ChatGPT fails at tasks in image processing. I tried to use a sprite image of about 19 x 7 sprites with a character as input and it returned a similar image with the same number of sprites but modified according to the 3D position of the character in each sprite
After a few days, until a week with the use of its limitation, I didn't manage to take any images at all.
Grok, however, managed to correctly go through the changes in proportion of 88% of the number of sprites, the other 12% are approximately 80%, in four queries, it seems that from the third query he started to forget and come up with a simplistic combination with a few characters, I came back with the input with the image and I barely got this percentage. The whole process took less than four minutes with my typing.
See the result of the Grok Beta :
