2,980 karma · joined August 10, 2010
I'm looking to move away from a QNAP box, and one of the driving reasons is the horrible "hard-plastic hard-mount everything" design that couldn't amplify hard drive noise any more if they'd done it on purpose.
I'm looking to move away from a QNAP box, and one of the driving reasons is the horrible "hard-plastic hard-mount everything" design that couldn't amplify hard drive noise any more if they'd done it on purpose.
(The other reasons are that I'd rather manage ZFS myself, and the need for more than gigabit ethernet)
Where does that appear? Its not on the button bar that appears over a text selection.
I think I read somewhere that Android 10 forbids apps from managing the clipboard. Can anyone confirm this? I hold out hope that clipboard integration might work again...
# Save a RTSP stream to file
ffmpeg -i rtps://someserver/somevideo -c copy out.mp4
# encoding quality settings examples
# H264
ffmpeg -i input.mp4 -a:c copy -c:v libx264 -crf 22 -preset slower -tune film -profile:v high -level 4.1 output.mp4
# H264 10bit (must switch to a different libx264 version)
LD_LIBRARY_PATH=/usr/lib/x86_64-linux-gnu/x264-10bit ffmpeg -i input.mp4 -a:c copy -c:v libx264 -crf 22 -preset slower -tune film -pix_fmt yuv420p10le -profile:v high10 -level 5.0 output.mp4
# resize
ffmpeg -i in.mp4 -s 480x270 -c:v libx264 out.mp4
# change source fps
ffmpeg -r 10 -i in.mp4 out.mp4
# resample fps
ffmpeg -i in.mp4 -r 10 out.mp4
# extract the audio from a video
ffmpeg -i inputfile.mp4 -vn -codec:a copy outputfile.m4a
# merge audio and video files
ffmpeg -i inputfile.mp4 -i inputfile.m4a -codec copy outputfile.mp4
# cut (reencoding)
# set the start time and duration as HH:MM:SS
ffmpeg -ss 00:00:40.000 -i input.mp4 -t 00:00:10 -c:v libx264 output.mp4
# set the start time and duration as seconds
ffmpeg -ss 40.0 -i input.mp4 -t 10.0 -c:v libx264 output.mp4
# skip an exact number of frames at the start (100)
ffmpeg -i input.mp4 -vf 'select=gte(n\,100)' -c:v libx264 output.mp4
# cut (w/o reencoding - cut times will be approximate)
ffmpeg -ss 40.0 -i input.mp4 -t 10.0 -c copy output.mp4
# save all keyframes to images
ffmpeg -i video.mp4 -vf "select=eq(pict_type\,I)" -vsync vfr video-%03d.png
# encode video from images (image numbering must be sequential)
ffmpeg -r 25 -i image_%04d.jpg -vcodec libx264 timelapse.mp4
# flip image horizontally
ffmpeg -i input.mp4 -vf hflip -c:v libx264 output.mp4
# crop and concatenate 3 videos vertically
ffmpeg -i cam_4.mp4 -i cam_5.mp4 -i cam_6.mp4 -filter_complex "[0:v]crop=1296:432:0:200[c0];[1:v]crop=1296:432:0:200[c1];[2:v]crop=1296:432:0:230[c2];[c0][c1][c2]vstack=inputs=3[out]" -map "[out]" out.mp4
# 2x2 mosaic (all inputs must be same size)
ffmpeg -i video1.mp4 -i video2.mp4 -i video3.mp4 -i video4.mp4 -filter_complex "[0:v][1:v]hstack=inputs=2[row1];[2:v][3:v]hstack=inputs=2[row2];[row1][row2]vstack=inputs=2[out]" -map "[out]" out.mp4
# picture-in-picture
ffmpeg -i video1.mp4 -i video2.mp4 -filter_complex "[1:v]scale=iw/3:ih/3[pip];[0:v][pip]overlay=main_w-overlay_w-20:main_h-overlay_h-20[out]" -map "[out]" out.mp4
# print framerate of every file in dir
for f in *.mp4; do echo $f; mediainfo $f|grep "Frame rate"; done
# print selected info of every file in dir in CSV format
for f in *.mp4; do echo -n $f,; mediainfo --Inform="Video;%Duration%,%FrameCount%,%FrameRate%" $f; done[Edit] doc: https://www-archive.mozilla.org/docs/end-user/keywords.html
The vast majority of the time the airplane is at cruising altitude, though. That air is as clean as you are going to get on planet earth.
"From October, Mastercard said it would increase these fees to 1.5% on every transaction, up from 0.3%. The EU introduced a cap on such fees in 2015 after concerns they pushed prices up for consumers and unfairly burdened companies."
"Mastercard is implementing the rises simply as it's no longer bound by the restrictions imposed by the UK being in the EU."
Thus proving the caps were indeed necessary. If MC was making money with a 0.3% fee the other 1.2% are pure sweet profit margin.
(I even think the DASH segmentation would allow them to this without reencoding the original video, but that's just speculation on my part)
It's funny because there's nothing new about those login popups - I already knew conceptually that kind of thing was possible - but seeing your little avatar picture show up where it doesn't belong provokes a much more direct reaction than abstract knowledge.
Reasons to buy a T4:
- You need the 16Gb RAM. You usually won't for inference, though.
- You need single slot cards to use in server chassis. Even here you are likely better off buying a Quadros than a T4
- You need the high power efficiency of the T4 (75W TDP)
It helps keeping to Ubuntu LTS versions though, that's what they support best.
But you're right about the randomization (although I don't think 9 seconds is right, I've never seen it take nearly as long as that)
It appears to be largely equivalent to the Calibre editor.
That's how I have this very laptop configured, and also how I configured the workstations for the people here using GPUs for DL development (because a composited desktop will get very janky if you try to run it off a GPU currently being brought to its knees by a training process).
Legacy BIOS had to go, it was designed at the time of 8086 and DOS and was really out of step with modern HW and OS needs. The replacement could have been even simpler though, now that OSes really want to take over everything themselves and no longer lean on the BIOS that way that DOS used to. Instead EFI created a monster and now we're stuck with it.
I'll definitely be creating a ListenBrainz account. As a long time last.fm user I occasionally worry about the future of the platform. (There have been long stretches of time where it seems to have been in maintenance mode). You seem to support bulk importing last.fm data right?
This has all the hallmarks of one of those changes driven by single-minded metric chasing (engagement!) instead of principled reasoning on what the product should do and feel like to use.
As for RAM, only you can know how big your datasets are. But if you're training models on GPUs the bottleneck is almost certainly going to be GPU RAM, not system RAM.