Re: [libdc] Flea3 on OSX 10.9 failing at dc1394_capture_setup()
Capture and control API for IIDC compliant cameras
Brought to you by:
ddouxchamps,
gordp
Menu
▾
▴
Re: [libdc] Flea3 on OSX 10.9 failing at dc1394_capture_setup()
|
From: Damien D. <dd...@do...> - 2014-03-06 03:40:11
|
Jim, Benedict,
Thanks a lot for your debugging work on USB3. I don't have the hardware
to help but I'm definitely following this closely. I've already pushed
the two new USB IDs in the repo.
Damien
On Wed, 2014-03-05 at 09:27 -0500, Benedict Brown wrote:
> This is very helpful Jim. Thank you. I will test these things out,
> hopefully next week, and see what happens. My Flea3 is 3MP, and I need to
> drive it at full resolution, so I'm curious if that will work on USB 2
> (which would be just fine--frame rate is not a concern). I think we're
> seeing the same problem of maximum buffer size in the mac USB 3 driver that
> exists in the Linux one. If we can figure out how to make those buffers
> larger, higher resolutions should work on USB 3 as well.
>
>
> On Tue, Mar 4, 2014 at 12:28 PM, jcassidy <jca...@no...>wrote:
>
> >
> > Here is an update on what I have found regarding the Flea3 USB camera on
> > OSX in case it saves
> > others some time.
> >
> > 1) Got it working at both 1280x960 (the default value) and 1280x1024
> > (requires cmd seq & usb2 hub)
> > 2) There are two firmwares for this camera: 1.32.3.2 and 2.7.3.0 and they
> > behave differently
> > and have different USB Interface numbers and USB product numbers!
> > Here is my usb_products
> > table:
> >
> > { 0x1e10, 0x2000 }, // Point Grey Firefly MV Color
> > { 0x1e10, 0x2001 }, // Point Grey Firefly MV Mono
> > { 0x1e10, 0x2002 }, // Point Grey High Res Firefly MV Color
> > { 0x1e10, 0x2003 }, // Point Grey High Res Firefly MV Mono
> > { 0x1e10, 0x2004 }, // Point Grey Chameleon Color
> > { 0x1e10, 0x2005 }, // Point Grey Chameleon Mono
> > { 0x1e10, 0x3000 }, // Point Grey Flea 3
> > { 0x1e10, 0x3005 }, // Point Grey Flea 3 (FL3-U3-13Y3M)
> > { 0x1e10, 0x3006 }, // Point Grey Flea 3 (FL3-U3-13S2C)
> > { 0x1e10, 0x3008 }, // Point Grey Flea 3 (FL3-U3-88S2C)
> > { 0x1e10, 0x300a }, // Point Grey Flea 3 (FL3-U3-13E4C)
> > { 0x1e10, 0x300b }, // Point Grey Flea 3 (FL3-U3-13E4M with 1.43.3.2
> > FW)
> > { 0x1e10, 0x3300 }, // Point Grey Flea 3 (FL3-U3-13E4M with 2.7.3.0
> > FW)
> > { 0, 0 }
> >
> > Firmware 2.7.x puts the 0x81 endpoint in interface 2, which breaks usb
> > capture.c. Firmware
> > 1.32.x puts the 0x81 endpoint in the normal interface 0 position and
> > requires no changes
> > to capture.c. My camera came with 2.7.x, so I did a firmware switch.
> > Both are on the PtGray
> > support site.
> >
> > 3) When the camera is initialized and frames grabbed, they are only
> > 1280x960. To get 1280x1024,
> > the camera will need to be switched to Format7 mode 0 and the offset
> > and frame size and
> > packet params set. I have the code to perform these steps in
> > ObjectiveC if anyone is interested.
> >
> > Note: This switch to 1280x1024 will not work with USB 3. The bulk
> > transfers will receive
> > no data. HOWEVER, putting a USB2 hub between the camera and mac makes
> > everything work.
> > I.e. initializing the camera and performing the switch to Format7 mode
> > 0 will produce proper
> > 1280x1024 frames. I am using libdc1394 2.2.2 and libusb 1.0.18 under
> > mavericks on a new
> > mac mini.
