- baseband NTSC output;
- color;
- extremely good in low light (1 lux or lower);
- pan, tilt, zoom, and focus controllable via a simple, documented serial port protocol.
Now, in my experience, there are only two varieties of cameras that are both color and good in low light: there are high-end pro cameras, which are typically $5000 or more; and there are Sony Nightshot 8MM camcorders, which are $250. (You'd think that a camera would be cheaper than a camcorder, since the recording mechanism with all its moving parts must be the most expensive piece to manufacture: but economies of scale and price points are involved, so your simple earth-logic does not apply.) Sony Nightshots are what I use for the club's webcams (though I don't run them in infrared mode.)
If a camera doesn't mention a lux rating, that almost certainly means it's 7lux or worse. Just about all digital webcams are terrible in low light, and X10 cams are terrible unless you're actually standing on the surface of the sun.
So I think what all this means is, I want a dingus that is a pan/tilt camera mount, and that can also control an arbitrary camcorder that is bolted on to it. I believe most (all?) Sony camcorders can be controlled by a protocol called Control-S and/or SIRCS and/or LANC and/or S-Link. (I suspect that those are four names for the same thing, but maybe not.) I gather it is basically a digital version of the IR remote control blips-and-bleeps.
The stuff on videobotics.com looks promising, though it's really expensive. Here's a summary. They sell a few things that look like the might work, like CamMotion PTU and (maybe?) CamMotion PTB plus Bescor MP101.
But I can't really tell whether any of these will actually do what I need. Which ones are computer-controllable and which ones only work with a physical joystick dingus? If they are computer controllable, do they truly require closed Windows software, or some Java/ActiveX nonsense, or will I be able to find a way to drive them from the serial port of a Linux box using a Perl script?
I can't tell. It all looks pretty damned pricey, though.
Then there's this Slink-E thing, which is a box with a Control-S plug on one end and a serial port on the other (but no servo motors.) Will I end up needing one of these too (and two serial ports on the controlling computer?) I don't know...
Surely one of you knows about this stuff? Gimme a clue...
Speaking of toys, are you still in the market for a stand-up arcade game? Because i have a working Millipede machine that i'm sick of looking at.
Actually, I have a Millipede. (And a Star Wars.) I wouldn't mind having a Tempest, though...
I went through this whole mess when I was trying to make FolsomCam work, and I decided that just about everything out there had the problems that you list above.
X-10 sucks, we know that for sure, and their motor controllers are slow as shit. Everyone else wants big $$$ to make a pan/tilt system, and the DIY solution seems so easy that the small amount of work to make go is just worth it.
What I decided on doing was very DIY, but simple enough to work without any soldering. I purchased two stepper motors, a power supply, and a stepper motor control board from steppercontrol.com (here: http://steppercontrol.com/a100.html). About $60 total.
I took that mess and mounted it onto two pieces of plastic (from TAP plastics). You basically get two pieces, one that looks like an L and then a second piece that looks like a U. You assemble these into an X/Y mount that'll screw into your tripod.
The torque of the steppers is 124oz/in, which is enough to move a reasonably large object (like a camera) around. The real problem is finding the right screws and hardware to attach the steppers to the plastic parts, as well as the bolt to fit a standard camera mount.
You can easily write a piece of code in linux to make all this go, and get full X/Y control over the unit. You'll have to figure out the stepping sequence of the stepper coils, but that's usually on the datasheet for the stepper. You'll use 4 bits of the parallel port for each motor.
I figure that you probably don't want to go this route, in which case you're probably going to have to shell out the cash for a prefab controller. Personally, I ended up aborting this project halfway through because I cooked my stepper card with by miswiring it. ugh!
I have a feeling I know what jwz's going to say about that approach...
... but thanks for the link. I want about a gross of those.
I pointed you to relatively cheap cameras.
or drop the $20 on the tiny 0.5 lux cam here and see if it will do the trick.
http://www.supercircuits.com/STORE/home.asp
get a DMX (lighting protocol) scanner head, slap a camera on that, then get one of the many PC DMX interface cards that has an SDK and write your own interface. I'm sure some of the lighting companies have an old light with an intact head that has some other cost-prohbative repair needed for it to function as a proper light. I've also seen, but wasn't able to find off hand, self-contained DMX X/Y Mirror heads.
It still doesn't meet your other requirements, but for the record it is pretty trivial to get NTSC composite output from X-10 cams. Open up the case, cut the yellow wire, splice into the black ground wire, attach an RCA connector.
As I understand it, it's taken as a given that you can get NTSC out of an X10 camera, it's just that the cameras, motors, and control software all suck.
you say dingus a lot. i like dingus.
