G
Guest 7037866
Guest
Yeah, it takes a bit of tinkering, and what works well with one AI tool often doesn't work at all with another.Heheh, I've tried various standing on things before (including, for the Rocket power armour pics, standing on a ruined building) but at best I only got 50/50 -- perhaps since the ruined building part could be inferred through context (of other ruined bits around). But for a white floor (and marble slab) there ain't much that can show that without actually showing that...
I've also found describing footwear or how a figure's feet are placed helps a bit, too. I've also used prompts like "full length" and "wide angle", sometimes combined nearly all of them until I get something that works. But, the nice thing is once I find the right combo, it tends to work whenever I generated new images of the same prompt.
