So, as we have been playing around with Text to image programs we have found that sitting is an issue. We're wondering if like with hands it's not a proportions thing? The task of not only creating a human body...hard enough on it's own. But then to add creating a chair, barstool, etc... that matches in size.
The image groupings below display how things got better when we gave AI better or should we say more familiar item to create. For example: when we typed in Large woman sitting in chair. AI created rather awkward matching human to chair ratios.....Barstools were a little off as well. ( guess AI doesn't drink much ). Yet when we gave it the prompt for sitting on the edge of bathtub.... They turned out to be the best of the grouping.
Click the buttons below to view the matching images:
Sitting in Chair
Click any Image to open larger views:
