About Render Net

[diamond0910]

Hi.

The paper says y = upsample(m) + delta m. But it seems that y is predicted by the RenderNet which takes m as one input. And I also don't find the L_mask. Is the code different from the article?

[seasonSH]

Hi zhouwy19,

The mask delta is implemented at models/semantic_stylegan.py:116-117
The mask loss is implemented at train.py:252

You may consider emailing me directly for more implementation questions.

[zhouwy19]

Hi,  Do you use the detach_texture in the LocalGenerator？ I see It's False in default, why not True？Because I think the img can just effect the texture feature not for depth feature.

…

------------------ 原始邮件 ------------------ 发件人: "seasonSH/SemanticStyleGAN" ***@***.***>; 发送时间: 2022年7月25日(星期一) 上午7:16 ***@***.***>; ***@***.******@***.***>; 主题: Re: [seasonSH/SemanticStyleGAN] About Render Net (Issue #10) Hi zhouwy19, The mask delta is implemented at models/semantic_stylegan.py:116-117 The mask loss is implemented at train.py:252 You may consider emailing me directly for more implementation questions. — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: ***@***.***>