Kind of a cheesey DIY idea, but do they still make those computer controllable robot arms for kids? Could they hold the wieght of a sony nightshot, as is or if you removed all the recornding equiptment from it? Probably more work that it's worth, but it would let use the existing cameras, and hopefully save $$$.
Maybe worth a look...
Cannon VC-C4
Claims 2 lux but I'm sure the quality wont be so good.
http://www.dperception.com/ptu_over.htm
Simple to use from any RS-232/485 host computer
Fast and accurate camera positioning at low cost
Precise control of position, speed and acceleration
Can network multiple pan-tilts from a single host computer port
ASCII commands for simplicity, binary commands for speed
http://mill-shaf.com/hardwaredoc.html
http://www.kintronics.com/neteye/PTZ%20cameras.htm
Though this stuff looks to be $pendy
Having just spent another two hours chasing links on those (and other) sites, I am no closer to having identified a setup that does the things I want.
At any price, let alone a reasonable price.
Some of the dome-enclosed cameras looked promising until you get to the fine print buried in the PDFs: "0.8 lux at 1/2 second exposure."
seriously, what about the one I mentioned: http://www.dperception.com/ptu_over.htm
it's a camera mount. you put any cam on it you want. you can talk to it in ascii or binary. shoudl eb a trivial fron-end to code. I think it's like $1000 though.
alas, it's not color, but jeebuz man, 0.0003 lux??!?!?!!?
http://www.supercircuits.com/STORE/prodinfo.asp?number=PC164C&variation=&aitem=2&mitem=2
oh, this pan/tilt system has a wireless IR remote, which could be controlled via IR transmitter on a PC.
http://www.supercircuits.com/STORE/prodinfo.asp?number=PC179PT&variation=&aitem=1&mitem=17
There are also some pan/tilt mounts that use controller boxes. any switches could probably be modded so they are controlled by serial or paralell output from a pc. Especially if it's just simple "keep panning left while button is held down" controllers.
It's very common for black-and-white cameras to have insanely good lux ratings.
ptu_over seems cool for a pan/tilt, but it doesn't do Control-S, so I'd still need to find a way to do zoom/focus. Maybe the only way to accomplish this is to talk to two serial ports at the same time (using a Slink-e as well), but I'm still hoping there's an integrated device that does both, plus is serial-controllable. (I've seen combinations of two of the three, but not yet all three.)
PC179PT is certainly economical, and 0.4 lux is very good, unless they mean at 1/2 sec shutter speed like that other camera. I don't know, because I can't find an actual spec sheet on this camera, only the one paragraph on that page.
Trying to get some Rube Goldberg PC IR transmitter-based setup working sounds like Pure Hell. And I do not have the electronics skill to tear apart a controller box and trigger a microswitch from a serial port. I understand this kind of thing is trivial for some people, but not for me. Just eyes, I just do eyes.
re: controlled zoom/focus I've seen standalone lenses that do that, and the cheap ones are like $900
re: PC179PT
http://www.supercircuits.com/STORE/returns.asp
re: gutting a control box, talk to JNA.
re: PC IR
how is that different than coding for pc lan? it's just serial communications. it's only semi-related, but I know there are programs for plam pilots that turn them into IR remote controls. Some people use those to send the magic Service Tech codes to certain DVD players to clear the region settings.
as long as you got a bidirectional IR box for your pc, all you'd do is capture the sequences for each button, then wire a little interface to send out a series of captured IR sequences for certain commands. which basically turns the pc into a programmable universal IR remote control. if you put the IR diode right near the other IR reciever, you could probably just encose them in something, and guarantee no interference.
http://www.midondesign.com/CIR-II/CIR-II.html
this just works off raw parallel port data. i.e. 2 pins for send, 2 pins for recieve.
more PC IR stuff
http://www.epanorama.net/links/irremote.html
http://www.armory.com/~spcecdt/remote/remote.html
oh look, Xremotecontrol
http://people.freebsd.org/~fsmp/HomeAuto/Ir.html
Linux IR control.
http://www.lirc.org/
I noticed something about a lot of the remote cam/video=server systems.... the camera and video server were two different units. so I dug in a bit, and found that this one
http://www.axis.com/products/cam_bundles/ptz_kit.htm
uses analog coax feed from the camera to the server. The server is still required, since that's how you talk to the camera, but the server has analog I/O, meaning you just take a feed off of it.
My guess is this is pretty common, as there would be need to integrate into traditional analog CCTV systems.
