Recherche avancée

Sur d’autres sites (10079)

• How to make a bot available to multiple users on blocking operations ? python, aiogram, pytube, ffmpeg

  27 février 2023, par Tony Reznik

  I'm trying to deal with multithreading and asynchrony, but I can't figure out how to deal with this situation.

  


  There is a telegram bot that downloads videos from YouTube and cuts only the first minute of the video into a separate file. Pytube and ffmpeg are used. The second bot user does not receive any response from the bot while the tasks of the first one are running. How to deal with such difficulties ?

  


  import subprocess

from aiogram import Bot, Dispatcher, types
from aiogram.utils import executor
from pytube import YouTube
import time

API_TOKEN = 'tkn'

bot = Bot(token=API_TOKEN)
dp = Dispatcher(bot)

video_folder = 'video/'


@dp.message_handler()
async def start_convert(message: types.Message):
    url = message.text
    yt = YouTube(url)

    await message.reply('accepted')

    video_name = time.strftime("%d-%m-%Y-%H-%M-%S") + '.mp4'

    video = yt.streams.get_highest_resolution()
    video.download(video_folder, filename=video_name)

    await message.reply('downloaded')

    subprocess.call(['ffmpeg', '-hide_banner', '-loglevel', 'error', '-i', f'{video_folder}{video_name}',
                     '-ss', '0', '-c:v', 'libx264', '-c:a', 'aac', '-b:v', '5M', '-to', '60',
                     f'{video_folder}1_min_{video_name}'])

    await message.reply('trimmed')

if __name__ == '__main__':
    executor.start_polling(dp, skip_updates=True)



  


  This is a code template.
In the current form, I want to understand how to implement the task correctly.

  


• I need assistance of making a automated script for ffmpeg [closed]

  18 décembre 2024, par user2614404

  Currently i use this script that GPT generated :

  


  ffmpeg -c:v av1_qsv -b:v 0 -global_quality 20 -preset veryslow -look_ahead 64 -c:a copy -c:s copy G:\1ARC-movies\A-View-to-a-kill.mkv -i Y:\media\Movies\A_VIEW_TO_A_KILL_t05.mkv

  


  But sadly GPT being gpt, it hallucinates answers, and none of it's deeper automation code scripts were working when it made the scripts for a .ps1 script file.

  


  And as i find most of the docs quite overwhelming, i need help to simplify what i need. As i learn best when i get explained what i currently try to do, and not every variable ffmpeg can do.

  


  what i want the script to do is :

  


  1 : convert movies with AV1_QSV with "veryslow", as that's apparently from what i've read up being the best it can do

  


  2 : Global quality as close to 50 as possible (smaller, the better), as 20-6 gave the same range of percentage of 97% VMAF score, 22 gave 95%, so 20 is the largest number i can go, as VMAF via NMKODER reported any higher number was sub 97%.

  


  3 : Give it a main folder for movies and shows to scan, and convert all the media in there to their respective subfolders to a designated output folder.

  


  2 : Automatic crop detection, as some crops it to a 21:9 ratio, others has very small letterboxes. As GPT's code for that attempt for some reason used cpu instead of arc's quicksync.

  


  As this will be ran from powershell, and it's to replace my plex library, by transcoding all movies and shows, starting with the least watched ones/shows that my family has seen, as then those will be unavailable for them until unraid 7 is out of beta to read intel arc for av1.

  


• ffmpeg - two very similar webm files, two very different conversion speeds to mp4

  28 juin 2018, par Nuthinking

  I generate webm files in two different ways. One using Chrome WebRTC MediaRecorder, the other one is using a js library which generates the webm video frame by frame (webm-writer-js). The file size of the videos generated is not that different, the fast one is 60% of the slow one but the difference in speed is 1000%

  Using the basic ffmpeg syntax -i input.webm output.mp4 the files created with Chrome’s media recorder take in fact almost 10x time to be converted. The conversion logs differ slightly but overall look very similar to my novice eyes. On the left the fast conversion and on the right the slow one.

  

  The fast one throws a little error but the conversion seems successful. In the slow conversion you can see many frames processed, in the fast one as if there was only one (very fast). Using -preset veryfast cuts the speed time by half to both but the loss of quality is visible.

  Any idea how I could speed up the conversion for the videos generated by Chrome without compromising much in quality ? Thanks a lot !