> >
> > Turning on the libdc debug messages, this is what I see for USB 3
> > connection after applying the
> > switch for Format 7:
> >
> > libdc1394 debug: Mode 88, 1280x1024, packet size 10004, packets per
> > frame 132
> > libdc1394 debug: usb: Frame size is 1320528
> > libdc1394 debug: usb: Helper thread starting
> > libdc1394 debug: usb: Bulk transfer 0 complete, 0 of 1320528 bytes
> > libdc1394 debug: usb: Bulk transfer 1 complete, 0 of 1320528 bytes
> >
> > But, just drop a USB 2 hub between the camera and the mac and the same
> > code produces:
> >
> > libdc1394 debug: Mode 88, 1280x1024, packet size 6560, packets per
> > frame 200
> > libdc1394 debug: usb: Frame size is 1312000
> > libdc1394 debug: usb: Helper thread starting
> > libdc1394 debug: usb: Bulk transfer 0 complete, 1312000 of 1312000
> > bytes
> > libdc1394 debug: usb: Bulk transfer 1 complete, 1312000 of 1312000
> > bytes
> > libdc1394 debug: usb: Bulk transfer 2 complete, 1312000 of 1312000
> > bytes
> > libdc1394 debug: usb: Bulk transfer 3 complete, 1312000 of 1312000
> > bytes
> >
> > And we have video frames. So I now have 1280x1024 as long as I use the
> > USB2 hub
> > to downgrade the connection. Looking at the usb prober output there is a
> > slight
> > different between how the camera is viewed based on its connection, but I
> > am not a USB guru.
> >
> > USB 2:
> > Interface #0 - Vendor-specific
> > Alternate Setting 0
> > Number of Endpoints 1
> > Interface Class: 255 (Vendor-specific)
> > Interface Subclass; 0 (Vendor-specific)
> > Interface Protocol: 0
> > Endpoint 0x81 - Bulk Input
> > Address: 0x81 (IN)
> > Attributes: 0x02 (Bulk)
> > Max Packet Size: 512
> > Polling Interval: 0 ( Endpoint never NAKs)
> >
> > USB 3
> > Interface #0 - Vendor-specific
> > Alternate Setting 0
> > Number of Endpoints 2
> > Interface Class: 255 (Vendor-specific)
> > Interface Subclass; 0 (Vendor-specific)
> > Interface Protocol: 0
> > Endpoint 0x81 - Bulk Input
> > Address: 0x81 (IN)
> > Attributes: 0x02 (Bulk)
> > Max Packet Size: 1024
> > SuperSpeed Endpoint Companion
> > Endpoint 0x82 - Interrupt Input
> >
> > Regards,
> > Jim.
> >
> >
> >
> >
> >
> > On 3/1/2014, 11:37 PM, Benedict Brown wrote:
> >
> > I recently had some success with the flea3 on linux (not yet on mac).
> > It turns out you have to add a kernel boot option to increase a maximum
> > buffer size in the USB driver. The info is buried in a PtGrey document
> > here:
> > http://www.ptgrey.com/support/downloads/documents/TAN2012007_UsingLinuxUSB3.pdf
> > .
> >
> > Once I added the necessary boot parameter, I could get full resolution
> > and frame rate in coriander. I'm using libdc 2.2.1 with the Flea3 USB ID
> > added, and the stock libusb and coriander packages from Min 16/Ubuntu 13.10.
> >
> > It's entirely possible there is a similar issue underlying the problems
> > on the mac, although I don't have any sort of solution or workaround to
> > suggest. I have yet to get the flea3 working on the mac myself. Still,
> > progress is progress.
> >
> >
> > On Mon, Feb 10, 2014 at 10:49 AM, Benedict Brown <bj...@gm...>wrote:
> >
> >> I've been working on and off on getting a flea3 to work as well. I get
> >> the same error as you, I think. The latest versions of libusb do support
> >> USB3, but I'm not sure the problem lies there. I run into the same
> >> problems with FlyCapture on linux, so I think the problem is at the driver
> >> level.
> >>
> >> I can get the flea3 to work in linux as long as the over all packet size
> >> is below some threshold, but that limits me to roughly 1.5Mpixels in RGB8.
> >> Again, I hit the same wall with both FlyCapture and libdc1394. I can also
> >> replicate it on linux distros, multiple USB chipsets and both USB 2 and USB
> >> 3 ports. On the other hand, FlyCapture on Windows will give me data at
> >> full resolution even over a USB port.
> >>
> >>
> >> On Mon, Feb 10, 2014 at 10:35 AM, Rodolphe Pineau <pi...@rt...>wrote:
> >>
> >>> I don't think libusb supports usb3 yet.
> >>> I don't have a usb3 camera so I can't really look into this.
> >>> You might want to ask in the libusb-dev list.
