From 2465f214714c3d7101626e31874a8c37b9239a41 Mon Sep 17 00:00:00 2001 From: JunHo Kim Date: Mon, 26 Jul 2021 16:44:27 +0900 Subject: [PATCH] add ivtc process. refine log, comment --- README.md | 33 +++-- pillar_detection.py | 293 +++++++++++++++++++++++++------------------- 2 files changed, 190 insertions(+), 136 deletions(-) diff --git a/README.md b/README.md index bd99974..c2f4421 100644 --- a/README.md +++ b/README.md @@ -3,10 +3,27 @@ Convenient Vid cropping tool ## Info - Window 용 Script -- 재생은 x3 배속 +- 재생은 x3 배속으로 고정 + +## ivtc Code +< 하나의 파일을 변환하는 코드 > + +ffmpeg -i in.mp4 -vf fieldmatch=mode=pc_n:combmatch=full,bwdif=0:-1:1,decimate -c:v libx264 -crf 0 -preset ultrafast out.mp4 + +< 폴더 내에 있는 모든 파일을 변환하는 코드 > + +for %a in ("*.*") do ffmpeg -i "%a" -vf fieldmatch=mode=pc_n:combmatch=full,bwdif=0:-1:1,decimate -c:v libx264 -crf 0 -preset ultrafast ".\out\%~na_ivtc.mp4" ## ToDo +1. 우선 IVTC 파일이 1:1 매칭 되는 지 확인 +2. 안되면 안된 파일들 목록 만들기 +3. 만든 목록 파일들 IVTC 처리 +4. 처리하는 동안 진행상황 Output - **Concealing bitstream error** 해결 +- ~~타이틀 창 등에 현재 재생 퍼센트 / 프레임 수 Display~~ +- 타이틀 한글 깨지는 것 +- 한 프레임씩 이동 하는 기능 +- 처음에 목록 보여주고 시작? (필요한지?) - Log Ouput 필요 ### 포함 해야 할 내용 - 날짜, 시간 정보 @@ -24,10 +41,10 @@ Convenient Vid cropping tool - 영상 파일 명.txt - 같은 파일에 매 편집 시 누적되는 식 - 각 편집 로그의 구별은 편집 시간정보별로. - ### 구조적 문제 - - Subprocess 기반으로 FFMPEG 동작 - - 현재 Encoding 중인 영상이 어느정도 진행 되었는 지 알 수 없음 - - 전체 프로그램이 종료되기 전까지는 파일이 생성되지 않음 (subprocess가 계속 돌고있는 듯 함) - - 엔터를 누를 때 마다 다음 영상으로 넘어가기 때문에, 이전 영상의 진행도를 알기 어려울 듯 하다. - - _생성되는 각 프로세스를 다른 변수명에 Mapping 하여 Log 관리를 각각 해야 하나?_ - - Encoding이 종료되면 프로세스를 종료시키는 것 같은 방법이 필요할듯.. \ No newline at end of file +### 구조적 문제 +- Subprocess 기반으로 FFMPEG 동작 + - 현재 Encoding 중인 영상이 어느정도 진행 되었는 지 알 수 없음 + - 전체 프로그램이 종료되기 전까지는 파일이 생성되지 않음 (subprocess가 계속 돌고있는 듯 함) + - 엔터를 누를 때 마다 다음 영상으로 넘어가기 때문에, 이전 영상의 진행도를 알기 어려울 듯 하다. + - _생성되는 각 프로세스를 다른 변수명에 Mapping 하여 Log 관리를 각각 해야 하나?_ + - Encoding이 종료되면 프로세스를 종료시키는 것 같은 방법이 필요할듯.. \ No newline at end of file diff --git a/pillar_detection.py b/pillar_detection.py index 10e07f0..2bdf180 100644 --- a/pillar_detection.py +++ b/pillar_detection.py @@ -1,6 +1,3 @@ -import io -from contextlib import redirect_stderr - import os from collections import Counter import cv2 @@ -9,14 +6,16 @@ import re import subprocess import argparse import copy -import base64 +import time parser = argparse.ArgumentParser() -parser.add_argument("-i", "--input_dir", default="./", help="Input vid directory") +parser.add_argument("-s", "--source_dir", default="./", help="Input source vid directory") +parser.add_argument("-i", "--ivtc_dir", default="./ivtc", help="Input ivtc vid directory") parser.add_argument("-o", "--output_dir", default="./out", help="Output vid directory") parser.add_argument("-d", "--debug", default=False, action="store_true", help="for debugging. short output") args = parser.parse_args() +# Adjust crop info in display layout. def