FF

Start with "A greyscale panel on a black background, in FF style, in high resolution, featuring" when the image is in black and white, or "A color panel on a white background, in FF style, in high resolution, featuring" when the image is in color.

When there is a speech bubble, specify where the tail points to.

Specify the view of characters (front, side, etc).
Don't specify the langage used in texts.
Do not correct spelling.
Avoid any description of style or medium.
Output a concise literal description.
Only describe things that are obvious from the image.




PP

Start with "A black and white panel on a white background in PP style, in high resolution, featuring".

When there is text in a rectangle on the top of the panel, name this rectangle a "narration box".

When there is a speech bubble, specify where the tail points to.

When there is a figure with curly short black hair, replace their physical description by the name of the character: "Zviane"

Specify the view of characters (front, side, etc).
Don't specify the langage used in texts.
Do not correct spelling.
Avoid any description of style or medium.
Output a concise literal description.
Only describe things that are obvious from the image.



***

SUGGESTIONS DE NIK:


- train a model in french (c’est plus précis comme description) (use my words)
- avoid numbers and dash and stuff that are not in natural language
- colle le nom et le nom de famille
- colle thirdstip halfstrip (thirder, halfer, fuller)

Output a concise literal description. (empêche les choses trop vagues)
Only describe things that are obvious from the image.

plug-in eagle: AI autotagger
zup pluggin: ask nik