List of webcams that don't work

Here is a thought why don't we list the ones that do work?

Chris Ward

@Theodore-Borromeo you are correct. I am looking at doing a FOV for a robot I am building. I see lots of people just sticking an android phone on the front. That I feel however is cheating, even though it might be cheaper. And it won't run OpenCV.

I understand where you are coming from too though. I was shocked at the price of web cams. I haven't bought one before. I figured I could find a year or two ago's models for cheap, but it is hard to know if they will work. (hence this thread?)

Maybe what we need is a "cheap webcams that work" thread if that is your goal (I would love to find that $10 web cam myself)

My C270 worked out of the box with the onion platform to stream video (for a rear van camera project). Getting it running on a RaspberryPi to stream was really difficult. (motion and opencv are not really streaming packages, and the one that comes on the onion took a lot of compiling on the pi) So I will just use the onion and the C270 for that project. For the robot, I just ordered the more expensive 'official' camera. I know that opencv is well supported there, and for now, I will just use the pi for the dedicated cam and opencv for the video FOV and face recognition for that project.

Good luck though.

Theodore Borromeo

@Chris-Ward Yeah, I agree that it is 'cheating' for purists/hobbyists, but we can't preclude economic facts for mass production. I mean, even if that weren't the case (since once you get a prototype working, you apply for funding and mass produce with specific BOMs anyway), opencv can be run on android phones. Their arm processors are more than capable, and arguably, it'd be much simpler to do as opencv itself does support android.

But yeah, I do acknowledge that streaming isn't supported by opencv out of the box. My idea is to have a low greyscale resolution and crank up the amount of snapshots I take per second. I'd then run classification over a series of stills to determine distance/faces/objects from a single camera system.

My overall goal is to simulate Tesla's 360 degree sonar ranging system for my own car. I can probably only run one camera per omega, but if I can prove out a system that can do the above, scaling will be the next step

Boken Lin

@Theodore-Borromeo I will include capturing grayscale images in the tutorial.

What kind of price point are you looking at for webcams? If your application doesn't require HD streaming, I should be able to source some fairly cheap webcams from China for you.

Theodore Borromeo

@Boken-Lin Thanks, but I really wanted to compile cheap COTS type stuff for cameras. I want to systematically narrow down the chipsets/manufacturers who are out of the box compatible or close to, just to make it so that older and hopefully cheaper webcams should lower the barrier to entry for this pretty accessible piece of hardware.

Case in point, though the raspberry pi is pretty versatile (and I have already bought an OV* camera module), they don't tend to be as straightforward as a usb plug-and-play solution. On top of that, even if I were to want to get cheap cameras, I'm nowhere near in need of the volume you'd be able to provide

Boken Lin

@Theodore-Borromeo Ok, that makes sense. If I find anything in the lower price range I'll give you a shout!

Does this one work with the omega
see pick below

Boken Lin

@Rudy-Trujillo Unfortunately I haven't tried this particular camera you are looking at. From experience though, earlier Logitech webcams don't support MJPEG output, so it's likely that this camera won't work. What kind of application are you thinking of, and what is your budget? Perhaps I can recommend you a camera that's compatible with the Omega.

Massimiliano della Rovere

@Rudy-Trujillo Just write an email to Logitech asking for specs, I had a problem in the past with a webcam and linux and they were very kind and replied.

It seems that the camera must be UPNP, be an UVC [1] device and support mjepg format for output.
@Boken-Lin are there other limitations?

[1] https://en.wikipedia.org/wiki/USB_video_device_class

Massimiliano della Rovere

http://www.acmesystems.it/video_streaming claims the followingf webcams to be compatible:
Microsoft Lifecam HD-3000
Microsoft Lifecam HD-5000
Microsoft Lifecam Cinema 720p HD webcam
Microsoft LifeCam Studio 1080p HD
Logitech Webcam C250 1.3 megapixel webcam
Logitech QuickCam Sphere AF. Motorised 2-megapixel HD sensor webcam with Carl Zeiss optics.

Massimiliano della Rovere

@Theodore-Borromeo you can try the -y switch for that camera: https://wiki.ubuntuusers.de/MJPG-Streamer/ even though, as @Boken-Lin wrote, it will use lots of CPU power.

Thank you. This project is just to kill time while I wait for the Onion camera to go on sale from the store.

Here it is in all its glory.

Massimiliano della Rovere

There are no specifications on the logitech website:
http://support.logitech.com/en_gb/product/quickcam-notebook

Try using this webcam with (mjpeg) and without (YUV) -y switch with mjpeg-streamer to see if you manage to stream live video

Looking for a "tutorial approach" for this project have to claim ignorance for this setup.

Boken Lin

@Massimiliano-della-Rovere No, I don't think there are any other limitations other than the fact that the hardware needs to support MJPEG output.

@Rudy-Trujillo It will be made available on our store in the next few weeks. Thanks for your patience!

Maurice

Logitech QuickCam Sphere AF works using mjpg_streamer and the webcam app on the latest firmware. Plugged an old one I had and worked out of the box but only when connected directly to the Omega (didn't work when trying it on a non-powered USB hub). Now on to look into controlling pan and tilt.