add ffhq 1024 link and scripts

README.md

@@ -21,7 +21,7 @@ Sky-Timelapse: [image generator](https://drive.google.com/file/d/16qHqShQMAdftVQ
 (FFHQ, VoxCeleb): [FFHQ image generator](https://drive.google.com/file/d/1Mw0HvFimDpdHjV1bX_VQXDxIfVKpmqC_/view?usp=sharing), [VoxCeleb](https://www.robots.ox.ac.uk/~vgg/data/voxceleb/vox1.html) <br>
 (AFHQ, VoxCeleb): [AFHQ image generator](https://drive.google.com/file/d/1dx5z0rp3UGyjLTWpf8P69DJy8JS9IPkE/view?usp=sharing), [VoxCeleb](https://www.robots.ox.ac.uk/~vgg/data/voxceleb/vox1.html), [motion generator](https://drive.google.com/drive/folders/17hnIg_KeuwK5gUSYa1sfrJAnXbOonUbO?usp=sharing) <br>
 (Anime, VoxCeleb): [Anime image generator](https://drive.google.com/file/d/1kd73588GaF1242_WPHQuIwQo2eFJQKwH/view?usp=sharing), [VoxCeleb](https://www.robots.ox.ac.uk/~vgg/data/voxceleb/vox1.html), [motion generator](https://drive.google.com/drive/folders/1X5jIUItBtsbkFDcL-gP0bby2JrG0xAZW?usp=sharing) <br>
-(FFHQ-1024, VoxCeleb): [FFHQ-1024 image generator](https://drive.google.com/file/d/1oljdihTnT53CbHrjAe2BMcbueNjH67Gd/view?usp=sharing), [VoxCeleb](https://www.robots.ox.ac.uk/~vgg/data/voxceleb/vox1.html) <br>
+(FFHQ-1024, VoxCeleb): [FFHQ-1024 image generator](https://drive.google.com/file/d/1oljdihTnT53CbHrjAe2BMcbueNjH67Gd/view?usp=sharing), [VoxCeleb](https://www.robots.ox.ac.uk/~vgg/data/voxceleb/vox1.html), [motion generator](https://drive.google.com/drive/folders/13gDzYnOXJb5t7bkDlO1MmVEFmFG9xiQt?usp=sharing) <br>
 (LSUN-Church, TLVDB): [LSUN-Church image generator](https://drive.google.com/file/d/1Vk3MEzsAl88cfpztKI2hc6jSTR41tTTw/view?usp=sharing), [TLVDB](http://people.csail.mit.edu/yichangshih/time_lapse/)
 
 Calculated pca stats are saved [here](https://drive.google.com/drive/folders/1zfowC7njgeWeb97eWMVE0VrfUaUHYrK-?usp=sharing).

script/ffhq-vox/run_evaluate_1024.sh

@@ -0,0 +1,9 @@
+python -W ignore evaluate.py \
+  --save_pca_path pca_stats/ffhq_1024 \
+  --latent_dimension 512 \
+  --style_gan_size 1024 \
+  --img_g_weights pretrained_models/stylegan2-ffhq-config-f.pt \
+  --load_pretrain_path /path/to/checkpoints \
+  --load_pretrain_epoch 0 \
+  --results results/ffhq_1024 \
+  --num_test_videos 10 \

script/ffhq-vox/run_get_stats_pca_1024.sh

@@ -0,0 +1,10 @@
+#!/usr/bin/env bash
+python get_stats_pca.py --batchSize 4000 \
+  --save_pca_path pca_stats/ffhq_1024 \
+  --pca_iterations 250 \
+  --latent_dimension 512 \
+  --img_g_weights pretrained_models/stylegan2-ffhq-config-f.pt \
+  --style_gan_size 1024 \
+  --gpu 0
+
+

script/ffhq-vox/run_train_1024.sh

@@ -0,0 +1,18 @@
+python -W ignore train.py --name ffhq_1024-voxel \
+  --time_step 2 \
+  --lr 0.0001 \
+  --save_pca_path pca_stats/ffhq_1024 \
+  --latent_dimension 512 \
+  --dataroot /path/to/voxel \
+  --checkpoints_dir checkpoints/ffhq_1024 \
+  --img_g_weights pretrained_models/stylegan2-ffhq-config-f.pt \
+  --multiprocessing_distributed --world_size 1 --rank 0 \
+  --batchSize 16 \
+  --workers 8 \
+  --w_match 5.0 \
+  --style_gan_size 1024 \
+  --total_epoch 30 \
+  --cross_domain \
+
+
+