4 ms·
It isn't going to impress the person in the street until it actually follows your instructions. I tried several times to express "a tall three-legged stool" but
by fernly 4y ago
It isn't going to impress the person in the street until it actually follows your instructions. I tried several times to express "a tall three-legged stool" but even with the "CFG" (how much the image will be like your prompt) at max, it gave me stools with four or ultimately, two legs. Also tried "a four-legged spider" (don't ask) and got first an eight-legged spider, and next, a spider with eight legs, but four of them were blurred. Sure, dumb, pedestrian requests, no imagination, but a five-year-old would quickly get impatient with its inability to follow simple directions.
- bogwog 4y agoAt the pace that these things are progressing, I bet we’ll have that kind of control in like a year or less
- astrange 4y agoYou could sketch a three legged stool in MSPaint and use img2img. I think looking at text prompts as an essential part of the technology is very limiting. You could be using a mix of text, storyboards, images of objects to place inside it, sketches of the desired layout, etc… in fact why not replace your game's renderer with it?
- jmiskovic 4y ago> in fact why not replace your game's renderer with it? What astrange idea, surely it would never work? /winkwink