When the system generates keyframes with an uploaded character shown full height, if often reduces the character's overall height within a scene, effectively squashing the character so they look like a shorter, wider version of themselves. This immediately and significantly reduces the character consistency and believabilty. In addition, it is often these full height scenes where the system begins to produce more CGI/animated look characters. I'm not sure why the system can't just shrink down the original upload image - this can be easily done in photoshop with no loss of character shape conformity - why can't the system do it? I have uploaded a screenshot from my latest video showing a 'squashed' character in half silhouette on the right, and the original character shrunk down to the same size in photoshop. Notice the difference in width of the body and size of the head in particular. Add in logic to check for 'squashed' full height characters and correct.