Add tracy's code

.vscode/launch.json

@@ -1,16 +0,0 @@
-{
-  // Use IntelliSense to learn about possible attributes.
-  // Hover to view descriptions of existing attributes.
-  // For more information, visit: https://go.microsoft.com/fwlink/?linkid=830387
-  "version": "0.2.0",
-  "configurations": [
-    {
-      "name": "Python: Current File",
-      "type": "python",
-      "request": "launch",
-      "program": "${file}",
-      "console": "integratedTerminal",
-      "justMyCode": true
-    }
-  ]
-}

OpenCV/README.md

@@ -1,8 +0,0 @@
-1. Converts RGB to 2D Black and White image
-2. Animate line drawing as a path
-
-
-INSTALLATIONS
-moviepy
-opencv
-

OpenCV/contourfinder.py

@@ -1,124 +0,0 @@
-import cv2 
-import sys
-from cv2 import CONTOURS_MATCH_I2
-import numpy as np
-
-#Takes two inputs as images
-#Original 1 and 2 are dummy data
-
-original =cv2.imread(r"Animate\images\flag (1).png")
-original2 =cv2.imread(r"Animate\images\brain2.png")
-
-# get key positions at which frame needs to be generated
-def list_of_positions(num_contours,num_frames=100):
-   positions=[]
-   for i in range(0, num_frames):
-      positions.append(int(num_contours/num_frames*i)) 
-   return positions 
-
-
-
-def contourfinder(image1, image2, text=None, num_frames=100):
-    #Create two blank pages to write into
-    blank = np.zeros(image1.shape, dtype='uint8')
-    blank2 = np.zeros(image1.shape, dtype='uint8')
-
-
-    #Threshold and contours for image 1 and 2
-    threshold=cv2.Canny(image=image1, threshold1=100, threshold2=200)
-    contours, hierarchies = cv2.findContours(threshold, cv2.RETR_LIST, cv2.CHAIN_APPROX_SIMPLE)
-
-    threshold2=cv2.Canny(image=original2, threshold1=100, threshold2=200)
-    contours2, hierarchies2 = cv2.findContours(threshold2, cv2.RETR_LIST, cv2.CHAIN_APPROX_SIMPLE)
-    
-
-    #Initialize three empty videos
-    vid1 = cv2.VideoWriter('vid1.mp4',cv2.VideoWriter_fourcc(*'mp4v'), 24, threshold.shape)
-    vid2 = cv2.VideoWriter('vid2.mp4',cv2.VideoWriter_fourcc(*'mp4v'), 24, threshold.shape) 
-    text_vid = cv2.VideoWriter('text_vid.mp4',cv2.VideoWriter_fourcc(*'mp4v'), 10, threshold.shape) 
-
-    
-    #Get positions
-    positions=list_of_positions((len(contours)))
-    frames=[]
-
-
-    #Loop over contours adding them to blank image then writing to video
-    for i in range(0, len(contours)):
-        cv2.drawContours(blank, contours=contours, contourIdx=i, color=(125, 200, 255), thickness=1) 
-
-        if i in positions:
-            frames.append(blank)
-            
-            #Complile to video
-            vid1.write(blank)
-    last_image1=blank
-    vid1.release()
-
-    positions=list_of_positions((len(contours2)))
-
-    for i in range(0, len(contours2)):
-        cv2.drawContours(blank2, contours=contours2, contourIdx=i, color=(125, 200, 255), thickness=1) 
-        if i in positions:
-            frames.append(blank2)
-            
-            vid2.write(blank2)
-    last_image2=blank2
-    vid2.release()
-
-    
-    #Next is the text vid
-
-    if text !=None:	
-        # Reading an image in default mode
-        image = np.zeros(original.shape, dtype='uint8')
-            
-        # font
-        font = cv2.FONT_HERSHEY_COMPLEX
-
-        # org
-        org = (10, 400)
-
-        # fontScale
-        fontScale = 3
-
-        # Blue color in BGR
-        color = (186, 184, 108)
-
-        # Line thickness of 2 px
-        thickness = 4
-
-
-
-
-        def text_frames(text,image,org):
-            spacing=55 #spacing between letters
-            blink=image
-            cv2.imwrite(f"blink.png", blink)
-            for i in range(0, len(text)-1):
-                
-                text_vid.write(blink)
-            
-                # Using cv2.putText() method
-                image = cv2.putText(image, text[i], org, font,
-                    fontScale, color, thickness, cv2.LINE_AA)
-
-                #Take care of org spacing
-                org=(org[0]+spacing,org[1])
-                if text[i].isupper():
-                    org=(org[0] +spacing +1, org[1])
-                    print(f"Upper {text[i]}")
-                print(org)
-
-                # Displaying the image
-                cv2.imwrite(f"text_im{i}.png", image)
-
-                #Complile to video
-                text_vid.write(image)
-            text_vid.release()
-
-        text_frames(text,image,org)
-    return last_image1,last_image2
-        
-
-#contourfinder(original, original2, "spies gui Fly")

