[Elphel-support] Asking the technical help

Andrey Filippov support-list at support.elphel.com
Tue Nov 15 09:31:12 PST 2011


On Tue, Nov 15, 2011 at 3:26 AM, GuangHai Jin <gjin at agiltron.com> wrote:

>  Hi Andrey,****
>
> ** **
>
> I understand you as our company grows up from a small company as well.
> Thank you very much for your kind help to consult us. I have a few basic
> questions per your answers.****
>
> 1) First of all you have to keep the window height constant…...****
>
> Does it means that the dynamic small window should have the format as
> X.vs.1936 such as 250x1936 from the 2592x1936 sensor?
>
GuangHai,
No - it does not need to be exactly 1936 - just don't change between the
frames


> ****
>
> As the full frame rate of your camera can realize 14fps with JP4 raw, does
> the maximum frame rate might reach up to 140fps approximately for 250x1936
> small window?
>

It would be almost correct if you reduced the height (1936), not the width
(2592) of the sensor WOI. The sensor has some minimal blanking - vertical
and horizontal, the exact numbers are in the Aptina datasheet (available on
their web site). The vertical one is small (8 lines if I remember
correctly), so reducing frame from 2592*1936 to say 2592*192 (each
dimension should be multiple of 16) will give you (1936+8)/(192+8)= 9.72
gain in fps. On the other hand, horizontal blanking is much larger -
several hundred of pixels, so reducing frame width has lesser effect on the
FPS. This large horizontal blanking time is caused by the need to transfer
analog signals from the selected line of pixels to the array output
multiplexers (they are not interleaved).

If on the camera home page you select "Parameter editor" (or just go to
http://camera_ip/autocampars.php) you can select "woi" (second line) and
then open the link "View/edit Current". Among other parameters it will show
"VIRT_WIDTH" and "VIRT_HEIGHT" that camera calculates for the WOI according
to Aptina specs. in some cases it is possible to modify these values
keeping the sensor operational (i.e. reduce VIRT_HEIGHT)


Additionally, for the really high FPS you may want to use the special
camera mode that allows combining several actual frames into one output
frame. In this mode you can combine even 2-pixel high "frames", converting
the camera into a virtual line-scan device (
http://blog.elphel.com/2010/11/elphel-as-a-line-scan-camera-for-river-panoramas/).
While the camera compressor itself is limited to average pixel rate of
80 MPix/sec in JP4 and 53MPix/sec in JPEG, the recorded frame rate can be
limited fy the network or recording to HDD, SSD or CF (limited by the ATA
implementation in the processor). Each compressed frame has the Exif
metadata attached - while it is negligible in size fro the large frames, it
can be relatively large for the small one, each frame causes frame
interrupt and some software activity. So when the FPGA combines several
small frames into one (with the single Exif header, single frame CPU
interrupt), you can get higher FPS in the cases when it is network/CPU
bandwidth-limited. Or be able to record higher quality (less compressed)
fro the same FPS.

