Cleanup in inference script, other minor changes

Code/Form1.cs

@@ -228,8 +228,11 @@ namespace Flowframes
         {
             UpdateOutputFPS();
             int guiInterpFactor = interpFactorCombox.GetInt();
-            if (!Program.busy && guiInterpFactor > 2 && !GetAi().supportsAnyExp && Config.GetInt("autoEncMode") > 0)
-                Logger.Log($"Warning: {GetAi().aiName.Replace("_", "-")} doesn't natively support 4x/8x and will run multiple times for {guiInterpFactor}x. Auto-Encode will only work on the last run.");
+            if (!initialized)
+                return;
+            string aiName = GetAi().aiName.Replace("_", "-");
+            if (!Program.busy && guiInterpFactor > 2 && !GetAi().supportsAnyExp && Config.GetInt("autoEncMode") > 0 && !Logger.GetLastLine().Contains(aiName))
+                Logger.Log($"Warning: {aiName} doesn't natively support 4x/8x and will run multiple times for {guiInterpFactor}x. Auto-Encode will only work on the last run.");
         }
 
         public void SetWorking(bool state, bool allowCancel = true)

Code/IO/Logger.cs

@@ -95,6 +95,8 @@ namespace Flowframes
         public static string GetLastLine ()
         {
             string[] lines = textbox.Text.SplitIntoLines();
+            if (lines.Length < 1)
+                return "";
             return lines.Last();
         }
 

Code/OS/AiProcess.cs

@@ -169,7 +169,7 @@ namespace Flowframes
             string rifeDir = Path.Combine(Paths.GetPkgPath(), Path.GetFileNameWithoutExtension(Packages.rifeCuda.fileName));
             string script = "inference_video.py";
             string uhdStr = InterpolateUtils.UseUHD() ? "--UHD" : "";
-            string args = $" --img {framesPath.Wrap()} --exp {(int)Math.Log(interpFactor, 2)} {uhdStr} --imgformat {InterpolateUtils.GetOutExt()} --output {Paths.interpDir}";
+            string args = $" --input {framesPath.Wrap()} --exp {(int)Math.Log(interpFactor, 2)} {uhdStr} --imgformat {InterpolateUtils.GetOutExt()} --output {Paths.interpDir}";
 
             if (!File.Exists(Path.Combine(rifeDir, script)))
             {

Pkgs/rife-cuda/inference_video.py

@@ -4,13 +4,11 @@ import cv2
 import torch
 import argparse
 import numpy as np
-#from tqdm import tqdm
 from torch.nn import functional as F
 import warnings
 import _thread
 import skvideo.io
 from queue import Queue, Empty
-#import moviepy.editor
 import shutil
 warnings.filterwarnings("ignore")
 
@@ -22,8 +20,8 @@ print("Added {0} to temporary PATH".format(dname))
 sys.path.append(dname)
 
 device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+torch.set_grad_enabled(False)
 if torch.cuda.is_available():
-    torch.set_grad_enabled(False)
     torch.backends.cudnn.enabled = True
     torch.backends.cudnn.benchmark = True
 else:
@@ -37,21 +35,13 @@ except:
     print("Failed to get hardware info!")
 
 parser = argparse.ArgumentParser(description='Interpolation for a pair of images')
-parser.add_argument('--video', dest='video', type=str, default=None)
-parser.add_argument('--img', dest='img', type=str, default=None)
+parser.add_argument('--input', dest='input', type=str, default=None)
 parser.add_argument('--output', required=False, default='frames-interpolated')
 parser.add_argument('--imgformat', default="png")
-parser.add_argument('--montage', default=False, dest='montage', action='store_true', help='montage origin video')
 parser.add_argument('--UHD', dest='UHD', action='store_true', help='support 4k video')
-parser.add_argument('--skip', dest='skip', default=False, action='store_true', help='whether to remove static frames before processing')
-parser.add_argument('--fps', dest='fps', type=int, default=None)
-parser.add_argument('--png', dest='png', default=True, action='store_true', help='whether to vid_out png format vid_outs')
-parser.add_argument('--ext', dest='ext', type=str, default='mp4', help='vid_out video extension')
 parser.add_argument('--exp', dest='exp', type=int, default=1)
 args = parser.parse_args()
-assert (not args.video is None or not args.img is None)
-if not args.img is None:
-    args.png = True
+assert (not args.input is None)
 
 from model.RIFE_HD import Model
 model = Model()
@@ -59,42 +49,23 @@ model.load_model(os.path.join(dname, "models"), -1)
 model.eval()
 model.device()
 
-path = args.img
+path = args.input
 name = os.path.basename(path)
 interp_output_path = (args.output).join(path.rsplit(name, 1))
 print("\ninterp_output_path: " + interp_output_path)
 
-if not args.video is None:
-    videoCapture = cv2.VideoCapture(args.video)
-    fps = videoCapture.get(cv2.CAP_PROP_FPS)
-    tot_frame = videoCapture.get(cv2.CAP_PROP_FRAME_COUNT)
-    videoCapture.release()
-    if args.fps is None:
-        fpsNotAssigned = True
-        args.fps = fps * (2 ** args.exp)
-    else:
-        fpsNotAssigned = False
-    videogen = skvideo.io.vreader(args.video)
-    lastframe = next(videogen)
-    fourcc = cv2.VideoWriter_fourcc('m', 'p', '4', 'v')
-    video_path_wo_ext, ext = os.path.splitext(args.video)
-    print('{} frames in total'.format(tot_frame))
-else:
-    videogen = []
-    for f in os.listdir(args.img):
-        if 'png' in f:
-            videogen.append(f)
-    tot_frame = len(videogen)
-    videogen.sort(key= lambda x:int(x[:-4]))
-    lastframe = cv2.imread(os.path.join(args.img, videogen[0]))[:, :, ::-1].copy()
-    videogen = videogen[1:]    
+videogen = []
+for f in os.listdir(args.input):
+    if 'png' in f:
+        videogen.append(f)
+tot_frame = len(videogen)
+videogen.sort(key= lambda x:int(x[:-4]))
+lastframe = cv2.imread(os.path.join(args.input, videogen[0]))[:, :, ::-1].copy()
+videogen = videogen[1:]    
 h, w, _ = lastframe.shape
 vid_out = None
-if args.png:
-    if not os.path.exists(interp_output_path):
-        os.mkdir(interp_output_path)
-else:
-    vid_out = cv2.VideoWriter('{}_{}X_{}fps.{}'.format(video_path_wo_ext, args.exp, int(np.round(args.fps)), args.ext), fourcc, args.fps, (w, h))
+if not os.path.exists(interp_output_path):
+    os.mkdir(interp_output_path)
     
