Re: [ros-users] call for testing: camera1394 in C-turtle

Top Page
Attachments:
Message as email
+ (text/plain)
Delete this message
Reply to this message
Author: Patrick Beeson
Date:  
To: Jack O'Quin
CC: ros-users
Subject: Re: [ros-users] call for testing: camera1394 in C-turtle
Jack, I really needed to fix this before leaving my robot with others
for a week. So, I have a patch that fixes this by default to
non-absolute mode if the device says absolute isn't available or the
absolute_set fails.

Should I check this in or are you actively working on this as well?



On 06/16/2010 11:11 AM, Patrick Beeson wrote:
>
>
> On 06/16/2010 10:05 AM, Jack O'Quin wrote:
>> I think what's happing is that the camera lies and says it accepts
>> "absolute" (i.e. float) values. The driver, seeing "absolute_capable
>> == true", tries to set the values using
>> dc1394_feature_set_absolute_value() instead of
>> dc1394_feature_set_value().
>>
>> Not sure how to handle this in a portable fashion that works for all
>> cameras.
>
> That's exactly what happens. Except that is doesn't fully lie. It return
> back the float of FPS (0.0001-16.0) for get_absolute_boundaries, but for
> get_boundaries, it returns back integers of 3-1150 (you don't have the
> get_boundaries call in that branch by the way, you always use 0-4095).
> So it can READ absolute values, but the set_absolute_value for shutter
> isn't working, even though it works for other features. Don't exactly
> see why yet.
>