Recherche avancée

Sur d’autres sites (8546)

• Revision 3130 : La page d’inscription ne doit être visible que si elle est autorisée dans ...

  24 mars 2010, par kent1 — Log

  La page d’inscription ne doit être visible que si elle est autorisée dans la configuration de SPIP

• Bootstrapping an AI UGC system — video generation is expensive, APIs are limiting, and I need help navigating it all [closed]

  24 juin, par Barack _ Ouma

  I’m building a solo AI-powered UGC (User-Generated Content) platform — something that automates the creation of short-form content using AI avatars, voices, visuals, and scripts. But I’ve hit a wall with video generation and API limitations.

  


  So far, I’ve integrated TTS and voice cloning (using ElevenLabs), and I’ve gotten image generation working. But video generation (especially talking avatars) has been a nightmare — both financially and technically.

  


  🛠️ Features I’m trying to build :

  


  AI avatars (face + lip-syncing)
Script generation (LLM-driven)
Image generation
Video composition

  


  I’m trying to build an AI faceless content creation automtion platform alternative to Makeugc.com or Reelfarm.org or postbridge.com — just trying to create a working pipeline for automated content.

  


  ❌ Challenges so far :

  


  Services like D-ID, Synthesia, Magic Hour, and Luma are either paywalled, have no trials, or are very expensive.

  


  D-ID does support avatar creation, but you need to pay upfront to even access those features. There's no easy/free entry point.

  


  Tools like Google Veo 3 are powerful but clearly not accessible for indie builders.
I’ve looked into open-source models like WAN 2.1, CogVideo, etc., but I have no clue how to run them or what infra is needed.

  


  Now I’m torn between buying my own GPU or renting compute power to self-host these models.

  


  💸 Cost is a huge blocker

  


  I’ve been looking through Replicate’s pricing, and while some models (especially image gen) are manageable, video models get expensive fast. Even GPU rental rates stack up quickly, especially if you’re testing often or experimenting with pipelines. Plus, idle time billing doesn’t help.

  


  💭 What I could really use help with :

  


  Has anyone successfully stitched together APIs (voice, avatar, video) into a working UGC pipeline ?

  


  Should I use separate services (e.g. ElevenLabs + Synthesia + WAN) or try to host my own end-to-end system ?

  


  Is it cheaper (long term) to buy a used GPU like a 4090 and run things locally ? Or better to rent compute short-term ?

  


  Any open-source solutions that are beginner-friendly or have minimal setup ?
Any existing frameworks or wrappers for UGC media pipelines that make all this easier ?

  


  I’ve spent weeks researching, testing APIs, and hitting walls — and while I’ve learned a lot, I’d really appreciate any guidance from folks who’ve been here before.
Thanks in advance 🙏

  


  And good luck to everyone else trying to build with AI on a budget — this stuff isn’t as plug-and-play as it looks on launch videos 💀

  


• In ffmpeg : Can't get peak bitrate values within 10% error tolerance for HLS playlist files

  5 octobre 2019, par ALS20394

  I’m using ffmpeg to produce several HLS variant playlists from an .mp4 file. When I check master.m3u8 file with mediastreamvalidator I get :

  Error : Measured peak bitrate compared to master playlist declared value exceeds error tolerance

  I understand that the error percentage needs to be less than 10%, and figured it out on 1 variant but not the three others. I’ve spent quite a bit of time adjusting the -maxrate and the -buffsize but the error percentage change is minimal. Beginning to wonder if I’m misunderstanding something ?

  This is the latest of what I enter :

  ffmpeg -i FHVid.mp4 \
  
  -b:v:0 5000k -maxrate 5250k -bufsize 5500k -profile:v main -c:v h264 -crf 20 -sc_threshold 0 -g 48 \
  
  -b:v:1 2800k -maxrate 2940k -bufsize 3100k -profile:v main -c:v h264 -crf 20 -sc_threshold 0 -g 48 \
  
  -b:v:2 1400k -maxrate 1540k -bufsize 1700k -profile:v main -c:v h264 -crf 20 -sc_threshold 0 -g 48 \
  
  -b:v:3 800k -maxrate 840k -bufsize 1050k -profile:v main -c:v h264 -crf 20 -sc_threshold 0 -g 48 \
  
  -b:a:0 192k \
  
  -b:a:1 128k \
  
  -b:a:2 128k \
  
  -b:a:3 96k \
  
  -c:a aac -ar 48000 -keyint_min 48 -map 0:v -map 0:a -map 0:v -map 0:a -map 0:v -map 0:a -map 0:v -map 0:a \
  
  -f hls -var_stream_map "v:0,a:0  v:1,a:1 v:2,a:2 v:3,a:3" \
  
  -master_pl_name FHVidmaster.m3u8 -hls_time 4 -hls_playlist_type vod \
  
  -hls_segment_filename 'file_%v_%03d.ts' out_%v.m3u8

  but I’ve also tried max rates that seem to be more standard

  -b:v:0 5000k -maxrate 5500k -bufsize 6500k
  
  -b:v:1 2800k -maxrate 3080k -bufsize 3200k
  
  -b:v:2 1400k -maxrate 1540k -bufsize 1900k
  
  -b:v:3 800k -maxrate 880k -bufsize 1050k

  The latest error message :

  Error: Measured peak bitrate compared to master playlist declared value exceeds error tolerance
  
  --> Detail:  Measured: 1111.54 kb/s, Master playlist: 1680.80 kb/s, Error: 33.87%
  
  --> Source:  /Users/Bun/Documents/CODING/CosmicPerspectiveAssets/01-FalconHeavy/HLS/FHVidmaster.m3u8
  
  --> Compare: out_2.m3u8
  
  
  
  --> Detail:  Measured: 1178.21 kb/s, Master playlist: 5711.20 kb/s, Error: 79.37%
  
  --> Source:  /Users/Bun/Documents/CODING/CosmicPerspectiveAssets/01-FalconHeavy/HLS/FHVidmaster.m3u8
  
  --> Compare: out_0.m3u8
  
  
  
  --> Detail:  Measured: 1109.03 kb/s, Master playlist: 3220.80 kb/s, Error: 65.57%
  
  --> Source:  /Users/Bun/Documents/CODING/CosmicPerspectiveAssets/01-FalconHeavy/HLS/FHVidmaster.m3u8
  
  --> Compare: out_1.m3u8

  Any help would be greatly appreciated on the -maxrate and -buffsize for variant playlists 0, 1, and 2. No adjustment I make seems to make any difference.