Error with photos fading to black about picamera

[waveform80]

Yes, I'd observed this myself (when capturing without preview) but thought it was a firmware bug. However, reading through that thread you linked to is most interesting - I'd been wondering why James bothered connecting the preview to a null-sink when the camera seemed to operate more or less fine without it, but it turns out that's the solution to this particular problem.

Anyway, many thanks for the excellent bug report! I'll schedule this for the 0.7 release (which I'm vaguely hoping might be tomorrow depending on how much time I get this weekend!)

[parkview]

Awesome! Looking fwd to trying it out next weekend :-)

[parkview]

Hi Dave,

do you think 0.7 version will be released before this Sat (Australia), or
Fri (USA)? If so, I would like to upgrade and have a good test with it on
my Sat. bush walk:

http://www.bdug.org.au/project/pipole-cam/about

Thanks.

-Parkview.

On Sat, Nov 9, 2013 at 6:48 PM, Dave Jones [email protected] wrote:

Yes, I'd observed this myself (when capturing without preview) but thought
it was a firmware bug. However, reading through that thread you linked to
is most interesting - I'd been wondering why James bothered connecting the
preview to a null-sink when the camera seemed to operate more or less fine
without it, but it turns out that's the solution to this particular problem.

Anyway, many thanks for the excellent bug report! I'll schedule this for
the 0.7 release (which I'm vaguely hoping might be tomorrow depending on
how much time I get this weekend!)

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/22#issuecomment-28124555
.

[waveform80]

Yes, it should be released tomorrow (Thursday GMT) - I was hoping to get the resizer code into the 0.7 release but I don't think that's realistic this week - I'm just not convinced I've got the API right yet. However, there's sufficient fixes and enhancements to justify a release anyway (fade-to-black, crop, video quantization, etc.) so whatever else happens I'll release 0.7 tomorrow evening.

[parkview]

Thats awesome. I could survive with out it ;-) as I could just set the
RPi Pole-Cam to to just sequence 50 odd photos and then restart it for
another lot.

Be nice to test it out though. I will report back & blog how it goes.

Cheers!

On Thu, Nov 14, 2013 at 7:36 AM, Dave Jones [email protected]:

Yes, it should be released tomorrow (Thursday GMT) - I was hoping to get
the resizer code into the 0.7 release but I don't think that's realistic
this week - I'm just not convinced I've got the API right yet. However,
there's sufficient fixes and enhancements to justify a release anyway
(fade-to-black, crop, video quantization, etc.) so whatever else happens
I'll release 0.7 tomorrow evening.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/22#issuecomment-28445482
.

[parkview]

Hi Dave,

yes, Picamera 0.7 seems to have fixed the problem I was having, although I
only tested a runs of 100 photos each run:
http://www.bdug.org.au/project/pipole-cam/pipole-cam-flys-again.

Thanks for that. I look fwd to future enhancements.

Cheers,

Parkview.

On Thu, Nov 14, 2013 at 7:36 AM, Dave Jones [email protected]:

Yes, it should be released tomorrow (Thursday GMT) - I was hoping to get
the resizer code into the 0.7 release but I don't think that's realistic
this week - I'm just not convinced I've got the API right yet. However,
there's sufficient fixes and enhancements to justify a release anyway
(fade-to-black, crop, video quantization, etc.) so whatever else happens
I'll release 0.7 tomorrow evening.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/22#issuecomment-28445482
.

[6by9]

Dave: Can you retest to see if this is really still an issue?
I'm looking into this whilst I still can, and whilst AGC is fixed across captures, they aren't going to black for me via V4L2. I suspect that a change I made back in March to stop AWB recalculating on every capture also means we're not getting rounding on AGC. (It did introduce the lens shading issue though)
If I can then I'll enable AGC between captures unless told otherwise by the client.

[waveform80]

@6by9 I think my assertion that fade-to-black was back with burst-mode was incorrect; last time I didn't wait for it to fade completely - I simply saw a darkening across a long-running timelapse and jumped to the conclusion. Rather stupidly however, I was just running this in the loft during the daytime and I wonder whether it wasn't simply a darkening due to changing daylight. I'll test this again tonight with artificial lighting only to make sure.