Natural color and/or IR camera shields. There are some really interesting and cheap tiny cameras on sparkfun that appear to be small enough to easily work in TinyDuino projects. Possibly even some TinyLily projects as well.

I can think of many uses for such a shield, such as surveillance applications and vision processing experiments.

Good idea. This could even be used for guiding a mini robot, like mounting it to the head of a delta robot for part recognition.

I would like to second this one. The idea of a small shield (potentially with a ribbon connector for proper placement http://forum.tiny-circuits.com/index.php?topic=114.msg158#msg158) that could be used to guide a small UAV or rover. I would really like some sort of visual sense for a robot powered by the Tinyduino platform.

Hmm, how much CPU do we need to drive a camera?

This one uses I2C for commands and has an 8-bit parallel output bus for data. The datasheet looks fairly straightforward to use, but I'm not sure if the microcontroller is fast enough to handle live video data at full framesize. https://www.sparkfun.com/products/8667

Here is an example of one that claims to be operable by an Arduino, so shouldn't have data transfer speed issues. It's a full module though, you'd likely not use it as-is (maybe use as a starting point, or at least connect via ribbon cable to a shield). https://www.sparkfun.com/products/11610

It would be nice to be able to get into the IR range with a camera as well

Most cameras can detect near IR. For example, a typical phone camera will show a purple glow from a remote control.

I was thinking of making a thermal IR camera cheaper than its current $2000 pricetag.

Thermal IR is something much different and much more difficult. It also doesn't help that the high-quality tech is tightly regulated, difficult to source, and the prices are held at an inflated level.

Lots of discussion here on it:

http://www.eevblog.com/forum/crowd-funded-projects/m-thermal-imager-real-or-fake/

looks like they are making something available with indiegogo ... Mu Thermal Camera. http://igg.me/p/312023. They are in Chicago. Would be nice to make it small... looks like it wont be tiny though.

Umm, exactly. That thread I linked has extensive discussion about it.

In a nutshell, there is no way in hell to produce a thermal camera in the same ballpark as their target specs while staying remotely close to their stated price point. The cost of the optics alone is enough to ensure that even the most modest design will still need to sell for over $500/unit to break even.

I was an early backer of the Mu Optics project on IGG, but asked for a refund not too long after funding ended. It was clear that they could not possibly succeed and were in WAY over their heads. Even now - three months past the supposed delivery date - they don't even have a partially functional prototype. Thermal imaging is not a novice-friendly technology.

If you were to add IR LEDs around the outside edge of a dirt cheap IR camera you would get a decent night-vision Pic/Video...More IR leds = better(brighter) night-vision.

(might be pointing out the obvious here :/)

I'll add my vote. How many would that be?