Andrey

> ****
>
> ** **
>
> Thanks again,****
>
> ** **
>
> Guanghai****
>
> Dr. Guanghai Jin****
>
> VP of R&D, Chief Scientist****
>
> Phone: 781-935-1200 ext 124****
>
> Fax: 781-935-2040****
>
> Agiltron, Inc****
>
> 15 Presidential Way****
>
> Woburn, MA 01801****
>
> www.agiltron.com****
>
> ** **
>
> *From:* elphel at gmail.com [mailto:elphel at gmail.com] *On Behalf Of *Andrey
> Filippov
> *Sent:* Monday, November 14, 2011 12:30 PM
> *To:* GuangHai Jin; Elphel Support List
>
> *Subject:* Re: [Elphel-support] Asking the technical help****
>
>  ** **
>
> Hi Guanghai,
>
> We will be happy to consult your engineers, but unfortunately we can not
> work on such project as a contract job. We are a small company and all our
> resources are used by our current projects.
>
> Andrey****
>
> On Sat, Nov 12, 2011 at 8:33 AM, GuangHai Jin <gjin at agiltron.com> wrote:**
> **
>
> Hi Andrey,****
>
>  ****
>
> Thank you so much for your clear explainations. It’s very helpful for me
> to decide which technical way we will chose. Our engineers will dig in the
> details by following your instruction. Meanwhile, I am wondering if your
> company would take this job as a contract for us.****
>
>    ****
>
> Thanks again and have a good weekend,****
>
>  ****
>
> Guanghai****
>
>  ****
>
> Dr. Guanghai Jin****
>
> VP of R&D, Chief Scientist****
>
> Phone: 781-935-1200 ext 124****
>
> Fax: 781-935-2040****
>
> Agiltron, Inc****
>
> 15 Presidential Way****
>
> Woburn, MA 01801****
>
> www.agiltron.com****
>
>  ****
>
> *From:* elphel at gmail.com [mailto:elphel at gmail.com] *On Behalf Of *Andrey
> Filippov
> *Sent:* Friday, November 11, 2011 1:26 PM
> *To:* GuangHai Jin; support-list at support.elphel.com
> *Subject:* Re: [Elphel-support] Asking the technical help****
>
>  ****
>
> Hello Jason,
>
> I'm sorry - we had a problem with the mailing list and I missed that your
> previous email.
>
> Yes, it is possible to dynamically re-program the small window, to the
> extend it is supported by the sensor.
>
> 1) First of all you have to keep the window height constant - changing it
> may cause sensor to output bad frame and it will be skipped by the camera.
>
> 2) There is a lag between sending the window parameters to the sensor, the
> output of the new frame and compressing it in the camera (when you get the
> frame and can analyze the objects there. The camera supports pipeline
> modification of the parameters, when you provide the new values you also
> specify to what frame it applies. On the hardware level the parameters are
> stored in an 8 element circular buffer - the software driver places the new
> values there and the FPGA code applies them to the compressor and send to
> the sensor at appropriate time. The default "program ahead" (if you do not
> specify when the new parameters should be applied) is 3 frames, but you can
> change that in the php function call.
>
> The PHP functions only allow you to program within the capacity of the
> hardware buffer (up to 5 ahead, as of 8 elements others are kept after the
> frame), but using parsedit.php - called directly or through autocampars.php
> you can experiment with other program-ahead times, the software extends the
> hardware command buffer. The parsedit.php allows you to open 9 (that number
> can be changed) last images, each image has the image number, timestamp and
> parameters, so you can verify if any images are lost or corrupted).
>
> Andrey****
>
> On Fri, Nov 11, 2011 at 7:17 AM, GuangHai Jin <gjin at agiltron.com> wrote:**
> **
>
> Dear Andrey,****
>
>  ****
>
> About one month ago, I sent you the e-mail to ask your help regarding the
> technical feasibility on your FPGA based camera, see the following e-mail.
> ****
>
> Could please you reply me kindly?****
>
>  ****
>
> Best regards, ****
>
>  ****
>
> Dr. Guanghai Jin****
>
> VP of R&D, Chief Scientist****
>
> Phone: 781-935-1200 ext 124****
>
> Fax: 781-935-2040****
>
> Agiltron, Inc****
>
> 15 Presidential Way****
>
> Woburn, MA 01801****
>
> www.agiltron.com****
>
>  ****
>
> *From:* GuangHai Jin [mailto:gjin at agiltron.com]
> *Sent:* Wednesday, October 05, 2011 11:24 AM
> *To:* 'Andrey Filippov'; 'support-list'
> *Subject:* [Elphel-support] Problem in using of LiveDVD****
>
>  ****
>
> Hi Andrey,****
>
>  ****
>
> We purchased your network cameras to build our detection system for
> checking a single object’s lateral motion at a refreshing rate about 10/s.
> The next goal is to increase the refreshing rate (>100/s) of the object’s
> position (x, y). My questions are****
>
> 1)      Is it possible to dynamically reconfigure a small date (ex.
> 320x240) window for high refreshing rate in your network camera around the
> object as it moves?****
>
> 2)      If “yes” in question 1, could we assign this kind of job to your
> company as a project contract?****
>
> 3)      Do you have any recommendation? ****
>
> Thank you very much in advance, and looking forward to hearing from you.**
> **
>
>  ****
>
> Dr. Guanghai Jin****
>
> VP of R&D, Chief Scientist****
>
> Phone: 781-935-1200 ext 124****
>
> Fax: 781-935-2040****
>
> Agiltron, Inc****
>
> 15 Presidential Way****
>
> Woburn, MA 01801****
>
> www.agiltron.com****
>
>  ****
>
> *From:* Jason Chen [mailto:jianzhong_chen at agiltron.com.cn]
> *Sent:* Thursday, September 15, 2011 6:35 AM
> *To:* Andrey Filippov; support-list
> *Cc:* jianhuiye; GuangHai Jin
> *Subject:* Re: Re: Re: [Elphel-support] Problem in using of LiveDVD****
>
>  ****
>
> Hi, Andrey,****
>
>  ****
>
> This time I removed the "sudo" from the front of "netboot", it looks like
> program ran some but I'm not sure if the camera is OK now.****
>
>  ****
>
> Regards****
>
> Jason Chen****
>
>  ****
>
>  ****
>
>  ****
>
>
> _______________________________________________
> Support-list mailing list
> Support-list at support.elphel.com
> http://support.elphel.com/mailman/listinfo/support-list_support.elphel.com
> ****
>
>  ****
>
>
> _______________________________________________
> Support-list mailing list
> Support-list at support.elphel.com
> http://support.elphel.com/mailman/listinfo/support-list_support.elphel.com
> ****
>
> ** **
>
> _______________________________________________
> Support-list mailing list
> Support-list at support.elphel.com
> http://support.elphel.com/mailman/listinfo/support-list_support.elphel.com
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://support.elphel.com/pipermail/support-list_support.elphel.com/attachments/20111115/4f37cca0/attachment-0002.html>


More information about the Support-list mailing list