Then I dug more, and looked for just the camera
http://www.theimagingsource.com/prod/cam/evid30d31/evid30d31.htm
has serial inteface, IR interface, remote pan/tilt/zoom/focus/exposure, up to 6 preset locations, Motion Detection, composite/s-video out, and "subject tracking". alas, it seems to be only 7 lux. price: 1095.00
Manual:
http://www.theimagingsource.com/prod/cam/evid30d31/evid30d31_man.pdf
so, on all those network cameras, dig deeper and look into the individual components. It seems most of them are stand-alone cam controlled via serial interface by video server. check this out http://www.nuspectra.com/vbcam/ they have a full working online demo, and they're based in alameda. you can control the camera online. it's using this camera http://www.canondv.com/vcc4/s.html
which claims "6 lux (2 lux at gain-up mode)". Again, serial and IR control. nuspectra sells ust the camera for $895.00
Since they're in alameda, you could probably arrange a trial use of one to see if it meets the lux needs.
You just keep telling me things I already know! "This has to be possible -- here are some things that are almost -- but not -- what you're looking for." I've seen dozens of multi-thousand-dollar 7lux cameras controllable with a proprietary box already.
And there's no fucking way I'm going to try and do this with an IR card in a PC. It cannot possibly be any less sucktastic than dealing with serial ports, and that's bad enough -- I do not need that level of pain in my life. I want to find a device that is already intended to be used this way: if I have to wander off into the "custom hardware hack" weeds, I'd rather just not do it at all. That shit does not sound like fun in any way shape or form. No matter how easy you keep telling me it is.
which is why I pointed out http://www.canondv.com/vcc4/s.html
which claims to be able to do 2 lux and has Serial interface. (software that works with that cam here. that place in alameda has them, it has everything you want except sub-1 lux. hell, you might be able to talk them into bringing you their demo unit to test it's performance. 5 lux handles most DNA lighting situations, and I'm guessing that what this is all for.
you realize that at really low lux, the ability to distinguish color gets worse and worse, hence why low-lux cams are monocrome.
I don't think you will find a sub-1 lux Color camera at 30 fps anywhere. or if such a beastie exists, it will cost multiple thousands of $$. Every "low light color camera" I've seen is a dual-mode camera, color in high lux, monochrome in low lux.
Sony Nightshots are 1 lux in color without flipping the switch for monochrome infrared mode. Those are what we're using today, and they are just barely good enough. So 2 lux would not work, let alone 6 lux.
are you sure about that?my sony has nighshot, and it's 5 lux in color mode.I think the 1 lux is nightshot mode, without activating the IR LED spotlights.here's a frame I lifted from video shot in the lounge against the East Wall with my 5 lux cam, arguably the darkest area of the club depending on the dance-lights and bar lights. and it was rather dark that night.
I can't find the manual online, but this table says that the CCD-TRV99, which is what we have, are "0.7 Lux / 0 Lux" ("0 lux" being their bullshit way of saying "infrared with built in IR spotlight".) The IR spotlights come on by default: if they were making a distinction with/without spotlights, there would be three numbers.
huh, go figure...
I think next time I'm at the club, I'd be interested in seeing how my cam matches up against those, shooting the same subject.
How about http://www.trackercam.com - this is a USB controlled PTZ camera mount - use it with the camera of your choice. If it corresponds to Visca sequences you can then use Netmeeting with the ETree software http://www.etree.com/tech006.html
I've used the ETree software to control a Polycom Viewstation from a PC with a $30 webcam - worked well.
It's only "PTZ" if you use a craptastic digital USB camera. Those all suck, as I said.
It could maybe be modified to be merely "PT" with a normal camcorder, but that's not enough.
Well, this isn't exactly what you're looking for, I know, but if you decide to go the camcorder and servos route, try to get a 3CCD DV camcorder. Lowlight performance is /insanely/ good in full color, because each of the 3 CCDs has light gathering capacity equal to the entire CCD in a standard camcorder. I have a TRV-900, which goes for about $1000 on ebay now. It has a shutter speed/lens aperture control as well, so you can actually adjust such things as needed. I've done event photography of local bands before with it and my PBG4 to produce a demo DVD with great success. But it is $1000ish used on ebay, so it would be a major cost commitment. (For more than you ever wanted to know about lowlight performance, check out http://www.bealecorner.com/trv900/dvfaq.html#Most-frequently)
If you decide you want a servo control system, check out the Acroname Brainstem. It's designed for robotics, but it has four servo control ports on it, runs off a wide range of voltages and has a serial protocol that doesn't totally suck (or you can download programs into the PIC processor onboard written in their kinda-java-C-ish-sorta-language called Tea (filled with so many bad pieces of terminology derived from "tea" it's not even funny, like naming compiling "brewing"). That would require thought and mechanical engineering time commitment you probably don't want to give, but if you do decide to go the "build your own control platform" route this is my favorite servo controller.