Update README.md

README.md

@@ -4,13 +4,19 @@ Implementation of A Style-Based Generator Architecture for Generative Adversaria
 
 Usage:
 
+You should prepare lmdb dataset
+
+> python prepare_data.py --out LMDB_PATH --n_worker N_WORKER DATASET_PATH
+
+This will convert images to jpeg and pre-resizes it. (For example, 8/16/32/64/128/256/512/1024) Then you can train StyleGAN.
+
 for celebA
 
-> python train.py --mixing -d {folder} PATH
+> python train.py --mixing LMDB_PATH
 
 for FFHQ
 
-> python train.py --mixing --loss r1 --sched -d {folder}
+> python train.py --mixing --loss r1 --sched LMDB_PATH
 
 ## Sample
 
@@ -22,4 +28,4 @@ I have mixed styles at 4^2 - 8^2 scale. I can't get samples as dramatic as samle
 ![Sample of the model trained on FFHQ](doc/sample_ffhq.png)
 ![Style mixing sample of the model trained on FFHQ](doc/sample_mixing_ffhq.png)
 
-Trained high resolution model on FFHQ. I think result seems more interesting.
+Trained high resolution model on FFHQ. I think result seems more interesting.