Monoup’s
I’ve been experimenting with Monoup’s image generator for a couple of weeks now, mostly to see if it can keep up with some reference shots I use in my design drafts. The results feel a bit inconsistent: sometimes it nails the lighting and texture, and then other times the skin looks too smooth or the shadows fall strangely. I’m wondering how others deal with this — do you tweak prompts endlessly, or is there some trick to getting more stable output?
11 Views
I’ve run into the same thing, especially when trying to generate portraits with tricky lighting. What helped me a bit was breaking the prompt into smaller, clearer parts instead of throwing everything in at once. For instance, when I needed a warm backlit profile shot, I first emphasized the lighting setup and only then added style notes. When I did that, the generator on monoup responded more predictably.
Another thing I’ve noticed is that the model reacts strongly to reference keywords. If I use terms like “photographic grain” or “soft diffused shadow,” it gives me closer-to-camera results. But if I mention too many artistic movements at once, the output becomes chaotic. It’s kind of like negotiating with it — the more precise I am, the less room it has to wander. I’m still trying to figure out how to avoid over-smoothing of facial details though; sometimes it looks like the model wants to erase texture by default. Curious if anyone found a workaround for that, maybe by adding noise or texture hints?