Re: video(4) landed. "No webcam driver" is dead.
Adrian Chadd <[email protected]>
| Newsgroups | gmane.os.freebsd.devel.hackers |
|---|---|
| Message-ID | <CAJ-Vmonm6QW+fzHoXzf95UNQkcQUnqJ5DuS3tV2-EJ1k+EV2HA@mail.gmail.com> |
hi! please file bugs at bugs.freebsd.org with the relevant information - camera hardware, dmesg, reproduction steps (especially cut/paste command lines) and screenshots. I need to do the same for our, err, "australia mode" (upside down) images that only happen to me on I think vlc, not with others. -a On Mon, 10 Aug 2026 at 09:03, Fernando Apesteguía <[email protected]> wrote: > > > > On Sun, Aug 9, 2026 at 11:41 PM Abdelkader Boudih <[email protected]> wrote: >> >> For years, the go-to move for dodging a video call was "well, I run FreeBSD, and there's no webcam driver, so." >> >> Camera stays off, nobody questions it, everyone assumes a serious systems person and not someone hiding from a standup. That excuse is now retired. >> By Adrian Chadd, Baptiste Daroussin, and me. >> We did this. On purpose. I don't know what we were thinking. >> >> video(4) is a new kernel-side capture framework: >> a common device model for things that produce frames, instead of every capture driver reinventing /dev/videoN from scratch. >> >> Initial consumers: >> >> uvideo(4): USB Video Class webcams, converted onto the new framework. Most of the hard logic here was already done for us - OpenBSD's Robert Nagy and Marcus Glocker wrote the original driver years ago, Baptiste ported it to FreeBSD. >> fwcam(4): Apple firewire camera. work perfectly for me, but Australian mode (upside down) for Adrian. >> >> Where its been tested: >> >> across basically the whole Apple fleet here : MacBooks, iMacs, Every one of those machines is now missing exactly one thing (WiFi, still catching up on some chipsets). >> Also tested on regular Laptops with integrated webcams, which mostly just worked . >> >> Diffs for the design: D58367-D58369 >> >> What's coming next: >> >> PCI capture cards - the "I have a real capture card, not a webcam". >> SBC-class cameras (think Raspberry Pi camera modules) - CSI-ish targets, for anyone doing embedded/SBC work on FreeBSD >> >> I'd love it if you: >> >> Threw a UVC device at uvideo(4) that has those non generic cameras. (especially 2k, 4k, thermal cameras, ect). >> Have an firewire webcam >> >> Once a few more hackers have tested this, I will post it to freebsd-current@. > > > Thank you all for this! > > I tested it a few weeks ago in current. It worked fine when using Google Meet but it does not work fine with pwcview. > I > tested again in FreeBSD hammer 16.0-CURRENT FreeBSD 16.0-CURRENT #8 main-n288009-d025b84268ec and pwcview (1.4.1_9) does not work properly, it looks like the image is quadrupled in four columns... It's difficult to explain, but I can provide a screenshot if necessary :-). This was run on a laptop with an embedded webcam. > > Similar effect in pwcview for a Logitech USB camera: > > ugen0.6: <vendor 0x046d C270 HD WEBCAM> at usbus0 > uvideo1 on uhub0 > uvideo1: <vendor 0x046d C270 HD WEBCAM, class 239/2, rev 2.00/0.21, addr 7> on usbus0 > uvideo1: 2 format(s), iface_index=1, endpoint=0x81, psize=3060, isoc > uvideo1: default format: pixfmt=0x56595559, 640x480, max_fbuf=2457600 > uvideo1: UVC camera on /dev/video1 > > This camera works fine in Google Meet, it does not with pwcview. > > pwcview worked fine with webcamd, though. > > Please, feel free to reach out should you need more details. > > Thanks! > > >> >> >> As for meetings: the next generation of excuse is already in production. "Sorry, can't turn my camera on, I'm running TempleOS this week, it's between me and God." Untestable & Unfalsifiable. Nobody's going to ask follow-up questions. You're welcome. >> >> Sent from HolyC Internet Missionary.