#!/usr/bin/env python3 __package__ = "vid_transition" import math import pathlib import enum import logging import datetime import shutil import subprocess import argparse import tempfile from PIL import Image, ImageOps, ImageEnhance, ImageFilter # default variables used in arg-parser INPUT_VIDEOS = [] OUTPUT = "" NUM_FRAMES = 10 ANIMATION = "rotation" MAX_ROTATION = 45 MAX_DISTORTION = 0.7 MAX_BLUR = 0.2 MAX_BRIGHTNESS = 1.0 MAX_ZOOM = 2.0 DEBUG = False ART = True REMOVE_ORIGINAL = False MERGE_PHASES = False # variable that cannot be changed by arg-parser _OUTPUT_VIDEO_TYPE = ".mp4" _OUTPUT_VIDEO_CODEC = "h264" _LIMITS = {"rotation": (5, 90), "brightness": (0.0, 2.0), "blur": (0.005, 1.0), "distortion": (0.3, 1.0), "zoom": (1.2, 2.0)} _ANIMATION_HELP = f""" This program supports multiple types of animation. The arguments 'max_blur', 'max_distortion' and 'max_brightness' \ affect all these types. Whereas, 'max_rotation' and 'max_zoom' only affect [rotation] and [zoom] animations \ respectively. Here is the list of supported animations and the number of frames needed for each (assuming that the \ argument 'num_frames' is kept at {NUM_FRAMES}): * [rotation]: clockwise rotation, has 2 phases and needs {NUM_FRAMES*2} frames. * [rotation_inv]: anti-clockwise rotation, has 2 phases and needs {NUM_FRAMES*2} frames. * [zoom_in]: zooms inwards, has 2 phases and needs {NUM_FRAMES*2} frames. * [zoom_out]: zooms outwards, has 2 phases and needs {NUM_FRAMES*2} frames. * [translation]: translation from left to right, has 2 phases and needs {NUM_FRAMES*2} frames. * [translation_inv]: translation from right to left, has 2 phases and needs {NUM_FRAMES*2} frames. * [long_translation]: translation from left to right, has 3 phases and needs {NUM_FRAMES*3} frames ({NUM_FRAMES} from \ the first video, and {NUM_FRAMES*2} from the second one). * [long_translation_inv]: translation from right to left, has 3 phases and needs {NUM_FRAMES*3} frames. """ def log_debug(msg): logging.getLogger(__package__).debug(msg) def log_info(msg): logging.getLogger(__package__).info(msg) def log_warning(msg): logging.getLogger(__package__).info("WARNING: " + msg) def log_error(msg): logging.getLogger(__package__).error("ERROR: " + msg + " (use --debug from more info)") def intro_print(in_art): """ Taken from https://patorjk.com/software/taag using 4MAX font""" intro = ''' Yb dP 88 8888b. 888888 88""Yb db 88b 88 .dP"Y8 88 888888 Yb dP 88 8I Yb 88 88__dP dPYb 88Yb88 `Ybo." 88 88 YbdP 88 8I dY 88 88"Yb dP__Yb 88 Y88 o.`Y8b 88 88 YP 88 8888Y" 88 88 Yb dP""""Yb 88 Y8 8bodP' 88 88 ''' if in_art: log_info(intro) log_debug("=" * 80) log_info((" starting video transition creation ".center(80, "="))) log_debug("=" * 80) log_info("") def end_print(in_art): end = """ ,d8PPPP 888 ,d8 88PPP. ______ d88ooo 888_dPY8 88 8 ______ XXXXXX ,88' 8888' 88 88 8 XXXXXX 88bdPPP Y8P Y8 88oop' """ log_info((" video transition finished ".center(80, "="))) if in_art: log_info(end) def progress(count, total, status=''): bar_len = 40 end_char = '' filled_len = int(round(bar_len * count / float(total))) percents = round(100.0 * (count + 1) / float(total), 1) bar = '=' * (filled_len - 1) + '>' + ' ' * (bar_len - filled_len) if count == total - 1: bar = '=' * bar_len end_char = '\n' msg = '\r[%s] %s%s [%s/%s] ... %s ' % (bar, percents, '%', count + 1, total, status) print(msg, end=end_char, flush=True) def format_list(in_list, format_str=""): format_str = "{:" + format_str + "}" return "[" + (", ".join([format_str.format(ii) for ii in in_list])) + "]" class Animations(enum.IntEnum): rotation = 0 rotation_inv = 1 zoom_in = 2 zoom_out = 3 translation = 4 