[Elphel-support] channel switching and exposure question

[Andrey Filippov]

>
> Currently I'm using two modes. The combined mode which works for 50%
> width, 50% height frames only (in my old version) and the alternating
> mode with buffering for full resolution frames. I haven't looked at the
> latest development for some time.
>
> What I need is the ability to capture two frames at exactly the same
> time. You said that the alternating mode with buffering(?) isn't
> supported any longer. The combining mode doesn't support full resolution
> which would be 2 * 5Mpix = 10Mpix, does it?
>

Andreas, current codes support image height up to 8192 and width of 4096
pixels so yes, you can use it with the 2 full resolution sensor frames


>
> I suppose that you have solved that for the Eyesis with three sensors,
> so I think it should work. Which mode are you using?
>

Combined mode with TRIG=4, so sensors are triggered. In Eyesis we
synchronize 3 10353 boards, so one "master" camera triggers itself and the
two other boards, all 9 sensors are triggered simultaneously (with better
than a microsecond precision). In addition to just synchronizing trigger
current fpga code distributes timestamp from the master camera, so all 3
composite images (each with 3 subframes) has exactly the same timestamp even
if the camera clocks may have some difference.


>
> >     Andrey, nor fpcf or reg_write.php follow the safety procedure you
> >     mentioned in your last mail, do they?
> >
> >
> > It is done on the driver write level, there were some updates recently
> > (mostly related to access to the sensor and 10359A boards EEPROMs -
> > their i2c is slower than sensor and 10359 FPGA)
> Do you mean the driver which handles the device /dev/xi2c16 ? If so, I
> don't have to take care of the sequencer, do I ?
>

You do not need to use sequencer directly, you just need to change parameter
as "exposure" or "quality" or "flip" (global or per-sensor) and specify - at
what frame number that change to take place. Writing directly to the
register can be useful only when testing/troubleshooting driver, when you
need to change those registers without driver "noticing" it. If you really
need to do it for some other reasons - please write us, maybe something is
missing in the driver.



> I have seen that in the code for autoexposure. Good to know that the
> queue is 7 frames.
>

both hardware sequencers have 8 frames (their registers are defined in
c353.h) but as one frame is "now" there are only 7 "ahead" left. On the
driver level one more frame is lost as the 8-frame parameter memory includes
"previous frame", "current frame" and 6 frames ahead.


> Ok, I will have a look at parsedit.php.
>

Parameters have "tooltips" - when you hold mouse pointer over it will show
some short info about it.
You can address sensor registers there as SENSOR_REGSNNN (where NNN is
decimal register number) or specify some bit filed in the parameter (i.e.
SENSOR_REGS32__0106) - two undercores folowed by 4 digits - first two
specify bit field width, second two - bit number of the lowest bit in a
field. when you open parsedit.php without parameters, it will show you a
mini-help that includes this explanations

Most sensor parameters in the multi-sensor cameras are treated as they
should have the same value and are written via i2c ina "broadcast mode"
(maintaned by the 10359, but some may have individual values (we'll likely
increase the number of such parameters). In parsedit.php you may write to
just the SENSOR_REGSNNN - it will be sent to all sesnors (and overwrite
each  shadow), but you may also address SENSOR_REGSNNN__A ...
SENSOR_REGSNNN__C and SENSOR_REGSNNN__a ... SENSOR_REGSNNN__c to access
individual sensor (A, B or C) Lower case suffx (__a, __b and __c) is strict
and result in an error if that particular SENSOR_REGSNNN does not have
individual capability, upper case suffix is just ignored without error in
such case.


MULTI_REGSNN allow you to view/edit 10359 registers during debug process,
normally you should use higher level parameters - currently there are 34 of
the parameters related to multisensor operation


Andrey
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://support.elphel.com/pipermail/support-list_support.elphel.com/attachments/20100907/9f38a470/attachment-0002.html>