience, with a manual (preferably geared) head.
The brand new Gigapan Epic Pro is ideal for this. It provides a very affordable solution for shooting extremely high resolution images.
Here’s a 3.21gigapixel image I shot with the Gigapan Epic Pro, a Canon EOS 1DS Mk III and a 70-200 f2.8 lens: http://photosynth.net/view.aspx?cid=2dbd60b0-8bc1-442e-8069-fcf122a40e24
Clearly there are complications when shooting a photo that make take 10 minutes or more to complete. But there still isn’t any accessible technology that makes this practical in a single shot.
There are also lots of straightforward and very effective techniques to create large group shots in segments. Smaller groups are shot on the same set of seats or bleachers, and then composited together in post. They’ve been using this approach since the earliest days of school class pictures shot on film.
One other reality check. Let’s assume you need a minimum of 64x64 pixels to make a recognizable face. When you have a bunch of people standing or sitting, even at very tight density, I’d be surprised if faces occupied more than 20% of the viewable image. So, to capture faces of a group of 80,000 people, I suspect you’ll actually need a minimum of 1.6 gigapixels! (64x64x80000x5) That would assume ideal positioning and no wasted space around the edges, with people arranged to exactly match the camera frame. Since that ideal configuration is highly unlikely, your requirements would most likely exceed 2 or even 3 gigapixels. You’re just not going to do that in a single frame.
I’m also curious how you get 80,000 people to all look at the camera, smile and not blink all at the same time!
Fun fun fun if you’re a photography maven!