Feature Requests for Photoshop and Generative AI

I haven't used Adobe Firefly extensively, but I noticed the tip "convert my sketch to vector artwork". But, it is not clear where or how to do that. I had a look at Adobe Firefly, and added an "image to match", several of pixel art, and one of a sketch drawing of mine, and asked it:

"convert my artwork to a digital painting style, with the help of watercolors"

What it came up with was not even remotely like my original image. From a nature scene, I got a girl sitting and drawing mostly, maybe with a pet, or a single pot plant. If I'm going about the process wrong, it's not clear how to do it correctly.

I tried the same thing with generative fill in Photoshop. Highlighted an image of pixel art, and asked it:

"convert to digital painting style"

What I got was blurred paint sploshes all over the screen. 

At this point, it's not very clear to me how to use the AI to get the kind of use cases I want, so here are things I would like the AI to be able to do, to make it useful to me, and for it to be very obvious and easy about how to achieve these (having different apps for AI also makes it a bit confusing which to use, I got a pretty lost):

• Convert sketch to colour (I want to be able to do a line drawing, maybe with the base colours filled in, and have AI convert it to a nicely painted image, so I'm still involved in the design process, but instead I'm using it to save time)
• Convert art from one style to another - I do a lot of pixel art. I would love to be able to take pixel art drawings, and convert them to beautiful digital paintings, or the reverse. My main skill area is in pixel art, but I need high quality promo art, and if I could use AI to helpw ith this process, it could save me a lot of time. 
• Create sprites of a certain size. Given I'm a pixel artist, I was a bit stuck the other day and tried to use FireFly to generate me ideas for a pixel art character sprite, of dimensons 20 x 20 pixels. I then discovered it can make pixel art of sorts, but it is not good at making images of a limited size, or rather, it keeps trying to make large images. And I can't see an option to do otherwise. So as a pixel artist, or even any artist, it would nice to have options to give more directions about the size of the canvas. And be able to get it to more easily create characters without backgrounds for potential use in video games.

With all of the above, I need the ability to give the AI an image, and to be able to work from it. It also seems like it needs to be a lot better at interpreting and possibly creating pixel art, particularly in the style of a particular image / artist. 

I've noticed you've started working on 3D to 2D. I'd also like to see the reverse. I have sometimes thought it would be cool and interesting to convert my pixel art game to 3D. Being able to input pixel art, and see it converted into 3D models, for use in a game, would be really cool and could save a lot of time, especially if they were rigged for animation.