MATT: Naštěstí máme ChatGPT aby ho vstřebal za nás! Píše se tam víc ne úplně relevantních věcí, tak jsem se ho zeptal co tam je konkrétně o rukách:
"The article explains that AI image generators have trouble with hands because they lack the understanding of the three-dimensional geometry of hands. These models can create generic images of hands, but they struggle with more complex scenarios involving context, holding objects, or specific hand gestures. Since the models don't understand the 3D nature of hands or the geometric shapes of objects they interact with, they often produce inaccurate or bizarre results."
Takže vlastně nic moc.