I don't think this is going in the rigbt direction....

I'm bringing to discussion an issue that was probably already discussed somewhere, and its about how Firefly is lagging hard behind other Text to Image Generators. In this case, ChatGPT.
Firefly and Adobe are my preferred work tools. I am paying to use them and I'm always hopeful that the tech evolves in the right direction. That the way the AI engine eats and chews the prompts is effective enough that I don't have to spend an awful amount of time refining prompts to get satisfactory results.
Firefly is armed with a vast array of style options and settings, and along with the prompt refining, it should do the trick in getting great results with little effort, but I keep getting bad results.
At the same time, I go into ChatGPT, and with a single prompt, no extra settings and presets, I get what I want at first try. I'm not paying for ChatGPT so I have a very limited daily amount of images I can generate.
Case in point, I'll give you the following example, from a series of personal things I've been goofing around in my spare free time.
With ChatGPT, and using the following prompt:
"can I have a vintage, pulp comic style illustration of a woman cutting her husband's hair.
the man is sitting on a chair, and the woman behind him, cutting his hair with a scissor. the hair of the husband is all messed up, lots of bald spots and weird cuts, because she is a terrible hair dresser.
the woman is clearly happy and having a great time cutting his hair, but the man looks miserable because his hair is all messed up.
it has to have that vintage style illustration look, vertical format."
The immediate result was this:
Without skipping a beat, the engine gave EXACTLY what I was looking for. I didn't have to refine the prompt, give examples or references, no use of presets or settings. Just the response to the prompt and it nailed it. The only thing I had to edit was remove the extra finger holding the scissors.

Now... Using the EXACT SAME PROMPT on Firefly, with no settings other than using the Portrait format, this is one of the results (the others are equally bad):

So. Not only the style and the look do not match in any way what is described in the prompt. The image itself, as usual is ridden with bizarre artifacts, distorted elements and outright weird stuff going on.
None of the results was even remotely satisfactory, and this frustrates me beyond reason.

I know these things use different learning models, but Adobe is supposed to be a tool for artists and designers who usually have a pretty good idea of what they are looking for. I try to make my prompts as specific as possible, and still I stumble in results that are frankly embarassing from a provider such as Adobe.

There are situations where Firefly outright ignores parts of the prompt that are well described. It simply ignores them and gives me back stuff that is not even in the prompt. This is beyond baffling...

I keep reporting bad results when I get them. I've come in here several times reporting bugs and trying to understand why Adobe is lagging so much behind other tools, when this was supposed to be a top of the shelf, state of the art product.

Sorry for the rant. But I needed to get this out there.