Snow!
December 8, 2005
I am trying out an experiment. If things go according to plan, I’ll have a timelapse movie of tonight’s (forecasted) massive snowfall. At the same time, the image up above should update with the current captured frame. We’ll see how that goes.
Oh, and yes, that is the best image I could get without a bunch of CCD noise. And no, I don’t have an easy way to remove the screen, and the only window I have without a screen (the doorwall) doesn’t have enough light outside of it for one to be able to see the snow. Ah well. :)
If you want to view this on it’s own page refreshing every 30 seconds or so, visit http://www.nuxx.net/webcam.
actually, removing CCD noise is easy, if you can fix the exposure of the camera.
Take one image with the lens covered. This is the “flat field” image.
For each image you take after that (with the same exposure), subtract the flat field from it. It’ll eliminate most of the noise.
You can usually do much longer exposures with much cleaner results.
If you leave it like it is above, be sure to either store the images non-compressed, or if jpeg, at like 90% or better quality… otherwise the details in the dark, very low contrast image will get lost in the jpeg artifacts.
(Yes, i’ve worked with stuff like this in the past; time lapse with a webcam at night, as well as the flat-field subtraction for astronomy work. :] )
Hehe. Well… That’d require me to write something to do that, not just fire up EvoCam. And I don’t know how. ;(
Also, I think that wouldn’t quite work for the noise, because if one watches the image with the gain turned up it actually sparkles. Maybe if I had the camera outside…
Ha ha! Evocam is sweet. I was actually demoing a customer it today at work. He owns a bunch of Dairy Queens around town and wants to use a Mac for the POS terminals and an iSight/evocam combo for security.
I made a time lapse video the other day from my window here: http://www.orbhead.com/timelapse/Dec%202%202005.m4v
My evocam is running here: http://webcam.orbhead.com using the built in java applet.
Hehe, neat. Just FYI, I think he’d probably be better off buying some Axis cameras that directly connect to the network. No need for a PC which could die or crash, they can either be accessed directly, aggregated, or just told to publish their photos somewhere. And all for about 2x to 3x the price of an iSight. (read: lots less than an iSight + Mac).
Thanks for the info! I will look into it.
From what I have read so far, the software client seems to be Windows only, although it did mention a web based interface also that might do the trick.
I’ve used their web-based stuff via OS X plenty of times. Works great.
With regards to the other, I was meaning the stop-motion footage. If you watch the seconds counter, it seems to cycle through numbers.
That’s cool then. I will definately check it out.
As for the time lapse, it was recorded at 1 frame every 5 seconds. It seemed to be a good balance. 1 frame per second dragged far too much, and 1 frame every 10 seconds missed out on some action shots (pedestrians, squirrels, blowing snow) making things a little too chunky.
Watch the seconds counter in the vid. It appears to increment other than I would expect for every five seconds.
Hmm, yeah, it looks like it offset about one second once every ten seconds or so. I didn’t notice that.
Hehe, yeah. Just kinda odd… It’s still nifty to watch.
I’m doing one still image every 30 seconds, then I’ll just combine all of those in Quicktime. (They are stamped so they should combine easily)
Cool. I look forward to seeing it!
It started snowing here about an hour ago, just starting to get heavier here now.
Heh, I hadn’t realized you lived so close. Yeah, you’ll be getting this, I imagine…
By the way, what was that? Every 10 seconds? If so, there is some skew in the time.
Actually, it’s a real time stream. The only thing holding back the refresh is available bandwidth, local traffic etc. From a local machne, it’s about 10fps. The machines at work get about 1 frame per second. A lot of factors come into play.
Fun experiment!