Recherche avancée

Sur d’autres sites (5805)

• Adjusting webcam settings before shooting a photo (v4l2 and fswebcam)

  18 septembre 2021, par Mikey

  I'm shooting photos with

  


  sudo fswebcam -d /dev/video0 -r 1920x1080 --no-banner /media/networkshare/public/"Temp Photo Holder SolidScape Right"/timelapse_$DATE.jpg


  


  on a Raspberry Pi 3 B+ running a recent version of Raspbian.

  


  The script is controlling a Logitech c920 that has adjustable features like focus, brightness, contrast, etc.

  


  I think the reason the 'manual' settings are not being saved is because I call the v4l2-ctl commands, add a delay, and then use fswebcam to shoot, like so :

  


  #!/bin/bash
DATE=$(date +"%y%m%d%H%M%S")
#sudo v4l2-ctl -d /dev/video0 -c focus_auto=false
#sudo v4l2-ctl -d /dev/video0 -c focus_absolute=35
sudo v4l2-ctl -d /dev/video0 -c brightness=128
sudo v4l2-ctl -d /dev/video0 -c contrast=128
sudo v4l2-ctl -d /dev/video0 -c saturation=128
sudo v4l2-ctl -d /dev/video0 -c gain=15
sudo v4l2-ctl -d /dev/video0 -c sharpness=128
sleep 2

sudo v4l2-ctl -c exposure_auto_priority=1

#fswebcam -d /dev/video0 -r 1920x1080 --no-banner /media/networkshare/public/RasPi/"Temp Photo Holder SS"/timelapse_$DATE.jpg

sudo fswebcam -d /dev/video0 -r 1920x1080 --no-banner /media/networkshare/public/"Temp Photo Holder SolidScape Right"/timelapse_$DATE.jpg

sleep 9

sudo fswebcam -d /dev/video0 -r 1920x1080 --no-banner /media/networkshare/public/"Temp Photo Holder SolidScape Right"/timelapse_$DATE.jpg

sleep 9


  


  How can I take a photo with manual settings appended ? Maybe I need to add commands to fswebcam inline instead of calling v4l2-ctl first ?

  


• Download Partial Video via HTTP (for Remote Thumbnailing)

  13 février 2012, par HuntedC

  I have videos hosted on Amazon S3. I encode them with Zencoder and store a thumbnail for the video then using Zencoder. However, I need a way to generate thumbnails at certain points in the video (i.e. 00:00:03, 00:10:32, 01:40:18) and store them either on S3 or my server.

  ffmpeg allows remote thumbnailing, however it takes a very long time (sometimes several minutes) to get a thumbnail from the middle of a file—I believe this is because it downloads the entire file up to that point to get the thumbnail.

  My plan is to somehow download the header of the video file via HTTP byte-range request, guesstimate the byte range where I should be looking for the thumbnail, download about a second of video from that part of the file via HTTP byte-range request, then save the header and tiny video locally. I pull the thumbnail from that using ffmpeg and delete the temporary video.

  I have no idea on how exactly this would work (I believe the H.264 MP4 files I'm working with have a dynamic length header, for another issue). Any suggestions or better ideas ?

  Edit : To clarify, Zencoder thumbnailing is great, but they only allow thumbnail creation in combination with transcoding. I don't want to transcode my video every time I create a new thumbnail, so I need to do this on my own without Zencoder.

• FFmpeg - How should I do to convince YouTube that it is receiving "endless" video stream ?

  16 octobre 2017, par K._

  I’m using FFmpeg and Youtube Live Streaming API to live-stream videos to YouTube.

  I made something generates video files continuously,
  and I want to live-stream them to YouTube without any delay.

  To feed them to YouTube’s RTMPT(rtmp ://a.../live2/...) server,
  I tried several things :

  1. Just while:; do ./ffmpeg ... rtmp://...; done.
  2. Piping

  YouTube streamed them continuously, but not seamlessly.
  It showed Buffering icon for about 20 seconds (as if a new YouTube stream gets created and starts to be fed), about 5 seconds before the currently showing video ends and the next video get showed.

  | ...===============(BUFFERING)===== | ...===============(BUFFERING)===== | ...

  It seems that YouTube just stops the stream when it detects the end of a video.
  How should I do to convince YouTube that it is receiving "endless" video stream ?