OpenCV/requirements.txt

@@ -1,15 +0,0 @@
-tensorflow==2.6.2 # The latest should include tensorflow-gpu
-tensorflow-datasets==4.4.0
-tensorflow-addons==0.15.0
-absl-py==0.12.0
-gin-config==0.5.0
-parameterized==0.8.1
-mediapy==1.0.3
-scikit-image==0.19.1
-apache-beam==2.34.0
-google-cloud-bigquery-storage==1.1.0 # Suppresses a harmless error from beam
-natsort==8.1.0
-image-tools
-opencv-python==4.6.0.66
-moviepy==1.0.3
-ffmpeg

OpenCV/videoedit.py

@@ -1,17 +0,0 @@
-from gc import freeze
-from moviepy.editor import VideoFileClip as vfc 
-from moviepy.editor import concatenate_videoclips, vfx, videotools,CompositeVideoClip\
-    ,TextClip, ImageSequenceClip, transfx
-#from contoutfinder import original,original2
-
-
-
-clip1=vfc("project.mp4").fx(vfx.fadeout,.5)
-#clip1to2=vfc(original).fx(vfx.fadeout,.5).fx(vfx.freeze,'end',3)
-clip2=vfc("next_vid.mp4").fx(transfx.crossfadein,2).fx(vfx.fadeout, .5)
-clip3=vfc("text_vid.mp4", True).fx(vfx.fadein,.5).fx(vfx.blink, 5, 10)
-
-
-combined=concatenate_videoclips([clip1,clip2, clip3])
-
-combined.write_videofile("combined.mp4")

