Company has tech to compensate for crappy cameras
The latest piece of technology originally intended for military / security applications to be rebranded for peaceful, consumer purposes comes from a Silicon Valley startup called MotionDSP, which has licensed technology from a military research project that can enhance low-resolution video and improve the output of phone cameras, webcams, and other low quality feeds. The technology, which will go into a consumer beta later this year, compares multiple frames in a video to replace lost pixels in any given frame without an increase in file size. Eventually the company plans to offer an online service where consumers can enhance their pixelated creations for free, although the company also wants to make deals with current online video services. Until that day comes, we'll continue to suffer through our daily fix of YouTube in all its aliased and pixelated glory.Without MotionDSP.
With MotionDSP.
















Reader Comments (Page 1 of 1)
Jonathan Keim @ Sep 2nd 2006 3:37PM
That's really interesting how they use the pixels from other frames of video to support the missing details! I always wondered how you could add information to a picture from nothing, but it looks like they have something here
Nick @ Sep 2nd 2006 3:54PM
garbage in, garbage out...
Even if they could fix compression issues, these are still cheap sensors with even cheaper lenses.
Erik @ Sep 2nd 2006 5:44PM
I love how you put a frame from David Hasselhoff's "High on Believing" video on the screen.
I also love how odd it is that I saw that.
JCasto @ Sep 2nd 2006 6:39PM
They should make that gun. Except definitely not by Nokia.
Rob @ Sep 2nd 2006 8:05PM
Seems like any technology that aids poor quality video should be welcomed. You can spew GIGO (garbage in garbage out) all day long, but relatively low-end video capture devices are not going away... Why knock something that makes the video from these prevalent devices more viewable? I would like to see this technology become client-based and real-time, so it can be applied to video downloads plagued by transmission/compression artifacts.
Fabre @ Sep 2nd 2006 9:38PM
Now people will think that CSI style technology is real...
ZOOM! ENHANCE! ZOOM! ENHANCE!
Excellent, that single pixel is now a clear enough photo of a face to arrest the guy!
Aaron @ Sep 2nd 2006 11:33PM
Man, I hate the zooming and enhancing that goes on in those stupid crime TV shows. People totally believe it, too.
Anyway, that's pretty impressive. There is a very noticeable difference in the videos. Even if you don't see much more than you did before, it's still a lot clearer and smoother, which makes for a better viewing experience. Well done.
Jacques @ Sep 3rd 2006 12:46AM
SSP Deblocking for DCT coefficients and warp-sharpening will do the same thing. Just play with ffdshow and a poor quality video for a while, and you'll see what I mean.
Hopefully these deblocking algorithms will soon make it into hardware decoders.
Matt Hadder @ Sep 3rd 2006 6:05AM
Will the camera now add 5, 10 or 20 pounds now??
Rob @ Sep 3rd 2006 9:51AM
This will add 3.5 pounds to your camera phone. BTW, ever notice how heavy your camera (or camera-phone) gets when you take a lot of pictures? People don't always realize how heavy data can be -- it really adds up!