Hi, thanks for sharing the code and paper!
I have a question about the text-guided image weather background editing stage in Section 3.1. The paper says that Qwen-Image is used with carefully designed prompts to generate multi-weather image sequences.
Could you provide an example of the prompts used for target weather conditions, such as rain, snow, or fog?
I am especially interested in whether the prompts include explicit constraints to preserve the original scene content, such as road layout, vehicles, pedestrians, lane markings, buildings, traffic signs, and camera viewpoint. Also, do you use different prompts for different weather severities, such as light rain vs. heavy rain?
Thanks!
Hi, thanks for sharing the code and paper!
I have a question about the text-guided image weather background editing stage in Section 3.1. The paper says that Qwen-Image is used with carefully designed prompts to generate multi-weather image sequences.
Could you provide an example of the prompts used for target weather conditions, such as rain, snow, or fog?
I am especially interested in whether the prompts include explicit constraints to preserve the original scene content, such as road layout, vehicles, pedestrians, lane markings, buildings, traffic signs, and camera viewpoint. Also, do you use different prompts for different weather severities, such as light rain vs. heavy rain?
Thanks!