resizeCropInfo(key, cropPos, frame_width, frame_height): def _IsOutOfBound(_cropPos): w, h, x, y = _cropPos @@ -69,27 +68,62 @@ def resizeCropInfo(key, cropPos, frame_width, frame_height): def main(): vid_types = ('*.avi', '*.mov', '*.mp4', '*.mxf') - vid_grapped = [] + source_dir = args.source_dir + ivtc_dir = args.ivtc_dir + output_dir = args.output_dir + + source_list = [] + ivtc_list = [] for types in vid_types : - vid_grapped += sorted(glob.glob(os.path.join(args.input_dir, types))) + source_list += sorted(glob.glob(os.path.join(source_dir, types))) - #-------------------------------------------------------- - f = io.StringIO() - #-------------------------------------------------------- + ivtc_list += sorted(glob.glob(os.path.join(ivtc_dir, "*.mp4"))) - for i, fpath in enumerate(vid_grapped): - # debugging - # fpath = "01. TRAX - PARADOX MV.avi" + if len(ivtc_list) > 0: + ivtc_input_list = [] + for i, fpath in enumerate(source_list): + vid_name = os.path.basename(fpath)[:-4] + ivtc_name_list = [os.path.basename(x)[:-4] for x in ivtc_list] - if i >= 68 : - return + if not vid_name + "_ivtc" in ivtc_name_list : + ivtc_input_list += glob.glob(fpath) + print(f"[INFO]\t Seems \"{os.path.basename(fpath)}\" is stil telecine... add to ivtc processing list") + else : + ivtc_input_list = source_list + for i, ivtc_input in enumerate(ivtc_input_list) : + ivtc_input_num = len(ivtc_input_list) + ivtc_input_name = os.path.basename(ivtc_input) + ivtc_output_path = os.path.join(ivtc_dir, ivtc_input_name[:-4]) + "_ivtc.mp4" + + if not os.path.exists(ivtc_dir) : + os.mkdir(ivtc_dir) + print(f'[INFO]\t {ivtc_dir} created') + + print(f"[INFO]\t Work ivtc on \"{ivtc_input_name}\" ===> \"{ivtc_output_path}\" ... [{i + 1}/{ivtc_input_num}]") + print("[INFO]\t Please wait...") + start_time = time.time() + subprocess.run(["ffmpeg", "-ss", "10", "-i", ivtc_input, "-vf", "fieldmatch=mode=pc_n:combmatch=full,bwdif=0:-1:1,decimate", "-c:v", "libx264", "-crf", "0", + "-preset", "ultrafast","-loglevel", "quiet", "-y", ivtc_output_path]) + t = time.time() - start_time + print("[INFO]\t Work done. [It takes %.2f seconds.]" % t) + + + ivtc_list = [] + ivtc_list += sorted(glob.glob(os.path.join(ivtc_dir, "*.mp4"))) + target_ratio = 16 / 9 + print("[INFO]\t Target ratio : %.7f" %target_ratio) + print("[INFO]\t Input ivtc dir :", ivtc_dir) + for i, fpath in enumerate(ivtc_list): vid_name = os.path.basename(fpath)[:-4] - vid_num = len(vid_grapped) - # vid_name = vid_name.encode('utf-8').decode('cp949') + vid_num = len(ivtc_list) - print(f"Video num : {i + 1} / {vid_num}") - print("File to detect crop: ", fpath) + # Korean title text bug fix (work in progress) + # vid_name = vid_name.encode('utf-8').decode('cp949') + + print(f"[INFO]\t Video num : {i + 1} / {vid_num}") + print("[INFO]\t File to detect crop area: ", fpath) + print("[INFO]\t Detecting crop area...") p = subprocess.Popen(["ffmpeg", "-ss", "10", "-i", fpath, "-vf", "cropdetect=limit=38", "-vframes", "3000", "-f", "null", "out.null"], stdout=subprocess.PIPE, stderr=subprocess.PIPE) infos = p.stderr.read().decode('utf-8') @@ -98,116 +132,119 @@ def main(): # print ("Crop Infos : ") # print (allCrops) mostCommonCrop = Counter(allCrops).most_common(1) - print("most common detected crop info: %s" % mostCommonCrop) - global crop - crop = mostCommonCrop[0][0] - # Delete "crop=" in list cropPos + if len(mostCommonCrop) > 0 : + crop = mostCommonCrop[0][0] + else : + print(f"[!!! ERROR !!!]\t Can`t read input vid. Skip to next vid \t ({fpath})") + continue + + # Delete "crop=" in list "crop" cropPos = crop.split(":") cropPos[0] = cropPos[0][5:] + print(f"[INFO]\t Most common detected crop area: [w : {cropPos[0]}, h : {cropPos[1]}, x : {cropPos[2]}, y : {cropPos[3]}]") cropPos = [int(x) for x in cropPos] - with redirect_stderr(f): - # Capture vid, Get Vid info - capture = cv2.VideoCapture(fpath) - frame_width = capture.get(cv2.CAP_PROP_FRAME_WIDTH) - frame_height = capture.get(cv2.CAP_PROP_FRAME_HEIGHT) - fps = capture.get(cv2.CAP_PROP_FPS) - frame_pos = 0 - frame_count = 0 - target_ratio = 16 / 9 - - # Crop Area even num check - if (cropPos[0] % 2) != 0 : - if (cropPos[0] + 1) >= frame_width: - cropPos[0] -= 1 - else: - cropPos[0] += 1 - if (cropPos[1] % 2) != 0 : - if (cropPos[1] + 1) >= frame_width: - cropPos[1] -= 1 - else: - cropPos[1] += 1 + # Capture vid, Get Vid info + capture = cv2.VideoCapture(fpath) + frame_width = capture.get(cv2.CAP_PROP_FRAME_WIDTH) + frame_height = capture.get(cv2.CAP_PROP_FRAME_HEIGHT) + fps = capture.get(cv2.CAP_PROP_FPS) + frame_pos = 0 + frame_count = 0 + + # Crop Area even num check + if (cropPos[0] % 2) != 0 : + if (cropPos[0] + 1) >= frame_width: + cropPos[0] -= 1 + else: + cropPos[0] += 1 + if (cropPos[1] % 2) != 0 : + if (cropPos[1] + 1) >= frame_width: + cropPos[1] -= 1 + else: + cropPos[1] += 1 - # Crop Area Multiple of 4 check - if (cropPos[0] % 4) != 0 : - if (cropPos[0] + 2) >= frame_width: - cropPos[0] -= 2 - else: - cropPos[0] += 2 - if (cropPos[1] % 4) != 0 : - if (cropPos[1] + 2) >= frame_width: - cropPos[1] -= 2 - else: - cropPos[1] += 2 - - print("Target Ratio : %.7f" %target_ratio) - print("Input dir :", args.input_dir) + # Crop Area Multiple of 4 check + if (cropPos[0] % 4) != 0 : + if (cropPos[0] + 2) >= frame_width: + cropPos[0] -= 2 + else: + cropPos[0] += 2 + if (cropPos[1] % 4) != 0 : + if (cropPos[1] + 2) >= frame_width: + cropPos[1] -= 2 + else: + cropPos[1] += 2 - key = None - isPaused = False - pausedFrame = None - frame = None - cv2.namedWindow(f'{vid_name}', cv2.WINDOW_NORMAL | cv2.WINDOW_KEEPRATIO) - while True: - key = cv2.waitKeyEx(10) + key = None + isPaused = False + pausedFrame = None + frame = None + cv2.namedWindow(f'{vid_name}', cv2.WINDOW_NORMAL | cv2.WINDOW_KEEPRATIO) + while True: + key = cv2.waitKeyEx(10) - # Handle close button event - if