[Elphel-support] Stereo Camera Systems: Distance to Image Sensor and Sub-sampling Ethernet Output

[Justin Burnett]

Olga,
Thank you for your prompt and detailed response.

1) Is there someway I can get more information on the custom sensor cables?
Specifically I am interested in the outer diameter and the jacket material?

2) For best performance in the JP4 format, what kind of write speed would
be required of the local storage device with the current model 353? What is
the anticipated maximum write speed for the upcoming 393 model?
Specifically I am wondering about the limitations of compact flash vs. SSD
for local storage.

3) We have seen your thermally compensated tri-clops camera system and the
associated software demo. Would it be possible(and relatively painless) to
multiplex three sensors to a single camera, and use the same software tool
for measurement and evaluation? This is a very attractive option, but has
some limitations for our application, namely we need to supply our own
(potentially non-calibrated or thermally compensated) lens, because the
optical properties of the pressure housing dome geometry, as well as the
transition to seawater, change the anticipated optics substantially. How
difficult would it be to calibrate a custom lens/assembly? For accurate
results, we would likely need to set up a test pattern in a water tank...

Thank you again for your continued support.

Regards,
-Justin

Justin Burnett
Mechanical Engineer
ANDRILL.org
WISSARD.org

ANDRILL Science Management Office
126 Bessey Hall
Lincoln NE 68588-0341


On Tue, Mar 18, 2014 at 2:23 PM, Elphel Info | Olga Filippova <
info at elphel.com> wrote:

> Hello Justin, Bob,
> it is nice to hear from you, and know you plan to keep exploring under the
> Antarctic with Elphel cameras.
>
> 1) The sensors can be connected to the system and interface board with the
> maximum 450mm cables. These are custom made FPCs with reduced EMI
>  (each signal pair is separated by grounding) and also more flexible then
> the standard FFC. These are the cables we use in Eyesis4Pi.
>
> 2) To have a 1/2 FPS previews you would need to write new FPGA code - it
> is very difficult and not feasible to do it right now when we are
> developing new NC393 camera.
> To increase the frame rate in general you can have each sensor board to be
> supported by it's own system board (when you connect 2 or 3 sensors to one
> system board the FPS drops 2-3 times).
> We do not have completely uncompressed images - the camera records in JP4
> format for the best image quality:
> http://wiki.elphel.com/index.php?title=JP4
> http://www3.elphel.com/importwiki?title=Eyesis4Pi_data_structure#Raw
>
> 3) Photogrammetry:
> Have you seen our photogrammetric cameras:
> http://www3.elphel.com/NC353-369-PHG3
> and  http://www3.elphel.com/nc353-phg
> with lenses corrected for distortions and thermally compensated.
> There is a blog article about calibration for photogrammetry we perform
> and a HTML5 demo for the use of information obtained from NC353-369-PHG3
> camera (triclops):
> http://blog.elphel.com/2012/10/heptaclops-camera-and-the-393/
> May be you will find this information useful for your photogrtammetric
> application.
>
> Let me know if you will have other questions. For more technical
> questions, please write to our support mailing list:
> http://www3.elphel.com/list which is checked and answered daily by our
> technical team, and they will be able to help you.
>
> Best Regards,
>
>
>
> Olga S Filippova
> Elphel, Inc.
> 801.783.5555x107
> www.elphel.com
>
>
> On Mon, Mar 17, 2014 at 6:30 PM, Justin Burnett <jburnett at andrill.org>wrote:
>
>> Dear Elphel Team,
>>
>> I am continuing development work for a submersible vehicle used for
>> science research in Antarctica, called SCINI.  We have been using your 353
>> cameras for many years and are happy with the platform, but looking at new
>> implementation options. We would like to use a single electronics package
>> to control two (or 3) sensors to provide photogrammetric images. We hope to
>> stitch these images together with the aid of tactical grade IMU data in
>> order to characterize population density of newly discovered ecosystems. I
>> think this is something that your platform we well suited for, however I
>> have a few questions as starting points for implementation:
>>
>> 1) Mechanically, it would be very convenient to create a single
>> underwater housing for the system and interface boards, along with local
>> data storage, while housing the sensor modules separately, transmitting
>> data over an underwater connector and cable. Is there anyway to feasibly do
>> this with your existing hardware or are the data rates too high for
>> reliable transmission in a noisy environment? I notice that both your
>> stereo/3D camera system and eyesis architectures have image sensors located
>> a comparable distance away from the system boards as we are hoping for.
>>
>> 2) Even on our single image sensor systems, we are heavily bandwidth
>> limited as far as what we can send through Ethernet. It would be highly
>> advantageous if we could write locally uncompressed images, while
>> compressing and sending ~1/2 frame rate images over Ethernet. Preferably
>> with minimal latency. Have you ever come across a similar need before, and
>> if so how difficult would this implementation be?
>>
>> Thank you for your continued support of the SCINI project!
>>
>> Regards,
>> Justin Burnett
>>
>> Justin Burnett
>> Mechanical Engineer
>> ANDRILL.org
>> WISSARD.org
>>
>> ANDRILL Science Management Office
>> 126 Bessey Hall
>> Lincoln NE 68588-0341
>>
>>
>> --
>> Justin Burnett
>> Mechanical Engineer
>> ANDRILL.org
>> WISSARD.org
>>
>> ANDRILL Science Management Office
>> 126 Bessey Hall
>> Lincoln NE 68588-0341
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://support.elphel.com/pipermail/support-list_support.elphel.com/attachments/20140329/1378d8d5/attachment-0002.html>