Upscaling Images in Python with ESRGANs

Поделиться
HTML-код
  • Опубликовано: 11 янв 2025

Комментарии • 16

  • @elu1
    @elu1 6 часов назад +1

    This is nice and useful. Thanks for the teaching!

  • @Solimanmohammed-hm9lb
    @Solimanmohammed-hm9lb 7 часов назад +1

    Awesome work. I appreciate your efforts ❤

  • @alvinlim86
    @alvinlim86 9 часов назад +4

    Thanks, i believe this model works well for photos, although more cartoonish. is this a one-size-fits-all approach?
    could you recommend other models for other images, based on your experience or opinion?
    images: maps, svg, text, graphs, x-ray scans, microscope (mildly transparent, with mild noise)

  • @AceOnlineMath
    @AceOnlineMath 10 часов назад +1

    auto compleate can be so annoying lol. Great video as allways. Thank you

  • @santhosh-j7e
    @santhosh-j7e 9 часов назад +2

    What if I pass the upscaled output image back to the same model and upscale it?

    • @thomasgoodwin2648
      @thomasgoodwin2648 9 часов назад +1

      You get back an x16 image. Wash rinse, repeat as needed. At some point you'll just start to magnify the defects in the original image encoding. This gets worse the lower resolution image you start with. JPGs are usually worse for this than PNGs as a matter of practice.

  • @methylithium
    @methylithium 11 часов назад +2

    thanks for tutorial! so helpful!

    • @numbah16
      @numbah16 11 часов назад

      ^ Comments on how helpful the 10-minute video is 1 minute after it's posted. Righttttt

  • @BALAHANUMAN
    @BALAHANUMAN 7 часов назад

    Great video. Could you please make a tutorial on speech recognition using our own custom voice samples?

  • @thomasgoodwin2648
    @thomasgoodwin2648 9 часов назад +1

    There are many many .pth model files out there without a lot of documentation. How do I know find out the correct parameters to use setting up the models? Is there some way to glean that info from the model file itself?
    I've been struggling with this stuff for quite a while now. Till now, the easiest way for me has been running open-cv upscale models. I know there are better ways (open cv is .. limited, at best), I was just never quite 🧠enough to grok it well enough to get past the inevitable slew of exceptions and plain old lack of understanding on my part. (Hey, I'm just an old skool code bum.)
    I would 💓 a deeper dive into the whole .pth and/or .onnx model systems (sorry, but the big can of worms is now open...) .
    ☮, 💓, n 🎸,
    🖖😎👍

  • @blacksundarkskies
    @blacksundarkskies Час назад

    do a video on whisper to to create subtitles of old japanese anime, thing is like magic

  • @gonzalolarios2755
    @gonzalolarios2755 35 минут назад

    Please keep the auto translation of your video title, it’s nice knowing the main topic in my own language so I could talk it with others

  • @subhanullahadelyar
    @subhanullahadelyar 11 часов назад +2

    Amazing

  • @Der_Rotsteiner
    @Der_Rotsteiner 5 часов назад +3

    Please turn off the auto translation of your video title

  • @RahulVyas-ds4uv
    @RahulVyas-ds4uv 8 часов назад

    Can you make a video on some important module of python ❤❤