Thomas Pang

How about the training image size and style image size? in the procedure of "Prepare a dataset", input images are all resized to size no more than 256, is this...

yes, that's a problem.

@HichameMoriceau thanks for help, i will try it!

training parameters: -model johnson -image_size 512 -style_size 512 -content_weight 1 -style_weight 50 -learning_rate 0.001 -normalize_gradients true -tv_weight 0.000085 dataset is ms coco

@SantaFlamel yes, exactly

make style_weight smaller such as 20 or 10, and i trained for 30000+ iterations

@markz-nyc , use cpu mode, with huge memory

@markz-nyc 4K size image takes me more than 70g memory

@0000sir not now, i tried training with large size style(such as size 1080) and get some larger scale but not satisfactory exactly, meanwhile you can not get so much video...

@0000sir i did not change any code of repo, i use follow param to train `th train.lua -data ./datasets -style_image input.jpg -model johnson -image_size 512 -style_size 512 -content_weight 1 -style_weight...