animate.py

@@ -0,0 +1,215 @@
+import os
+import sys
+
+import cv2
+import mediapy
+import numpy as np
+from eval import interpolator, util
+from huggingface_hub import snapshot_download
+from image_tools.sizes import resize_and_crop
+from moviepy.editor import CompositeVideoClip
+from moviepy.editor import VideoFileClip as vfc
+from PIL import Image
+
+# My installs
+os.system("git clone https://github.com/google-research/frame-interpolation")
+sys.path.append("frame-interpolation")
+
+
+# get key positions at which frame needs to be generated
+def list_of_positions(num_contours, num_frames=100):
+    positions = []
+    for i in range(0, num_frames):
+        positions.append(int(num_contours / num_frames * i))
+    return positions
+
+
+def contourfinder(image1, image2, text=None, num_frames=100):
+    # Create two blank pages to write into
+    # I just hardcoded 1024*1024 as the size, ideally this should be np.shape(image1)
+    blank = np.zeros((2048, 2048, 3), dtype="uint8")
+    blank2 = np.zeros((2048, 2048, 3), dtype="uint8")
+
+    # Threshold and contours for image 1 and 2
+    threshold = cv2.Canny(image=image1, threshold1=100, threshold2=200)
+    contours, hierarchies = cv2.findContours(
+        threshold, cv2.RETR_LIST, cv2.CHAIN_APPROX_SIMPLE
+    )
+
+    threshold2 = cv2.Canny(image=image2, threshold1=100, threshold2=200)
+    contours2, hierarchies2 = cv2.findContours(
+        threshold2, cv2.RETR_LIST, cv2.CHAIN_APPROX_SIMPLE
+    )
+
+    # Initialize three empty videos
+    vid1 = cv2.VideoWriter(
+        "vid1.mp4", cv2.VideoWriter_fourcc(*"mp4v"), 24, threshold.shape
+    )
+    vid2 = cv2.VideoWriter(
+        "vid2.mp4", cv2.VideoWriter_fourcc(*"mp4v"), 24, threshold.shape
+    )
+    text_vid = cv2.VideoWriter(
+        "text_vid.mp4", cv2.VideoWriter_fourcc(*"mp4v"), 10, threshold.shape
+    )
+
+    # Get positions
+    positions = list_of_positions((len(contours)))
+    frames = []
+
+    # Loop over contours adding them to blank image then writing to video
+    for i in range(0, len(contours)):
+        cv2.drawContours(
+            blank, contours=contours, contourIdx=i, color=(125, 200, 255), thickness=1
+        )
+
+        if i in positions:
+            frames.append(blank)
+
+            # Complile to video
+            vid1.write(blank)
+
+    vid1.release()
+    clip1 = vfc("/home/user/app/vid1.mp4")
+    # print(f"Type clip 1{type(clip1)}")
+
+    positions = list_of_positions((len(contours2)))
+
+    for i in range(0, len(contours2)):
+        cv2.drawContours(
+            blank2, contours=contours2, contourIdx=i, color=(125, 200, 255), thickness=1
+        )
+        if i in positions:
+            frames.append(blank2)
+
+            vid2.write(blank2)
+
+    vid2.release()
+    clip3 = vfc("/home/user/app/vid2.mp4")
+
+    # Next is the text vid
+
+    if text != None:
+        # Reading an image in default mode
+        image = np.zeros(original.shape, dtype="uint8")
+
+        # font
+        font = cv2.FONT_HERSHEY_COMPLEX
+
+        # org
+        org = (10, 400)
+
+        # fontScale
+        fontScale = 3
+
+        # Blue color in BGR
+        color = (186, 184, 108)
+
+        # Line thickness of 2 px
+        thickness = 4
+
+        def text_frames(text, image, org):
+            spacing = 55  # spacing between letters
+            blink = image
+            cv2.imwrite(f"blink.png", blink)
+            for i in range(0, len(text) - 1):
+
+                text_vid.write(blink)
+
+                # Using cv2.putText() method
+                image = cv2.putText(
+                    image, text[i], org, font, fontScale, color, thickness, cv2.LINE_AA
+                )
+
+                # Take care of org spacing
+                org = (org[0] + spacing, org[1])
+                if text[i].isupper():
+                    org = (org[0] + spacing + 1, org[1])
+                    print(f"Upper {text[i]}")
+                print(org)
+
+                # Displaying the image
+                cv2.imwrite(f"text_im{i}.png", image)
+
+                # Complile to video
+                text_vid.write(image)
+            text_vid.release()
+
+        text_frames(text, image, org)
+    return clip1, clip3
+
+
+def load_model(model_name):
+    model = interpolator.Interpolator(snapshot_download(repo_id=model_name), None)
+    return model
+
+
+model_names = [
+    "akhaliq/frame-interpolation-film-style",
+    "NimaBoscarino/frame-interpolation_film_l1",
+    "NimaBoscarino/frame_interpolation_film_vgg",
+]
+
+models = {model_name: load_model(model_name) for model_name in model_names}
+
+ffmpeg_path = util.get_ffmpeg_path()
+mediapy.set_ffmpeg(ffmpeg_path)
+
+
+def resize(width, img):
+    basewidth = width
+    img = Image.open(img)
+    wpercent = basewidth / float(img.size[0])
+    hsize = int((float(img.size[1]) * float(wpercent)))
+    img = img.resize((basewidth, hsize), Image.ANTIALIAS)
+    return img
+
+
+def resize_img(img1, img2):
+    img_target_size = Image.open(img1)
+    img_to_resize = resize_and_crop(
+        img2,
+        (
+            img_target_size.size[0],
+            img_target_size.size[1],
+        ),  # set width and height to match img1
+        crop_origin="middle",
+    )
+    img_to_resize.save("resized_img2.png")
+
+
+def predict(frame1, frame2, times_to_interpolate, model_name):
+    model = models[model_name]
+
+    img1 = cv2.imread(frame1)
+    img2 = cv2.imread(frame2)
+
+    frame1 = resize(256, frame1)
+    frame2 = resize(256, frame2)
+
+    frame1.save("test1.png")
+    frame2.save("test2.png")
+
+    resize_img("test1.png", "test2.png")
+    input_frames = ["test1.png", "resized_img2.png"]
+
+    frames = list(
+        util.interpolate_recursively_from_files(
+            input_frames, times_to_interpolate, model
+        )
+    )
+
+    mediapy.write_video("out.mp4", frames, fps=5)
+
+    print(f"TYPES....{type(img1)},{type(img2)} SHAPES{img1.shape} Img {img1}")
+    clip1, clip3 = contourfinder(img1, img2)  # has a third text option
+
+    # Use open CV and moviepy code
+    # So we move from open CV video 1 to out.mp4 to open CV video2
+    clip1 = clip1
+    clip2 = vfc("out.mp4").resize(8).set_start(clip1.duration - 0.5).crossfadein(2)
+    clip3 = clip3.set_start((clip1.duration - 0.5) + (clip2.duration)).crossfadein(2)
+
+    new_clip = CompositeVideoClip([clip1, clip2, clip3])
+    new_clip.write_videofile("out.mp4")
+
+    return "out.mp4"

app.py

@@ -104,6 +104,10 @@ def add_static_image_to_audio(image, audio_input) -> str:
     video_clip.write_videofile(path, audio_codec="aac")
     return path
 
+def add_openv_animation_to_audio(image_paths: list[str], audio_path: str) -> str:
+    
+
+
 
 def get_stable_diffusion_image(prompt) -> str:
     stable_diffusion = gr.Blocks.load(name="spaces/stabilityai/stable-diffusion")
@@ -124,4 +128,4 @@ iface = gr.Interface(
 )
 
 
-iface.launch()
+iface.launch()

requirements.txt

@@ -5,4 +5,18 @@ ftfy
 --extra-index-url https://download.pytorch.org/whl/cu113 torch
 moviepy
 gradio
-spodcast
+spodcast
+tensorflow==2.6.2 # The latest should include tensorflow-gpu
+tensorflow-datasets==4.4.0
+tensorflow-addons==0.15.0
+absl-py==0.12.0
+gin-config==0.5.0
+parameterized==0.8.1
+mediapy==1.0.3
+scikit-image==0.19.1
+apache-beam==2.34.0
+google-cloud-bigquery-storage==1.1.0 # Suppresses a harmless error from beam
+natsort==8.1.0
+image-tools
+opencv-python==4.6.0.66
+moviepy==1.0.3