• CrayonRosary@lemmy.world
    link
    fedilink
    English
    arrow-up
    9
    arrow-down
    5
    ·
    edit-2
    8 months ago

    Absolutely not! ChatGPT is a large language model and cannot generate images.

    ChatGPT can have a little image gen once in a while as a treat.

    • june@lemmy.world
      link
      fedilink
      English
      arrow-up
      19
      arrow-down
      2
      ·
      8 months ago

      It’s awful at text in images though. Pretty sure it draws the text rather than writes it, if that makes sense lol. I had it try 4 times and it got it wrong every time

      • just another dev@lemmy.my-box.dev
        link
        fedilink
        English
        arrow-up
        11
        arrow-down
        1
        ·
        8 months ago

        That’s GPT talking to DALL-E though - GPT is just the messenger, and has no idea what’s in the image, other than the prompt it generated for you.

        • srecko@lemm.ee
          link
          fedilink
          English
          arrow-up
          7
          arrow-down
          3
          ·
          8 months ago

          ChatGPT talks to GPT something (3 or 4 with or without turbo) and Dall-e, and ChatGPT isnt generating anything at all but that is just being pedantic for the sake of it. We all know what the OP meant.

      • fidodo@lemmy.world
        link
        fedilink
        English
        arrow-up
        7
        arrow-down
        3
        ·
        8 months ago

        The llm is executing a function on a diffusion image model. The llm does not generate the image itself

        • kelvie@lemmy.ca
          link
          fedilink
          English
          arrow-up
          9
          arrow-down
          2
          ·
          8 months ago

          This doesn’t contradict what the OP said. ChatGPT is now an interface to both an LLM and a diffusion-based image generator.

        • CrayonRosary@lemmy.world
          link
          fedilink
          English
          arrow-up
          1
          ·
          8 months ago

          ChatGPT is just a front-end that maintains a session that gets fed to an LLM each time you add a reply, and now has access to image gen, too, so I was wrong.

        • tsonfeir@lemm.ee
          link
          fedilink
          English
          arrow-up
          3
          arrow-down
          2
          ·
          8 months ago

          You’re being pedantic—and confidently ignorant. The product is called “ChatGPT” and through that you can access multiple models. Like ChatGPT 3.5, or DALL•E.

      • h3rm17@sh.itjust.works
        link
        fedilink
        English
        arrow-up
        4
        arrow-down
        3
        ·
        8 months ago

        Yeah, but the model that does the images is actually Dall-e, you are just using gpt’s interface to create them

      • Nexz@feddit.nl
        link
        fedilink
        English
        arrow-up
        3
        arrow-down
        2
        ·
        8 months ago

        I mean, the GPT model is a LLM and ChatGPT uses DALL-E in the background to create images. So depending on definition you’re both correct :-)

        • tsonfeir@lemm.ee
          link
          fedilink
          English
          arrow-up
          1
          arrow-down
          1
          ·
          8 months ago

          Depending on how I define anything means I’m always correct I guess. 🤷‍♂️