I’m looking at all the text 2 image generators, and i can only find the few with Flux in them to render words correctly, but the other models sare most peferable. any way to get them to spell correctly? murder

Using this in my prompts: Only These Letters “JUNGLISM”, In this Sequence “JUNGLISM” “J U N G L I S M”, “J + U + N + G + L + I + S + M”

I have these in my neg prompt: incorrect spelling, bad spelling, error spelling, typo

and it gets all silly. any advice by chance? TIA

  • VioneT@lemmy.worldMEnglish
    2·
    8 days ago

    The models in Perchance are quite old. Getting to generate ‘text’ in the images are likely impossible with the current models. I would suggest opting to use other image generation services instead.

    • 1234567ATEUP@lemmy.worldOPBEnglish
      1·
      6 days ago

      now it just keeps pumping out black chicks with dreads. halfway decent cartoony black chicks with dreads, nothing like what i input. and i didnt go black, i dont think. SFW

      • VioneT@lemmy.worldMEnglish
        1·
        6 days ago

        Are you using a lot of commas. The new model works best in descriptive instruction, and not comma separated tags or keywords.

  • allo@lemmy.worldEnglish
    2·
    7 days ago

    edit: responded b4 knowing perchance img update just happened. :)

    • 1234567ATEUP@lemmy.worldOPBEnglish
      1·
      6 days ago

      now it just keeps pumping out black chicks with dreads. halfway decent cartoony black chicks with dreads, nothing like what i input. and i didnt go black, i dont think. SFW