Fortunately, webcam image technology has improved,
and so for 2005, the webcam IS real ... although
you would certainly be justified in being skeptical of that! ;-)
So maybe the new hoax is that everyone will think it is fake,
but in actual fact it is real ...
this comment from DIGG says it all:
"This is so fake. Cracks me up those that are trying to pretend that
it's real. Think about it. Really think hard about it.. actually, don't
think hard about it and you can still figure out that it's fake.
I actually did "really think hard about it"
and regardless if I'm doing it for real or not, it's totally doable.
Heck, we sent men to the moon 30 years ago, so it's certainly possible
to make holidays lights turn on and off. What would we do without these
Einsteins on the Internet to set us straight on this technology stuff ...
or at least provide some good comedy. Speaking of comedy, I put togather some
spooky
time-lapse halloween movies and will be adding some
christmas movies
that are entertaining ... and oh yeah,
if this is a simulation again, think how much work it was to make those! ;-)
SO ... join in the fun ... add a link on your website and/or blog
to the
Hulk'in halloween webcam and/or the
christmas lights webcam
and see if people think it is a Trick ... or a Treat! ;-)
2. So if it's really a webcam, how come you don't show streaming video?
While it's not too difficult to let a few of your friends "directly"
view your webcam, that doesn't scale ... and bazillions of people
came by to look at my
christmas lights. Scalability is a challenge, and I haven't figured
out a good way to do this "right" and be able to handle a heavy load.
Drop me a note if you know how to
rebroadcast an MPEG4 video stream from a Dlink DCS-6620G webcam to
a Linux/Apache web server. Right now, it is FTP'ing JPEG's every
couple of seconds and then meta refresh's the web page for the
user to the next image every 5 seconds. So hey, that's not too bad,
and a back-of-an-envelope calculation indicates that this might
be able to support thousands of viewers before the 100 MBit/second
connection saturated ... although the Apache web server would
probably croak before then.
BTW, If you think Frankenstein looks funny being inflated/deflated in
stop-action, it's frickin' hilareous watching it in video as I can do
on my local LAN. Yea, I sure wish I could do this for everyone.
3. When I turn a light on & off (or clobber Frankenstein), it doesn't happen right away or not at all?
In the real-world, there are latencies. When you press that ON/OFF button
on my web page, a request
is made from the web server in Dallas, Texas to the Linux box in my house in
Colorado which then uses a serially attached
CM17A Firecracker wireless transmitter to send the appropriate X10 signal
using the nice Flipit
program from Matt Armstrong.
X10 isn't designed for high-speed applications, so in a best-case scenerio,
this adds up to a 1.6 second delay - yea, I tested it.
Then, the webcam has to shoot the changed picture, upload it to the
image server (a seperate machine also in Dallas),
and then you have to see the final result in the next screen refresh.
For the technies, the ping times from Colorado to Dallas is 32 milli-seconds
and the typical image size is about 50KBytes
and I paid the extra $10 a month for the higher speed Comcast
residential service which has an uplink speed of 768Kbits/second.
All of this can take anywhere from 1-5 seconds more when things are operating
normally ... which is actually the vast majority of the time.
Also keep in mind that during that interval, someone may command the
opposite action. Finally, X10 signals don't always go thru - stuff happens
in the real-world.
4. There is no way your lights are being changed a bazillion times a day!
You are RIGHT. X10 is more typically used to turn your lights on at dusk,
and off at dawn ... so (as alluded above) it is relatively slow
for this intense application.
The Web server rate-limits requests and if more than one request comes in
during that time period (I've tested anywhere from 1-5 seconds), only one is
actually "acted upon" and the others are dropped with a note back
to the user to try again. BTW, this rate-limiting was also implemented
in the simulation because ... well, that's how things actually work in the
real world! Per entry #74
in the Halloween Blog, requests were averaging one every 3 seconds on
October 19th. I suspect this will increase as it gets closer to Halloween.
When I "manually" tested the X10 control code, it actually changed the lights
90%+ of the time ... but when the requests come in hot-n-heavy, my guess is
that some of those aren't going through - there is no logging on the X10
side and it's a one-way protocal, so I don't know ... but I can tell you
the lights blink a lot!
Drop me a note if you know of a powerline control technology that has
milli-second response time, uses a two way protocal, and has 99.99%
reliability ... I'd love to implement it!
5. Is it easier to do the simulation or do it for real?
While the simulation took a bit of coding, once it's done, it's done.
So in some respects, it was easier in that real-world aspects
don't come into play. I've had problems with the wireless signal
dropping out, futz on the webcam lens, general weather issues, inability
of the webcam to optimize exposure & focus, etc. I am fortunate
enough to have great neighbors who don't mind the lights really
blinking this year.
Overall, I'm pleased with how well the whole setup has held togather,
but don't underestimate the challenges of the analog real world.
6. What's this about a campaign to "kill Frankenstein?"
The DIGG story used that term - maybe they didn't realize
that Frankenstein already died ... and even if you "kill" him again
(by keeping him deflated),
one bolt of lightening will bring him back to life anyway.
And just for grins, I brought 'ol Franky back for Christmas.
He'll need a brain, but I'll pass on one from that Einstein who said
to "really think hard about it" - not sure how many
brain cells he has left after all that hard thinking ... ;-)
7. OK Alek, I gotta know, is it really real this time?
It's real ... but HEY, I would be skeptical of anything I say! ;-)
A hint/clue is that
in previous years, I only showed the holiday lights at night ... darkness
is your friend when it comes to image trickery. This year, I turn the
webcam on when there is still light outside,
which is MUCH harder to simulate. Another hint/clue is in the
real world, leaves and trees move, plus with thousand of lights,
there's an occasional light strand than blinks out - didn't happen
in the virtual world. Finally, look at the time-lapse
halloween movies and
christmas movies
- that would be a LOT of work to simulate!
Check out what others are saying and chime in with your two cents
at the halloween blog or
the christmas blog about the
christmas lights extravagnza ;-)
P.S. As mentioned, Halloween is a warm-up for Christmas - not only will there
be a lot more lights, but come back after Thanksgiving to see me take
the technology up a notch!
Update: Halloween had one webcam - Christmas has three ... holy outa-control ChristmasCams Batman! ;-)