 def clear_write_buffer(user_args, write_buffer):
     cnt = 1
@@ -102,20 +73,14 @@ def clear_write_buffer(user_args, write_buffer):
         item = write_buffer.get()
         if item is None:
             break
-        if user_args.png:
-            print('=> {:0>8d}.{}'.format(cnt, args.imgformat))
-            cv2.imwrite('{}/{:0>8d}.{}'.format(interp_output_path, cnt, args.imgformat), item[:, :, ::-1])
-            #cv2.imwrite('vid_out/{:0>7d}.png'.format(cnt), item[:, :, ::-1])
-            cnt += 1
-        else:
-            vid_out.write(item[:, :, ::-1])
+        print('=> {:0>8d}.{}'.format(cnt, args.imgformat))
+        cv2.imwrite('{}/{:0>8d}.{}'.format(interp_output_path, cnt, args.imgformat), item[:, :, ::-1])
+        cnt += 1
 
 def build_read_buffer(user_args, read_buffer, videogen):
     for frame in videogen:
-        if not user_args.img is None:
-            frame = cv2.imread(os.path.join(user_args.img, frame))[:, :, ::-1].copy()
-        if user_args.montage:
-            frame = frame[:, left: left + w]
+        if not user_args.input is None:
+            frame = cv2.imread(os.path.join(user_args.input, frame))[:, :, ::-1].copy()
         read_buffer.put(frame)
     read_buffer.put(None)
 
@@ -128,9 +93,6 @@ def make_inference(I0, I1, exp):
     second_half = make_inference(middle, I1, exp=exp - 1)
     return [*first_half, middle, *second_half]
 
-if args.montage:
-    left = w // 4
-    w = w // 2
 if args.UHD:
     print("UHD mode enabled.")
     ph = ((h - 1) // 64 + 1) * 64
@@ -139,10 +101,7 @@ else:
     ph = ((h - 1) // 32 + 1) * 32
     pw = ((w - 1) // 32 + 1) * 32
 padding = (0, pw - w, 0, ph - h)
-#pbar = tqdm(total=tot_frame)
 skip_frame = 1
-if args.montage:
-    lastframe = lastframe[:, left: left + w]
 
 write_buffer = Queue(maxsize=200)
 read_buffer = Queue(maxsize=200)
@@ -159,44 +118,20 @@ while True:
     I0 = I1
     I1 = torch.from_numpy(np.transpose(frame, (2,0,1))).to(device, non_blocking=True).unsqueeze(0).float() / 255.
     I1 = F.pad(I1, padding)
-    #p = (F.interpolate(I0, (16, 16), mode='bilinear', align_corners=False)
-    #     - F.interpolate(I1, (16, 16), mode='bilinear', align_corners=False)).abs().mean()
-    #if p < 5e-3 and args.skip:
-    #    if skip_frame % 100 == 0:
-    #        print("Warning: Your video has {} static frames, skipping them may change the duration of the generated video.".format(skip_frame))
-    #    skip_frame += 1
-    #    #pbar.update(1)
-    #    continue
-    #if p > 0.2:             
-    #    mid1 = lastframe
-    #    mid0 = lastframe
-    #    mid2 = lastframe
-    #else:
+
     output = make_inference(I0, I1, args.exp)
-    if args.montage:
-        write_buffer.put(np.concatenate((lastframe, lastframe), 1))
-        for mid in output:
-            mid = (((mid[0] * 255.).byte().cpu().numpy().transpose(1, 2, 0)))
-            write_buffer.put(np.concatenate((lastframe, mid[:h, :w]), 1))
-    else:
-        write_buffer.put(lastframe)
-        for mid in output:
-            mid = (((mid[0] * 255.).byte().cpu().numpy().transpose(1, 2, 0)))
-            write_buffer.put(mid[:h, :w])
-    #pbar.update(1)
-    lastframe = frame
-if args.montage:
-    write_buffer.put(np.concatenate((lastframe, lastframe), 1))
-else:
     write_buffer.put(lastframe)
+    for mid in output:
+        mid = (((mid[0] * 255.).byte().cpu().numpy().transpose(1, 2, 0)))
+        write_buffer.put(mid[:h, :w])
+
+    lastframe = frame
+write_buffer.put(lastframe)
 import time
 while(not write_buffer.empty()):
     time.sleep(0.1)
-#pbar.close()
 if not vid_out is None:
     vid_out.release()
-
-# move audio to new video file if appropriate
-if args.png == False and fpsNotAssigned == True and not args.skip:
-    outputVideoFileName = '{}_{}X_{}fps.{}'.format(video_path_wo_ext, args.exp, int(np.round(args.fps)), args.ext)
-    transferAudio(video_path_wo_ext + "." + args.ext, outputVideoFileName)
+    
+    
+