This was taken two days ago using a Librem 5 Evergreen running PureOS byzantium, using the “megapixels” app version 0.16.0-1pureos2, it’s one of my favorite views of Stockholm, from the top of the Västerbron bridge, looking east:
This was on a bright sunny day (but the sun was behind me, not shining into the camera), the sky was a beautiful blue color that is for some reason lost, it looks mostly white/gray in the picture. In the top-left corner of the picture a bit of the blue sky color is seen, but really most of the sky had that color.
Does anyone here have some idea about what happens here, is there some way to tweak things to make this kind of picture look better?
Here is what it looked like when I aimed a shot more down towards the water, the dark part at the bottom is the shadow of the bridge:
You’re probably getting some vignetting in the corner, where the light intensity is lower, and it doesn’t get overexposed. Just drop down sensitivity settings until color appears. The three primary colors are not captured evenly, leading to weird stuff like that.
Thanks! Just to clarify, which are the sensitivity settings?
It looks like the megapixels app allows tuning the following four settings:
Gain
Exposure
Balance
Focus
By “sensitivity settings” do you mean all the top three, all except Focus?
Anyway, when I took the Stockholm pictures above I had not touched any of the settings, they were all at their default values. I was unaware that they could be changed, but I see now that it’s easy, just tap on one of the numbers and a slider appears to tune it. I’ll make a new attempt next time I’m on that bridge.
From the comment about the image being “overexposed” I’d start with the exposure setting
But also I have no idea what “gain” would be in an image context. I’m used to that as a microphone setting, maybe that setting is for audio gain for recording video and just not hidden when taking still images?
I do thank you for sharing these images so those of us still waiting can see the progress made while we wait
Just curious to know: what about to use same technics used by Google in Pixel 2?
I don’t remember the name of this technique but I read Google “stolen” from NASA: using only one camera + micro movements of the hands + AI, it’s possible to simulate 2 or more cameras to have a lot of pics from different view/position merged in one. Using it, it’s possible to have effects like Bokeh/Portrait and exceptional improvement of night pics and videos.
Yup - when it comes to the big camera it seems like there’s some race condition in the driver and it works better on some devices but worse on others. On one of my phones it works pretty much perfectly, on another one it sometimes needs some encouragement, and on another one it mostly fails. It obviously needs some more work.
The selfie cam driver should already be reliable everywhere though - at least I’m not aware of any issues with it.
About the selfie cam, when I use Megapixels, it goes to 250% CPU usage, and after around 1 min, the battery is notified to 0% and the phone shuts down (when I reboot, the percentage battery is back to normal)
I was able to save a front camera picture for the first time today.
Trying to switch to the back camera does nothing; whatever the front camera was seeing remains frozen in view. If I close the app and then come back to it, I can tell that the back camera is engaged, because the “Focus” control has been added to the screen. The view is just pitch black, though.
One curious thing: there’s a delay in viewing the subject through the front camera, e.g. the subject moves a bit and the same action is seen a second or two later on the screen view.
It also seems that I have to crank the phone’s brightness all the way up to see the subject, and again, only present with the front camera. Back camera view sees nothing.
Yes, I thought about that. Or a USB webcam. (Part of the problem is that there is often a queue for the COVID check-in, so getting a photo of the QR code needs to be fast and efficient. What happens after that can be slow and clunky i.e. running a shell script to do something with the photo. I can leave the queue to do that.)
I’m not. Even though I can’t save main camera photos (most of the time), whereas selfie camera photos seem to work reliably, I have not experienced anything like what you did. I just have a malfunctioning application that is annoying but, no drama, I can close it and continue to enjoy use of the phone.
That said, I haven’t tested the cameras since the updates that were applied just this morning.
Some ppl who had trouble with it earlier now can take photos after installing the mentioned kernel on their Byzantium phones. If you test it, don’t hesitate to mention it in the issue, so we can close it and celebrate (or get back to the drawing board).
The kernel got an update to 5.13.0-1-librem5 on Amber but still can not save images from the main camera. It seems we must wait for Byzantium to become the default.
By the way is Autofocus coming to megapixels anytime soon?
Yes, Byzantium is now where development takes place. It should be available in the near future.
No, I don’t think we’re going to have enough capacity to work on autofocus. There’s a lot of foundational work left - media interface, switch handling, calibration, libcamera. But I might be wrong and it might pop up as the next logical step.