cv2.getWindowProperty(f'{vid_name}', cv2.WND_PROP_VISIBLE) < 1 : - return + # Close button(X button) event + if cv2.getWindowProperty(f'{vid_name}', cv2.WND_PROP_VISIBLE) < 1 : + return - if key > 0 : - # 13 == Enter key ASCII Code - if key == 13: - break - # Pause - elif key == ord('p') or key == ord('P'): - if isPaused : - isPaused = False - else : - isPaused = True - frame = copy.deepcopy(pausedFrame) - # Replay - elif key == ord('r') or key == ord('R'): - if not isPaused: - capture = cv2.VideoCapture(fpath) - continue - # Move Forward (10s) - elif key == ord('e') or key == ord('E'): - if not isPaused: - capture.set(cv2.CAP_PROP_POS_FRAMES, frame_pos + (fps * 10)) - # Move Backward (10s) - elif key == ord('q') or key == ord('Q'): - if not isPaused: - capture.set(cv2.CAP_PROP_POS_FRAMES, frame_pos - (fps * 10)) - # Skip this Vid - elif key == ord('n') or key == ord('N'): - break; - else: - # Crop Area control - cropPos = resizeCropInfo(key, cropPos, frame_width, frame_height) - - frame_pos = capture.get(cv2.CAP_PROP_POS_FRAMES) - frame_count = capture.get(cv2.CAP_PROP_FRAME_COUNT) - - if(frame_pos >= frame_count): + if key > 0 : + # 13 == Enter key ASCII Code + if key == 13: break - - print("Current Crop Info : ", end="") - print(cropPos, end="") - print(f"\t Frame Count ({frame_pos}/{frame_count})", end="\r") + # Pause + elif key == ord('p') or key == ord('P'): + if isPaused : + isPaused = False + else : + isPaused = True + frame = copy.deepcopy(pausedFrame) + # Replay + elif key == ord('r') or key == ord('R'): + if not isPaused: + capture = cv2.VideoCapture(fpath) + continue + # Move Forward (10s) + elif key == ord('e') or key == ord('E'): + if not isPaused: + capture.set(cv2.CAP_PROP_POS_FRAMES, frame_pos + (fps * 10)) + # Move Backward (10s) + elif key == ord('q') or key == ord('Q'): + if not isPaused: + capture.set(cv2.CAP_PROP_POS_FRAMES, frame_pos - (fps * 10)) + elif key == ord('h') or key == ord('H'): + print("[INFO]\t ### KEY GUIDE ########################################################################################################") + print("[INFO]\t # (MOVE : wasd), (RESIZE : ijkl), (FOR / BACKWARD : q, e), (RESET : r), (PAUSE : p), (SKIP : n), (CONFIRM : enter) #") + print("[INFO]\t ###################################################################################################################### ") + # Skip this Vid + elif key == ord('n') or key == ord('N'): + break; + else: + # Crop Area control + cropPos = resizeCropInfo(key, cropPos, frame_width, frame_height) - if not isPaused: - ret, frame = capture.read() - pausedFrame = copy.deepcopy(frame) - cv2.rectangle(frame, (cropPos[2], cropPos[3]), (cropPos[2] + cropPos[0], - cropPos[3] + cropPos[1]), (0, 0, 255), thickness=1, lineType=cv2.LINE_8) - cv2.imshow(f'{vid_name}', frame) - else : - pausedFrame = copy.deepcopy(frame) - cv2.rectangle(pausedFrame, (cropPos[2], cropPos[3]), (cropPos[2] + cropPos[0], - cropPos[3] + cropPos[1]), (0, 