The truth about AI: Midjourney vs. Adobe Firefly

Поділитися
Вставка
  • Опубліковано 28 вер 2024

КОМЕНТАРІ • 19

  • @designisblank
    @designisblank  Рік тому +1

    This is a deep dive into the details of these two leaders in the AI Image space. Adobe is behind, what are your AI predictions?

  • @jonathanhackett2095
    @jonathanhackett2095 Рік тому

    Great video man. Really glad I came across this. Informative and well delivered.

  • @themightyflog
    @themightyflog Рік тому

    Midjourney makes great images but starting to suck. It has no control. If Midjourney doesn't get with the program they will be left behind.

    • @designisblank
      @designisblank  Рік тому

      I have noticed that it used to be more accurate in terms of the prompt, and it does seem to be moving backwards a bit. Are you using the latest version? Adobe has had a ROUGH start with text to image but has a big advantage with generative fill

  • @arreshubham
    @arreshubham Рік тому

    Totally agree but I don't think AI primetime is far away, we already have a stable diffusion control net which is far more customizable than the Firefly and all those custom ai models publicly available. AI development going too fast imo what do you think, how's the future gonna be for designers?

    • @designisblank
      @designisblank  Рік тому

      I think you’re right and adobe may not be as dominant long term. I’ve played with some of the less mainstream options and its getting better so fast. Next vid may be how a designer skillset is likely going to change, that seems inevitable.

    • @bluedragontoybash2463
      @bluedragontoybash2463 Рік тому

      Ai is going to be indispensable tool for designer. It is going harder and harder to master AI

    • @designisblank
      @designisblank  Рік тому

      @@bluedragontoybash2463 yeah it will be indispensable, what do you mean by harder to master?

    • @bluedragontoybash2463
      @bluedragontoybash2463 Рік тому

      @@designisblank When the client demand more and more specific designed it would be harder to produce specific result (beyond what controlnet can do). It is easier when the result demanded is more or less random or non specific

    • @designisblank
      @designisblank  Рік тому

      @@bluedragontoybash2463 agree. Using a combo of midjourney and adobe generative fill seems to give me a good result, but still difficult when it is ultra specific

  • @bluedragontoybash2463
    @bluedragontoybash2463 Рік тому

    Pretty soon the lawyers are the killer of AI

    • @designisblank
      @designisblank  Рік тому

      The lawyers try to kill everything haha. I’m not sure they can stop this one

    • @bluedragontoybash2463
      @bluedragontoybash2463 Рік тому

      @@designisblank Reddit is going under as we speak because, the data is being sold to feed AI. Pretty soon lawyer will guard their learning data even more tighter

  • @Goodhello369
    @Goodhello369 Рік тому

    The whole “allies” thing is bs. Of course they say that for marketing. These tools are meant to replace artists when time is right.

    • @designisblank
      @designisblank  Рік тому +1

      Actions will have to speak loudly over the next few years for sure. Everything follows money at the end of the day. Will smaller AI companies succeed?

    • @Goodhello369
      @Goodhello369 Рік тому

      @@designisblank no but the cream will get bought out and acquired by mega corps. The information i am hearing is the AI industry is super small. Like a couple blocks radius at best. But thats only going to grow and its mindblowing what is happening.

    • @designisblank
      @designisblank  Рік тому +1

      @@Goodhello369 yeah, it will be interesting that’s for sure. Hopefully it is at least not some kind of monopoly on creative…

    • @ekaterinavalinakova2643
      @ekaterinavalinakova2643 Рік тому +1

      Sure, but a big part of the appeal of firefly is that they are train off material that is either CC0 or they have permission to, like their adobe stock. One of the major criticisms is mid journey were training off random internet images and paintings.