I expect 1. and 2. to benefit developers because they allow issue report quality to increase a lot in certain areas.
=> More informations about this toot | View the thread
I'm very happy with the initial results I get from #postmarketos switching to #systemd and think it will help a lot improving the platform going forward. My current favorites after playing around a bit over the last days:
journalctl -e
is so much more informative and powerful than what we previously had.=> More informations about this toot | View the thread
So everybody wins from my mistake and I wonder why my #db app doesn't tell me about such options. Not sure how usual cases like this are - but I guess we could improve the train experience quite a bit, just with a bit of software and much less privacy-invasive than the car counter-part (which exists for more than a decade now if I'm not mistaken?)
=> More informations about this toot | View the thread
Interesting #db / train experience: I just forgot to change trains, noticing half an hour late. Turns out the train I was supposed to take has "exceptionally high demand" and I'd probably would have had to sit on the floor. Now I took the next alternative route and
=> More informations about this toot | View the thread
=> More informations about this toot | View the thread
I mainly like this one because it shows so nicely how some caring person can keep old HW running in an evolving environment.
2/3
=> More informations about this toot | View the thread
The #mesa 24.3 release is, as usual, pretty neat. Some random personal highlights:
1/3
=> More informations about this toot | View the thread
And what's really cool here: the matrix will be part of the tuning file which can be created and edited without rebuilding libcamera, making that process more inviting.
=> More informations about this toot | View the thread
For #linuxmobile #postmarketos #mobian folks interested in cameras: there's a first version of patches for #libcamera softwareISP color correction matrix (CCM) support now: https://patchwork.libcamera.org/cover/22041/
Together with autofocus (AF) this probably the main missing feature to get a somewhat reasonable image quality on devices like the #pixel3a or #fairphone5 with open drivers / close to #mainlinelinux.
=> More informations about this toot | View the thread
Only reason it hasn't been done by others already seems to be just lack of time - so let's try to share work among more shoulders, especially as it's a task that's likely going to be quite rewarding :)
=> More informations about this toot | View the thread
For #linuxmobile / #postmarketos / #mobian etc. folks: there's a rather beginner friendly task to improve battery life for qcom devices like the #oneplus6, #pixel3a etc. by making sure sensors are idle when not needed.
If you have some spare cycles, motivation, basic C literacy (or want to learn it) and know how to build and deploy/test a package with pmbootstrap (or want to learn it) - this one could be for you.
The details are here: https://gitlab.postmarketos.org/postmarketOS/pmaports/-/issues/2455#note_453587
=> More informations about this toot | View the thread
The new #pipewire 1.0.9 release backports a bunch of fixes for cameras - if your distro ships 1.0, I strongly recommend to pick up that update :)
=> More informations about this toot | View the thread
Notably focus is missing on the Pixel3a rear camera - and the image is zoomed because we're still figuring out the right register values for the mode used by default by Gnome Camera / Snapshot (1080p).
=> More informations about this toot | View the thread
[#]postmarketos just landed an update for the #pixel3a enabling the rear camera \o/
That, AFAIK, makes it the first #exandroid (close to mainline kernel) device were both cameras work OOTB.
To set expectations right here: there's still a lot of work in various components in the stack before the camera experience comes even close to what you get on the same devices running Android (kernels).
Short comparison with my Pixel 6a running #lineageos - the rear cameras on both devices use the IMX363:
=> View attached media | View attached media | View attached media | View attached media
=> More informations about this toot | View the thread
Those using the experimental nvidia-vaapi-driver will additionally need VaapiOnNvidiaGPUs.
If things go well I hope we can enable them by default in the next months, 🤞
=> More informations about this toot | View the thread
[#]chromium on #Linux nightly (upcoming 132) now has better #V4L2 hardware decoders support - used by #arm devices instead of VA-API. It's still disabled by default, but the flags to enable it have been unified. To enable it you'll need:
https://chromium-review.googlesource.com/c/chromium/src/+/5872600
[#]linuxmobile
=> More informations about this toot | View the thread
Similarly, we'll need a bunch of simple VCM drivers for focus.
If you're interested, maybe take a look at https://gitlab.com/sdm845-mainline/linux/-/issues/32#note_2141816416, https://gitlab.com/sdm670-mainline/linux/-/issues/2 and similar issues.
=> More informations about this toot | View the thread
To follow up on a recent post of mine (https://mastodon.social/@rmader/113234428076202804): if you want to see working cameras on more devices using a #mobilelinux distro of you choice, consider helping with sensor driver upstreaming. Apparently it a bunch of cases this really isn't all to difficult and the main work you need to do is combine an existing driver with data from downstream kernels, test that it works - et voilà, a big part of the work is done!
[#]postmarketos #mobian
=> More informations about this toot | View the thread
For those running #postmarketos edge on a device with working camera - the number of which is raising fast enough to make me lose track - and you want to play around with the great new #gnomecamera / #gnomesnapshot #gnome47 release: you'll need to install gst-plugins-rs manually until some build issues are fixed.
The release features heavily improved rendering performance, mainly thanks to improvements in #gtk4 and the #gstreamer gtk4paintablesink (also for apps using the aperture library)
=> More informations about this toot | View the thread
Hm, looks like I was wrong about de- and encoding parts and it's actually hard to speed those up with opencl. One claim I read was that modern codecs are just too branchy and CPU ends up being better - does anyone know more about that, specifically regarding AV1?
=> More informations about this toot | View the thread
=> This profile with reblog | Go to rmader@mastodon.social account This content has been proxied by September (ba2dc).Proxy Information
text/gemini