0, 255), thickness=1, lineType=cv2.LINE_8) - cv2.imshow(f'{vid_name}', pausedFrame) + frame_pos = capture.get(cv2.CAP_PROP_POS_FRAMES) + frame_count = capture.get(cv2.CAP_PROP_FRAME_COUNT) + + if(frame_pos >= frame_count): + break + + print("[INFO]\t Current crop area : ", end="") + print(cropPos, end="") + print(f"\t Frame Count ({frame_pos}/{frame_count})\t[PRESS \"h\" TO VIEW KEY GUIDE]", end="\r") + + if not isPaused: + ret, frame = capture.read() + pausedFrame = copy.deepcopy(frame) + cv2.rectangle(frame, (cropPos[2], cropPos[3]), (cropPos[2] + cropPos[0], + cropPos[3] + cropPos[1]), (0, 0, 255), thickness=1, lineType=cv2.LINE_8) + cv2.imshow(f'{vid_name}', frame) + else : + pausedFrame = copy.deepcopy(frame) + cv2.rectangle(pausedFrame, (cropPos[2], cropPos[3]), (cropPos[2] + cropPos[0], + cropPos[3] + cropPos[1]), (0, 0, 255), thickness=1, lineType=cv2.LINE_8) + cv2.imshow(f'{vid_name}', pausedFrame) print("") capture.release() @@ -222,13 +259,13 @@ def main(): # ffmpeg filter args crop_arg = f"crop={cropPos[0]}:{cropPos[1]}:{cropPos[2]}:{cropPos[3]}," - print(f"[CropInfo] (W : {cropPos[0]}), (H : {cropPos[1]}), (X : {cropPos[2]}), (Y : {cropPos[3]})") + print(f"[INFO]\t [CropInfo] (W : {cropPos[0]}), (H : {cropPos[1]}), (X : {cropPos[2]}), (Y : {cropPos[3]})") crop_ratio = cropPos[0] / cropPos[1] scale_arg = "" - # Set padding argument - # Offset for 16:9 display ratio + # Set padding(letter box) argument + # variable "offset" for 16:9 output display ratio if target_ratio < crop_ratio: offset = (9 * cropPos[0] - 16 * cropPos[1]) / 16 tmp_height = cropPos[1] + offset @@ -237,7 +274,7 @@ def main(): tmp_height -= (tmp_height % 4) # scale_arg = f"scale=w={cropPos[0]}:h={tmp_height}," pad_arg = f"pad={cropPos[0]}:{tmp_height}:(ow-iw)/2:(ih-oh)/2" - print(f"[PadInfo] (W : {cropPos[0]}), (H : {tmp_height})") + print(f"[INFO]\t [PadInfo] (W : {cropPos[0]}), (H : {tmp_height})") else: offset = (16 * cropPos[1] - 9 * cropPos[0]) / 9 tmp_width = cropPos[0] + offset @@ -246,15 +283,15 @@ def main(): tmp_width -= (tmp_width % 4) # scale_arg = f"scale=w={tmp_width}:h={cropPos[1]}," pad_arg = f"pad={tmp_width}:{cropPos[1]}:(ow-iw)/2:(ih-oh)/2" - print(f"[PadInfo] (W : {tmp_width}), (H : {cropPos[1]})") + print(f"[INFO]\t [PadInfo] (W : {tmp_width}), (H : {cropPos[1]})") - print("Encoding \"%s\" ..." % vid_name) + print("[INFO]\t Encoding \"%s\" ..." % vid_name) - if not os.path.exists(args.output_dir) : - os.mkdir(args.output_dir) - print(f'{args.output_dir} created') + if not os.path.exists(output_dir) : + os.mkdir(output_dir) + print(f'[INFO]\t {output_dir} created') - output_path = os.path.join(args.output_dir, f"{vid_name}_box.mp4") + output_path = os.path.join(output_dir, f"{vid_name}_box.mp4") if args.debug: # for Debug p = subprocess.Popen(["ffmpeg", "-t", "15", "-ss", "30", "-i", fpath, "-vf", crop_arg + scale_arg + pad_arg, "-y",