Recherche avancée

Sur d’autres sites (8793)

• 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 💀

  


• How to encode video with ffmpeg using AMD h264_amf

  10 novembre 2022, par Ivy Growing

  Given :

  


  

  • Win10

  


  • AMD CPU

  


  • Video capturing card Avermedia Live Gamer Extreme 3

  


  • ffmpeg versions and encoders :

  


  


  >ffmpeg.exe -encoders | find "264"
ffmpeg version 5.1-full_build-www.gyan.dev Copyright (c) 2000-2022 the FFmpeg developers
// cut
 V....D libx264              libx264 H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10 (codec h264)
 V....D libx264rgb           libx264 H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10 RGB (codec h264)
 V....D h264_amf             AMD AMF H.264 Encoder (codec h264)
 V....D h264_mf              H264 via MediaFoundation (codec h264)
 V....D h264_nvenc           NVIDIA NVENC H.264 encoder (codec h264)
 V..... h264_qsv             H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10 (Intel Quick Sync Video acceleration) (codec h264)


  


  Required to capture the video into H.264 encoded file using AMD's hardware accelerator in the CPU (AMF, or VCE).
Tried : ffmpeg -y -f dshow -rtbufsize 2002000k -framerate 30 -i video="Live Gamer EXTREME 3"  -t 00:00:10  -c:v h264_amf output.ts
Result :

  


  Input #0, dshow, from 'video=Live Gamer EXTREME 3':
  Duration: N/A, start: 88548.973998, bitrate: N/A
  Stream #0:0: Video: rawvideo (YUY2 / 0x32595559), yuyv422(tv, bt709/bt709/unknown), 1280x720, 30 fps, 30 tbr, 10000k tbn
Stream mapping:
  Stream #0:0 -> #0:0 (rawvideo (native) -> h264 (h264_amf))
Press [q] to stop, [?] for help
[h264_amf @ 000002404328c700] DLL amfrt64.dll failed to open
Error initializing output stream 0:0 -- Error while opening encoder for output stream #0:0 - maybe incorrect parameters such as bit_rate, rate, width or height
Conversion failed!


  


  For some reason ffmpeg uses resolution 1280x720... When trying to specifiy the capture card resolution the following error appears :

  


  >ffmpeg -y -f dshow -rtbufsize 2002000k -framerate 30 -video_size 3840x2160 -i video="Live Gamer EXTREME 3" -r 30 -t 00:00:10   -c:v h264_amf -f mpegts output.ts
//cut
[dshow @ 0000029d7c0f84c0] Could not set video options
video=Live Gamer EXTREME 3: I/O error


  


  This is not unique error for Avermedia card. The same error appears with Dell web cam and for Magewell.

  


  From this answer the extra flags to be used with h264_amf. I guessed the default values should be good enough. It seems something needs to be configured or initialized when using AMF/VCE.

  


  The video encoding in software (without AMF) works just fine but loads the CPU. The goal is using dedicated hardware module and release computational power of the CPU for the other apps.

  


  Command example will be appreciated.

  


• UVC webcam with ffplay outputs only noise ?

  1er mars 2024, par Abdulla Masud

  (My end goal is to use a UVC webcam with esp32 or raspberry pi. I was hoping to learn while doing some fun projects.)

  


  I have an old UVC webcam (Creative model ct6840) but I can't seem to get it to work with ffplay. I have tried looking through the documentation and other questions here but nothing is working for me. So far I have only been able to achieve a noisy-jittery output.

  


  Running ffplay -f rawvideo -video_size 670x480 /dev/video1, I get :

  


  

  


  Can someone help me understand how to make the camera work with ffplay ?

  


  The following is the information of my webcam :

  


  $ ffmpeg -f v4l2 -list_formats all -i /dev/video2

  


  [video4linux2,v4l2 @ 0x17eb3c0] Compressed: Unsupported :          GSPCA OV511 : 320x240 640x480


  


  $ v4l-info /dev/video2

  


  ### v4l2 device info [/dev/video2] ###
general info
    VIDIOC_QUERYCAP
    driver                  : "ov519"
    card                    : "USB Camera (05a9:0511)"
    bus_info                : "usb-0000:00:14.0-8.2"
    version                 : 6.1.79
    capabilities            : 0x85200001 [VIDEO_CAPTURE,?,READWRITE,STREAMING,(null)]

standards

inputs
    VIDIOC_ENUMINPUT(0)
    index                   : 0
    name                    : "ov519"
    type                    : CAMERA
    audioset                : 0
    tuner                   : 0
    std                     : 0x0 []
    status                  : 0x0 []

video capture
    VIDIOC_ENUM_FMT(0,VIDEO_CAPTURE)
    index                   : 0
    type                    : VIDEO_CAPTURE
    flags                   : 1
    description             : "GSPCA OV511"
    pixelformat             : 0x3131354f [O511]
    VIDIOC_G_FMT(VIDEO_CAPTURE)
    type                    : VIDEO_CAPTURE
    fmt.pix.width           : 640
    fmt.pix.height          : 480
    fmt.pix.pixelformat     : 0x3131354f [O511]
    fmt.pix.field           : NONE
    fmt.pix.bytesperline    : 640
    fmt.pix.sizeimage       : 614400
    fmt.pix.colorspace      : JPEG
    fmt.pix.priv            : 4276996862

controls
    VIDIOC_QUERYCTRL(BASE+0)
    id                      : 9963776
    type                    : INTEGER
    name                    : "Brightness"
    minimum                 : 0
    maximum                 : 255
    step                    : 1
    default_value           : 127
    flags                   : 48
    VIDIOC_QUERYCTRL(BASE+1)
    id                      : 9963777
    type                    : INTEGER
    name                    : "Contrast"
    minimum                 : 0
    maximum                 : 255
    step                    : 1
    default_value           : 127
    flags                   : 32
    VIDIOC_QUERYCTRL(BASE+2)
    id                      : 9963778
    type                    : INTEGER
    name                    : "Saturation"
    minimum                 : 0
    maximum                 : 255
    step                    : 1
    default_value           : 127
    flags                   : 32
    VIDIOC_QUERYCTRL(BASE+24)
    id                      : 9963800
    type                    : MENU
    name                    : "Power Line Frequency"
    minimum                 : 0
    maximum                 : 2
    step                    : 1
    default_value           : 0
    flags                   : 0
    VIDIOC_QUERYCTRL(BASE+32)
    id                      : 9963808
    type                    : BOOLEAN
    name                    : "Brightness, Automatic"
    minimum                 : 0
    maximum                 : 1
    step                    : 1
    default_value           : 1
    flags                   : 8


  


  Can someone guide me here please ? Any advice will be greatly appreciated 

  


  (P.S. the camera works perfectly with "guvcview" gtk application but since I want to use the camera with raspberry pi, I want it to work with ffplay...)