update README

README.md

@@ -160,8 +160,7 @@ python inference.py \
 --reload_swinir --swinir_ckpt weights/general_swinir_v1.ckpt \
 --steps 50 \
 --sr_scale 4 \
---image_size 512 \
---color_fix_type wavelet --resize_back \
+--color_fix_type wavelet \
 --output results/demo/general \
 --device cuda [--tiled --tile_size 512 --tile_stride 256]
 ```
@@ -205,8 +204,7 @@ python inference.py \
 --reload_swinir --swinir_ckpt weights/general_swinir_v1.ckpt \
 --steps 50 \
 --sr_scale 4 \
---image_size 512 \
---color_fix_type wavelet --resize_back \
+--color_fix_type wavelet \
 --output results/demo/general \
 --device cuda \
 --use_guidance --g_scale 400 --g_t_start 200
@@ -238,9 +236,9 @@ Since the proposed two-stage pipeline is very flexible, you can utilize other aw
 python inference.py \
 --config configs/model/cldm.yaml \
 --ckpt [full_ckpt_path] \
---steps 50 --sr_scale 1 --image_size 512 \
+--steps 50 --sr_scale 1 \
 --input [img_dir_path] \
---color_fix_type wavelet --resize_back \
+--color_fix_type wavelet \
 --output [output_dir_path] \
 --disable_preprocess_model \
 --device cuda

inference.py

@@ -62,6 +62,9 @@ def process(
         control = model.preprocess_model(control)
     model.control_scales = [strength] * 13
     
+    if cond_fn is not None:
+        cond_fn.load_target(2 * control - 1)
+    
     height, width = control.size(-2), control.size(-1)
     shape = (n_samples, 4, height // 8, width // 8)
     x_T = torch.randn(shape, device=model.device, dtype=torch.float32)