Per Dave Sparks request, bug 1986 was entered for this issue.
On Jan 28, 1:58 pm, Dave Sparks <[email protected]> wrote:
> I agree we need some work in this area.
>
> We get around the problem right now by forcing the stride to be the
> same as the display. On G1, this means preview must always be a
> multiple of 16 pixels.
>
> On Jan 26, 11:54 am, steve2641 <[email protected]> wrote:
>
>
>
> > Hello,
>
> > I have a question about the preview frames that can be returned by
> >CameraServices to the applications. There seems to be two different
> > means of returning the frame, copied or not copied, ignoring the one
> > shot case which is not important for this discussion. In both of
> > these cases, the stride of the buffer does not seem to be taken into
> > account. The application will need the stride in order to properly
> > parse the buffer, unless the width of the frame data and the stride
> > are equal.
>
> > If one were to use hardware or a software feature which pads the
> > frame, the Surface Flinger would be able to properly parse the buffer,
> > since it has the width stride, but I have not found any place where
> > the application can get this key piece of data.
>
> > This issue could be quickly solved for the cases where a copy of the
> > preview buffer is requested by the application. In this case the
> > copyFrameAndPostCopiedFrame() function would just need to copy the
> > buffer with the knowledge of the stride. The stride could be stored
> > off when the Surface Flinger is given the heap and then reused in this
> > function to properly copy the preview data from the preview frame to
> > the buffer that will be passed to the application.
>
> > In the case of returning the actual preview memory, the issue gets
> > more complicated. The application would need to be given (maybe in
> > the callback) the actual frame stride. I'm not sure how the
> > application could parse the preview frame without it.
>
> > One feature that will tend to cause preview frames to be padded is
> > image stabilization, but I can also imagine hardware limitation that
> > may also cause padding.
>
> > Thanks,
>
> > Steve.- Hide quoted text -
>
> - Show quoted text -
--~--~---------~--~----~------------~-------~--~----~
unsubscribe: [email protected]
website: http://groups.google.com/group/android-porting
-~----------~----~----~----~------~----~------~--~---
