Converting one scene of the drama 'Witcher' to Studio Ghibli style with Stable Diffusion
An animation video that converts one scene of the drama version `` The Witcher '' distributed on Netflix into a Studio Ghibli work style using the image generation AI ``
A scene from Witcher as a Studio Ghibli animation (Stable Diffusion)-YouTube
The original is the battle scene in Blaviken. In the video below, Henry Cavill, who plays Geralt, tells us that it was a one-cut shoot.
Henry Cavill talks behind the Braviken battle scene 'The Witcher'-Netflix-YouTube
This is the image after conversion. Geralt is a silver-haired character with a dull look.
The production is 'img2img' of Stable Diffusion, and it is said that the image is converted to Studio Ghibli style for each frame. It is said that the same seed value was used when converting so that the results would be similar, but the character's face is still not consistent.
The back of Geralt walking. For some reason, a face appears on the back of the head, whether the hair is judged to be a ``beard face''.
The face on the back of the head may or may not come out, and it changes into various faces.
In the scene facing the enemy, Geralt changes into a silver-haired young man.
The enemy on the other hand has no clear face.
For some reason Geralt has changed into an uncle with glasses. Because of my silver hair, I often end up looking like an old man.
It's more like a live-action film than Ghibli.
…… I thought, and the next moment my eyes were patchy and anime-like.
Although it is not Ghibli, it has a villainous face that looks like an enemy character in anime.
During battle scenes, the movements are intense, and the faces and everything else are unstable.
In the scene where there is an enemy on the other side of the aisle, a round cat-like character that does not exist there appears. Is it a misidentification of sand dust?
Again, a mysterious creature appears at the back of Geralt's head.
Geralt (right front) who has become like a kemono character.
It seems that stable conversion is difficult.
Related Posts:
in Video, Posted by logc_nt