search

yeungchenwa / FontDiffuser

[AAAI2024] FontDiffuser: One-Shot Font Generation via Denoising Diffusion with Multi-Scale Content Aggregation and Style Contrastive Learning
https://yeungchenwa.github.io/fontdiffuser-homepage/
299 stars 26 forks source link

How to increase the training image size to 256 instead of the definite 96 ？ #42

Open zhangzhenhuazju opened 8 months ago

PetrByvsh commented 4 months ago

Did you manage to increase the resolution to 256. For me changing the resolutions to anything other than 96 does not work in train_phase_1.sh

PetrByvsh commented 4 months ago

this 

    style_content_feat = style_structure_features[-self.upblock_index-2]

in unet_blocks.py breaks for resolutions other than 96 since the length of style_structure_features is not the same for different resolutions