> >>>
> >>> Rodolphe
> >>>
> >>>
> >>> On Sun 2/9/14, at 11:42, jcassidy <jca...@ga...> wrote:
> >>>
> >>> >
> >>> > Hello All,
> >>> >
> >>> > I was wondering if anyone has had success getting a Flea camera
> >>> working on 10.9.1
> >>> > with a usb3 port?
> >>> >
> >>> > The camera is a: PtGrey Flea FL3-U3-13E4M, vendor: 0x1e10, prod:
> >>> 0x3300. It is recognized and
> >>> > responds to initial function calls in the helloword.c example program
> >>> but fails at:
> >>> >
> >>> > err = dc1394_capture_setup( camera, 4,
> >>> DC1394_CAPTURE_FLAGS_DEFAULT );
> >>> >
> >>> > with the message:
> >>> >
> >>> > libdc1394 error: usb: Failed to submit initial transfer 0
> >>> >
> >>> > I am using the latest dc1394 source via a git clone and libusb-1.0.18
> >>> on a recent mac
> >>> > mini with usb3 ports.
> >>> >
> >>> > Any insights/help would be greatly appreciated!
> >>> >
> >>> > Regards,
> >>> > Jim.
> >>> >
> >>> >
> >>> >
> >>> >
> >>> ------------------------------------------------------------------------------
> >>> > Managing the Performance of Cloud-Based Applications
> >>> > Take advantage of what the Cloud has to offer - Avoid Common Pitfalls.
> >>> > Read the Whitepaper.
> >>> >
> >>> http://pubads.g.doubleclick.net/gampad/clk?id=121051231&iu=/4140/ostg.clktrk_______________________________________________
> >>> > Mailing list for libdc1394-devel
> >>> > lib...@li...
> >>> > https://lists.sourceforge.net/lists/listinfo/libdc1394-devel
> >>>
> >>> --
> >>> | Rodolphe Pineau RTI-Zone |
> >>> | http://www.rti-zone.org/ |
> >>> | Robotics / Unix / Mac OS X / Astronomy |
> >>>
> >>>
> >>>
> >>> ------------------------------------------------------------------------------
> >>> Managing the Performance of Cloud-Based Applications
> >>> Take advantage of what the Cloud has to offer - Avoid Common Pitfalls.
> >>> Read the Whitepaper.
> >>>
> >>> http://pubads.g.doubleclick.net/gampad/clk?id=121051231&iu=/4140/ostg.clktrk
> >>> _______________________________________________
> >>> Mailing list for libdc1394-devel
> >>> lib...@li...
> >>> https://lists.sourceforge.net/lists/listinfo/libdc1394-devel
> >>>
> >>
> >>
> >
> >
> > ------------------------------------------------------------------------------
> > Flow-based real-time traffic analytics software. Cisco certified tool.
> > Monitor traffic, SLAs, QoS, Medianet, WAAS etc. with NetFlow Analyzer
> > Customize your own dashboards, set traffic alerts and generate reports.
> > Network behavioral analysis & security monitoring. All-in-one tool.http://pubads.g.doubleclick.net/gampad/clk?id=126839071&iu=/4140/ostg.clktrk
> >
> >
> >
> > _______________________________________________
> > Mailing list for lib...@li...https://lists.sourceforge.net/lists/listinfo/libdc1394-devel
> >
> >
> >
> >
> > ------------------------------------------------------------------------------
> > Subversion Kills Productivity. Get off Subversion & Make the Move to
> > Perforce.
> > With Perforce, you get hassle-free workflows. Merge that actually works.
> > Faster operations. Version large binaries. Built-in WAN optimization and
> > the
> > freedom to use Git, Perforce or both. Make the move to Perforce.
> >
> > http://pubads.g.doubleclick.net/gampad/clk?id=122218951&iu=/4140/ostg.clktrk
> > _______________________________________________
> > Mailing list for libdc1394-devel
> > lib...@li...
> > https://lists.sourceforge.net/lists/listinfo/libdc1394-devel
> >
> >
> ------------------------------------------------------------------------------
> Subversion Kills Productivity. Get off Subversion & Make the Move to Perforce.
> With Perforce, you get hassle-free workflows. Merge that actually works.
> Faster operations. Version large binaries. Built-in WAN optimization and the
> freedom to use Git, Perforce or both. Make the move to Perforce.
> http://pubads.g.doubleclick.net/gampad/clk?id=122218951&iu=/4140/ostg.clktrk
> _______________________________________________
> Mailing list for libdc1394-devel
> lib...@li...
> https://lists.sourceforge.net/lists/listinfo/libdc1394-devel
--
Damien 高原 Douxchamps http://damien.douxchamps.net/
|