• Truscape@lemmy.blahaj.zone
      link
      fedilink
      arrow-up
      5
      ·
      1 day ago

      Upscalers usually don’t have context for text or smaller graphics, so the patches don’t look much better. Hopefully OP can provide a better source.

      • howrar@lemmy.ca
        link
        fedilink
        arrow-up
        1
        arrow-down
        1
        ·
        17 hours ago

        If the upscaler was trained on data that contains that exact patch in other contexts, it’ll theoretically be capable of upscaling it properly.

      • Draupnir@lemmy.world
        link
        fedilink
        English
        arrow-up
        1
        arrow-down
        1
        ·
        1 day ago

        Yeah that’s fine, but by doing so someone else will have an easier time recognizing the image or matching the details to their memory