Clarification on character used for style image during sampling in your experiments

[tyuukau]

Hello,

I’m trying to understand the sampling process used in your experiments. Specifically, I have a question regarding the character selection for the style image during testing.

From what I gather, during the test process, the model takes a content image of character c in style 1 and a style image of character s in style 2, and then generates an image of character c in style 2.

Could you clarify if you consistently used the same character s across multiple different styles when generating the outputs? Or did the style image vary depending on the experiment? If so, what is the character?

Thank you for your help!

[yeungchenwa]

Hello,

I’m trying to understand the sampling process used in your experiments. Specifically, I have a question regarding the character selection for the style image during testing.

From what I gather, during the test process, the model takes a content image of character c in style 1 and a style image of character s in style 2, and then generates an image of character c in style 2.

Could you clarify if you consistently used the same character s across multiple different styles when generating the outputs? Or did the style image vary depending on the experiment? If so, what is the character?

Thank you for your help!

Hi, thanks for your attention! During testing, for a fair comparison, I used the same character '蜓' in style image.