translation_inv = 5 long_translation = 6 long_translation_inv = 7 class FramesActions: class Type(enum.IntEnum): mirror = 0 zoom = 1 crop = 2 rotation = 3 blur = 4 distortion = 5 brightness = 6 class Function(enum.IntEnum): linear = 0 polynomial = 1 polynomial_inv = 2 class MirrorDirection(enum.IntEnum): all_directions_1 = 0 left_1 = 1 right_1 = 2 left_3 = 3 right_3 = 4 def __init__(self, action_type=Type.mirror): self.action_type = action_type self.values = [] self.function = FramesActions.Function.linear class AnimationActions: def __init__(self, max_zoom, max_brightness, max_rotation, max_blur, max_distortion, half_animation_num_frames): self.phase1_actions = [] self.phase2_actions = [] self.max_zoom = max_zoom self.max_brightness = max_brightness self.max_rotation = max_rotation self.max_blur = max_blur self.max_distortion = max_distortion self.half_animation_num_frames = half_animation_num_frames def get_actions_values(self, animation_type): if animation_type == Animations.rotation: self._get_rotation_actions(clockwise=True) elif animation_type == Animations.rotation_inv: self._get_rotation_actions(clockwise=False) elif animation_type == Animations.translation: self._get_translation_actions(left2right=True) elif animation_type == Animations.translation_inv: self._get_translation_actions(left2right=False) elif animation_type == Animations.zoom_in: self._get_zoom_actions(inward_direction=True) elif animation_type == Animations.zoom_out: self._get_zoom_actions(inward_direction=False) elif animation_type == Animations.long_translation: self._get_long_translation_actions(left2right=True) elif animation_type == Animations.long_translation_inv: self._get_long_translation_actions(left2right=False) self._print_info(animation_type) return self.phase1_actions, self.phase2_actions def _get_long_translation_actions(self, left2right=True): num_frames1 = self.half_animation_num_frames num_frames1_30p = int(round(num_frames1 * 0.3, 0)) num_frames1_50p = int(round(num_frames1 * 0.5, 0)) num_frames2 = 2 * self.half_animation_num_frames num_frames2_30p = int(round(num_frames2 * 0.3, 0)) # --- mirror frames --- direction1, direction2 = FramesActions.MirrorDirection.right_1, FramesActions.MirrorDirection.left_3 if not left2right: direction1, direction2 = FramesActions.MirrorDirection.left_1, FramesActions.MirrorDirection.right_3 fa_mirror1 = FramesActions(FramesActions.Type.mirror) [fa_mirror1.values.append(direction1) for _ in range(num_frames1)] self.phase1_actions.append(fa_mirror1) fa_mirror2 = FramesActions(FramesActions.Type.mirror) [fa_mirror2.values.append(direction2) for _ in range(num_frames2)] self.phase2_actions.append(fa_mirror2) # --- crop --- crop_1_values, crop_2_values = (0, 1), (0, 2, 3) if not left2right: crop_1_values, crop_2_values = (1, 0), (3, 2, 0) fa_crop1 = FramesActions(FramesActions.Type.crop) self._polynomial(fa_crop1, crop_1_values[0], crop_1_values[1], num_frames1) fa_crop1.values = [(v, 0) for v in fa_crop1.values] self.phase1_actions.append(fa_crop1) fa_crop2 = FramesActions(FramesActions.Type.crop) # self._linear(fa_crop2, crop_2_values[0], crop_2_values[1], num_frames1) self._polynomial_inv(fa_crop2, crop_2_values[0], crop_2_values[2], 2 * num_frames1) fa_crop2.values = [(v, 0) for v in fa_crop2.values] self.phase2_actions.append(fa_crop2) # --- brightness --- if _LIMITS["brightness"][0] <= self.max_brightness <= _LIMITS["brightness"][1] and self.max_brightness != 1: fa_br1, fa_br2 = FramesActions(FramesActions.Type.brightness), FramesActions(FramesActions.Type.brightness) self._polynomial(fa_br1, 1.0, self.max_brightness, num_frames1) fa_br2.values = [self.max_brightness for _ in range(num_frames1)] self._polynomial_inv(fa_br2, self.max_brightness, 1.0, num_frames1) self.phase1_actions.append(fa_br1) self.phase2_actions.append(fa_br2) # --- blur --- if _LIMITS["blur"][0] < self.max_blur <= _LIMITS["blur"][1]: fa_bl1, fa_bl2 = FramesActions(FramesActions.Type.blur), FramesActions(FramesActions.Type.blur) fa_bl1.values = [0 for _ in range(num_frames1_30p)] self._polynomial(fa_bl1, 0.0, self.max_blur, num_frames1 - num_frames1_30p) self._polynomial(fa_bl2, self.max_blur, 0.0, num_frames2) self.phase1_actions.append(fa_bl1) self.phase2_actions.append(fa_bl2) # --- distortion --- if _LIMITS["distortion"][0] < self.max_distortion <= _LIMITS["distortion"][1]: fa_ds1, fa_ds2 = FramesActions(FramesActions.Type.distortion), FramesActions(FramesActions.Type.distortion) self._polynomial(fa_ds1, 0.0, self.max_distortion, num_frames1 - num_frames1_50p) fa_ds1.values += [self.max_distortion for _ in range(num_frames1_50p)] fa_ds2.values = [self.max_distortion for _ in range(num_frames1 + num_frames1_50p)] self._polynomial(fa_ds2, self.max_distortion, 0.0, num_frames1 - num_frames1_50p) self.phase1_actions.append(fa_ds1) self.phase2_actions.append(fa_ds2) def _get_zoom_actions(self, inward_direction=True): num_frames = self.half_animation_num_frames num_frames_30p = int(round(num_frames * 0.3, 0)) num_frames_50p = int(round(num_frames * 0.5, 0)) # --- mirror frames --- for phase_fas in [self.phase1_actions, self.phase2_actions]: fa_mirror = FramesActions(FramesActions.Type.mirror) for _ in range(num_frames): fa_mirror.values.append(FramesActions.MirrorDirection.all_directions_1) phase_fas.append(fa_mirror) # --- zoom --- zoom_1v, zoom_2v = self.max_zoom, 1 / self.max_zoom if not inward_direction: zoom_1v, zoom_2v = zoom_2v, zoom_1v if _LIMITS["zoom"][0] <= self.max_zoom <= _LIMITS["zoom"][1]: fa_zoom1, fa_zoom2 = FramesActions(FramesActions.Type.zoom), FramesActions(FramesActions.Type.zoom) self._polynomial(fa_zoom1, 1.0, zoom_1v, num_frames) self.phase1_actions.append(fa_zoom1) self._polynomial_inv(fa_zoom2, zoom_2v, 1.0, num_frames) self.phase2_actions.append(fa_zoom2) # --- crop --- for phase_fas in [self.phase1_actions, self.phase2_actions]: fa_crop = FramesActions(FramesActions.Type.crop) for _ in range(num_frames): fa_crop.values.append((1, 1)) phase_fas.append(fa_crop) # --- brightness --- if _LIMITS["brightness"][0] <= self.max_brightness <= _LIMITS["brightness"][1] and self.max_brightness != 1: self._symmetric_action_value(self._linear, FramesActions.Type.brightness, 1, self.max_brightness, num_frames) # --- blur --- if _LIMITS["blur"][0] < self.max_blur <= _LIMITS["blur"][1]: self._symmetric_action_value(self._polynomial, FramesActions.Type.blur, 0, self.max_blur, num_frames, num_f_a_duplicates=num_frames_30p) # --- distortion --- if _LIMITS["distortion"][0] < self.max_distortion <= _LIMITS["distortion"][1]: self._symmetric_action_value(self._polynomial_inv, FramesActions.Type.distortion, 0, self.max_distortion, num_frames, num_f_b_duplicates=num_frames_50p) def _get_translation_actions(self, left2right=True): num_frames = self.half_animation_num_frames num_frames_30p = int(round(num_frames * 0.3, 0)) num_frames_50p = int(round(num_frames * 0.5, 0)) # --- mirror frames --- direction1, direction2 = FramesActions.MirrorDirection.right_1, FramesActions.MirrorDirection.left_1 if not left2right: direction1, direction2 = direction2, direction1 fa_mirror1 = FramesActions(FramesActions.Type.mirror) [fa_mirror1.values.append(direction1) for _ in range(num_frames)] self.phase1_actions.append(fa_mirror1) fa_mirror2 = FramesActions(FramesActions.Type.mirror) [fa_mirror2.values.append(direction2) for _ in range(num_frames)] self.phase2_actions.append(fa_mirror2) # --- crop --- crop_f_a, crop_f_b = 0, 1 if not left2right: crop_f_a, crop_f_b = 1, 0 fa_crop1 = FramesActions(FramesActions.Type.crop) self._polynomial(fa_crop1, crop_f_a, crop_f_b, num_frames) fa_crop1.values = [(v, 0) for v in fa_crop1.values] self.phase1_actions.append(fa_crop1) fa_crop2 = FramesActions(FramesActions.Type.crop) self._polynomial_inv(fa_crop2, crop_f_a, crop_f_b, num_frames) fa_crop2.values = [(v, 0) for v in fa_crop2.values] self.phase2_actions.append(fa_crop2) # --- brightness --- if _LIMITS["brightness"][0] <= self.max_brightness <= _LIMITS["brightness"][1] and self.max_brightness != 1: self._symmetric_action_value(self._linear, FramesActions.Type.brightness, 1, self.max_brightness, num_frames) # --- blur --- if _LIMITS["blur"][0] < self.max_blur <= _LIMITS["blur"][1]: self._symmetric_action_value(self._polynomial, FramesActions.Type.blur, 0, self.max_blur, num_frames, num_f_a_duplicates=num_frames_30p) # --- distortion --- if _LIMITS["distortion"][0] < self.max_distortion <= _LIMITS["distortion"][1]: self._symmetric_action_value(self._polynomial_inv, FramesActions.Type.distortion, 0, self.max_distortion, num_frames, num_f_b_duplicates=num_frames_50p) def _get_rotation_actions(self, clockwise=True): num_frames = self.half_animation_num_frames num_frames_30p = int(round(num_frames * 0.3, 0)) num_frames_50p = int(round(num_frames * 0.5, 0)) # --- mirror frames --- for phase_fas in [self.phase1_actions, self.phase2_actions]: fa_mirror = FramesActions(FramesActions.Type.mirror) for _ in range(num_frames): fa_mirror.values.append(FramesActions.MirrorDirection.all_directions_1) phase_fas.append(fa_mirror) # --- rotation --- if _LIMITS["rotation"][0] < self.max_rotation <= _LIMITS["rotation"][1]: mul = 1 if clockwise else -1 fa1_rot, fa2_rot = FramesActions(FramesActions.Type.rotation), FramesActions(FramesActions.Type.rotation) self._polynomial(fa1_rot, 0, - mul * self.max_rotation, num_frames) self._polynomial_inv(fa2_rot, mul * self.max_rotation, 0, num_frames) self.phase1_actions.append(fa1_rot) self.phase2_actions.append(fa2_rot) # --- crop --- for phase_fas in [self.phase1_actions, self.phase2_actions]: fa_crop = FramesActions(FramesActions.Type.crop) for _ in range(num_frames): fa_crop.values.append((1, 1)) phase_fas.append(fa_crop) # --- brightness --- if _LIMITS["brightness"][0] <= self.max_brightness <= _LIMITS["brightness"][1] and self.max_brightness != 1: self._symmetric_action_value(self._linear, FramesActions.Type.brightness, 1, self.max_brightness, num_frames) # --- blur --- if _LIMITS["blur"][0] < self.max_blur <= _LIMITS["blur"][1]: self._symmetric_action_value(self._polynomial, FramesActions.Type.blur, 0, self.max_blur, num_frames, num_f_a_duplicates=num_frames_30p) # --- distortion --- if _LIMITS["distortion"][0] < self.max_distortion <= _LIMITS["distortion"][1]: self._symmetric_action_value(self._polynomial_inv, FramesActions.Type.distortion, 0, self.max_distortion, num_frames, num_f_b_duplicates=num_frames_50p) def _symmetric_action_value(self, func, action_type, f_a, f_b, length, num_f_a_duplicates=0, num_f_b_duplicates=0, phase2_multiplier=1): p2m = phase2_multiplier phase1_fa = FramesActions(action_type) phase2_fa = FramesActions(action_type) [phase1_fa.values.append(f_a) for _ in range(num_f_a_duplicates)] [phase2_fa.values.append(f_b * p2m) for _ in range(num_f_b_duplicates)] func(phase1_fa, f_a, f_b, length - num_f_a_duplicates - num_f_b_duplicates) if phase1_fa.function == FramesActions.Function.linear: self._linear(phase2_fa, f_b * p2m, f_a * p2m, length - num_f_a_duplicates - num_f_b_duplicates) elif phase1_fa.function == FramesActions.Function.polynomial: self._polynomial_inv(phase2_fa, f_b * p2m, f_a * p2m, length - num_f_a_duplicates - num_f_b_duplicates) elif phase1_fa.function == FramesActions.Function.polynomial_inv: self._polynomial(phase2_fa, f_b * p2m, f_a * p2m, length - num_f_a_duplicates - num_f_b_duplicates) else: log_error("this should never happens") [phase1_fa.values.append(f_b) for _ in range(num_f_b_duplicates)] [phase2_fa.values.append(f_a * p2m) for _ in range(num_f_a_duplicates)] self.phase1_actions.append(phase1_fa) self.phase2_actions.append(phase2_fa) def _print_info(self, animation_type): log_info("") log_debug("".center(80, "=")) log_info(" Transition animation info ".center(80, "=")) log_debug("".center(80, "=")) animation_types = {0: "clockwise rotation", 1: "anticlockwise rotation", 2: "zoom in", 3: "zoom out", 4: "translation (left to right)", 5: "translation (right to left)", 6: "long translation (left to right)", 7: "inverse long translation(right to left)"} log_info(f"transition animation type: [{animation_types[animation_type]}]") log_info(f"transition activated effects (in order):") phase1_activated = [action.action_type.name for action in self.phase1_actions] phase2_activated = [action.action_type.name for action in self.phase2_actions] log_info(f"* phase 1 activated effects: [{', '.join(phase1_activated)}]") log_info(f"* phase 2 activated effects: [{', '.join(phase2_activated)}]") log_debug("") for phase_idx, actions in enumerate([self.phase1_actions, self.phase2_actions]): log_debug(f" transition phase_{phase_idx + 1} ".center(80, "-")) for action in actions: if action.action_type == FramesActions.Type.mirror: log_debug(f"mirroring frames, type: [{action.action_type.name}], " f"function: [{action.function.name}]") log_debug(f"* values: [{action.values[0].name}] - num frames: [{len(action.values)}]") elif action.action_type == FramesActions.Type.zoom: log_debug(f"zoom effect, max value: [{self.max_zoom:.1%}], function: [{action.function.name}]") log_debug(f"* values: {format_list(action.values, '.1%')}") elif action.action_type == FramesActions.Type.crop: log_debug(f"crop effect") cropped = [f"({v[0]:g}, {v[1]:g})" for v in action.values] log_debug(f"* values: {format_list(cropped, 's')}, function: [{action.function.name}]") elif action.action_type == FramesActions.Type.rotation: log_debug(f"rotation effect, max value (in degrees): [{self.max_rotation:.1f}], " f"function: [{action.function.name}]") log_debug(f"* values (degree): {format_list(action.values, '.1f')}") elif action.action_type == FramesActions.Type.blur: log_debug(f"blur effect, max value: [{self.max_blur:.1%}], function: [{action.function.name}]") log_debug(f"* values: {format_list(action.values, '.1%')}") elif action.action_type == FramesActions.Type.distortion: log_debug(f"len pincushion distortion effect, max value: [{self.max_distortion:.1%}], " f"function: [{action.function.name}]") log_debug(f"* values: {format_list(action.values, '.1%')}") elif action.action_type == FramesActions.Type.brightness: log_debug(f"brightness effect, max value: [{self.max_brightness:.1%}], " f"function: [{action.function.name}]") log_debug(f"* values: {format_list(action.values, '.1%')}") log_debug("") log_debug("") @staticmethod def _linear(frame_action, f_a, f_b, length): xa, xb = 0, length - 1 c1 = (f_b - f_a) / (xb - xa) c2 = f_a - c1 * xa for xi in range(length): frame_action.values.append(c1 * xi + c2) frame_action.function = FramesActions.Function.linear @staticmethod def _polynomial(frame_action, f_a, f_b, length, strength=3.0): xa, xb = 0, length - 1 c1 = (f_b - f_a) / ((xb - xa) ** strength) c2 = f_a for xi in range(length): frame_action.values.append(c1 * ((xi - xa) ** strength) + c2) frame_action.function = FramesActions.Function.polynomial @staticmethod def _polynomial_inv(frame_action, f_a, f_b, length, strength=3.0): AnimationActions._polynomial(frame_action, f_a, f_b, length, 1 / strength) frame_action.function = FramesActions.Function.polynomial_inv class AnimationImages: class PincushionDeformation: def __init__(self, strength=0.2, zoom=1.2, auto_zoom=False): self.correction_radius = None self.zoom = zoom self.strength = strength if strength <= 0: self.strength = 0.00001 self.auto_zoom = auto_zoom self.half_height = None self.half_width = None def transform(self, x, y): new_x = x - self.half_width new_y = y - self.half_height distance = math.sqrt(new_x ** 2 + new_y ** 2) r = distance / self.correction_radius if r == 0: theta = 1 else: theta = math.atan(r) / r source_x = self.half_width + theta * new_x * self.zoom source_y = self.half_height + theta * new_y * self.zoom return source_x, source_y def transform_rectangle(self, x0, y0, x1, y1): return (*self.transform(x0, y0), *self.transform(x0, y1), *self.transform(x1, y1), *self.transform(x1, y0)) def determine_parameters(self, img): width, height = img.size self.half_width = width / 2 self.half_height = height / 2 self.correction_radius = (min(self.half_width, self.half_height) * 10) * (1 - self.strength) ** 2 + 1 # print(f"correction radius => {self.correction_radius}") if self.auto_zoom: r = math.sqrt(min(self.half_height, self.half_width) ** 2) / self.correction_radius self.zoom = r / math.atan(r) def get_debug_info(self, img): self.determine_parameters(img) w, h = img.size msg = [" lens distortion debug info ".center(80, '=')] msg += [f"input image size: [w:{w}, h:{h}]"] if not self.auto_zoom: msg += [f"strength: [{self.strength:.0%}] , automatic zoom: [Off] , provided zoom: [{self.zoom:.0%}]"] else: msg += [f"strength: [{self.strength:.0%}] , automatic zoom: [On] , calculated zoom: [{self.zoom:.0%}]"] msg += ["corner points displacement:"] points = {"top-left": (0, 0), "top-center": (self.half_width, 0), "top-right": (w, 0), "left": (0, self.half_height), "right": (w, self.half_height), "bottom-left": (0, h), "bottom-center": (self.half_width, h), "bottom-right": (w, h)} for key, value in points.items(): res = self.transform(value[0], value[1]) msg += [f"* {key:<13s} [x:{res[0]:<6.1f}, y:{res[1]:<6.1f}] => [{value[0]:<4.0f}, {value[1]:<4.0f}]"] msg += [""] return msg def getmesh(self, img): self.determine_parameters(img) width, height = img.size grid_space = 20 target_grid = [] for x in range(0, width, grid_space): for y in range(0, height, grid_space): target_grid.append((x, y, x + grid_space, y + grid_space)) source_grid = [self.transform_rectangle(*rect) for rect in target_grid] return [t for t in zip(target_grid, source_grid)] @staticmethod def make_transition(working_dir, in_images1, in_images2, in_actions1, in_actions2, debug=False): log_info("") log_debug("".center(80, "=")) log_info(" Transition image processing ".center(80, "=")) log_debug("".center(80, "=")) images_path = [in_images1, in_images2] res_folder = [None, None] peak_distortion_msg = [] peak_distortion_value = 0.0 peak_distortion_img = None for phase_idx, actions in enumerate([in_actions1, in_actions2]): log_debug("=" * 80) log_info(f"processing transition phase_{phase_idx+1} images") for img_idx, img_path in enumerate(images_path[phase_idx]): if not debug: progress(img_idx, len(images_path[phase_idx]), f"phase_{phase_idx+1} images") img = