Myeh, they're clearly AI-generated but that's not my complaint here - the absolute inconsistency in style and ultimate lack of 'overall style' or visual grammar's what bugs here. It may as well be a poorly-considered selection of clip art, from multiple disparate sources as used to be a problem back in the 90s.
It'll take a while for people to get used to considering graphic generation as more than mere visual placeholders, I suppose. Early days.
Yes, I struggled to get a semblant of similar styles between the pictures, despite the same suffixes or playing with 'remixes'.
Sometimes, adding one extra word to the prompt (even though it doesn't describe the style) will change the whole visual style or change the thing you're trying to highlight.
It was my first time with Midjourney, I'm sure there are tricks to improve on that. Might revisit the images, I do agree that the styling inconsistencies don't look too good.
As a 'sometime designer' who has dabbled in MJ quite a lot, I'd suggest taking the outputs and then running them through something like Photoshop and tweaking them in there to get them to somehow match a bit more - eg. you could add a texture, or halftone them, or alter the hue and saturation to standardise them a bit. Whilst MJ can texturise things, it does tend to fall over a bit when doing so.
Thank you! I did spent a substantial amount of time playing with prompts, variations and upscales to try and get decent results. The space elevator one for example, way harder than I thought.
