Closed graphl closed 1 year ago
(1) To train the image super resolution model in pkl mode, put rendered_128
(low resolution) and rendered_256
(high resolution) together in pkl files. You code seems correct.
(2) There was a problem with your validation (testing) dataset . Try to print the value of trg_seqlen
to find the problem.
手顺
使用命令将ttf/otf文件做成数据集
步骤2:train the neural rasterizer:
使用命令
步骤3. train the image super-resolution model:
使用命令: cp -r ./data/vecfont_dataset ./data/glyphss_dataset python train_sr.py --mode train --name image_sr
出现下面错误
修改错误
在data_util/write_data_to_pkl.py文件的93行追加以下代码,解决错误
步骤4: train our main model
发生错误:
问题