The forum does not like to me to have separate posts (even in my own thread) and I can only have three in succession, but I like to keep different experiments separate, because I think that is easier to read.
Anyway, one obvious limitation is the length of AI-generated videos, so how to create something longer? One way is to take the last frame of a video and use that as reference frame for the next video. I’ve done that previously with mixed success but yesterday I created a 40-second video from 4 10-second segments, which turned out well.
Here is the initial image:
It was a result where Seedream deviated from the composition of the original iClone render of the bar scene, but I decided to use it anyway.
Here is the final video, generated with Kling 2.6:
(For some reason the thumbnail for the video is not showing)
I used the following prompts:
PART 1:
“A bar scene as in @image 1 with one man and one woman. Through the window we see an urban setting.
The camera is focused on the young woman sitting at the bar, who is dressed in a revealing top, short skirt and boots.
The young woman is holding a clear glass with red wine. She has a serious expression.
A young man stands on the left, watching the young woman.
After a few moments, the young man walks up to the young woman and stands there.
The young woman ignores the young man as in her own world.
Meanwhile, the young woman takes a sip from the wine and savors it.
The young woman puts the glass back on the bar surface and smiles with satisfaction.
The young woman then looks at the young man with a warm smile and they kiss.”
PART 2:
“Bar scene as in @image 1 with a young woman and a young man kissing. After a few seconds the young woman turns to her glass of wine and takes a sip. The young man looks disappointed.”
PART 3:
“Bar scene as in @image 1 with a young woman holding a glass of wine and a young man looking disappointed.
The young woman puts down the glass, smiles and turns to the young man. She embraces him and they kiss.”
PART 4:
“Bar scene with a young woman and a young man kissing as in @Image 1. After a few moments they end their embrace and then walk away together to leave the bar. The camera follows them.”
The parts fit well, but it is noticeable that the characters become a bit more cartoonish and that the text on the man’s t-shirt gets distorted. With an option for additional reference images this could be avoided.
Sound is generated as part of the video, but I also added additional AI-generated sound.
What is interesting is how the AI interpret the prompts. The young man looks disappointed when the young woman turns away to have a sip of wine, but then she turns back and says: “Don’t be sad.” I didn’t tell her to say that so it all came from the AI.