search

ZYM-PKU / UDiffText

UDiffText: A Unified Framework for High-quality Text Synthesis in Arbitrary Images via Character-aware Diffusion Models
MIT License
192 stars 16 forks source link

Does it work withSD-Turbo? #6

Open Shengguang-Zhou opened 8 months ago

Shengguang-Zhou commented 8 months ago

Also, any plan to integrate with diffuser?

I can here for help working around for those two features

ZYM-PKU commented 8 months ago
  1. We can expect a faster inference stage using SD-turbo, but the model must be re-trained based on the new structure.
  2. The codes need a major revision to be transferred to the diffuser framework. It will be great if you can help.
Shengguang-Zhou commented 8 months ago

  1. We can expect a faster inference stage using SD-turbo, but the model must be re-trained based on the new structure.

  2. The codes need a major revision to be transferred to the diffuser framework. It will be great if you can help.

Will do in recent. It's quite bit an important project for my research. I'm planning to integrated in diffuser as well and trained Chinese characters.

May I have you WeChat or gmail for further contact?

ZYM-PKU commented 8 months ago

My email address is 82540602@qq.com.