You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The prompt enhancer has a random choice of style, which can contradict the input image and the user prompt. If a user prompts for a real-life video, and it adds "The scene is computer-generated imagery.", then it becomes a contradictory prompt. It took me a while to work out that it was this node that was sometimes causing bad/unexpected results, so I then had to remove it and use my own solution.
Probably caused by this part:
def _get_random_scene_type():
"""
Randomly select a scene type to add to the prompt.
"""
types = [
"The scene is captured in real-life footage.",
"The scene is computer-generated imagery.",
"The scene appears to be from a movie.",
"The scene appears to be from a TV show.",
"The scene is captured in a studio.",
]
return random.choice(types)
This should either be removed or a switch option added to the node to toggle random styles.
The text was updated successfully, but these errors were encountered:
The prompt enhancer has a random choice of style, which can contradict the input image and the user prompt. If a user prompts for a real-life video, and it adds "The scene is computer-generated imagery.", then it becomes a contradictory prompt. It took me a while to work out that it was this node that was sometimes causing bad/unexpected results, so I then had to remove it and use my own solution.
Probably caused by this part:
This should either be removed or a switch option added to the node to toggle random styles.
The text was updated successfully, but these errors were encountered: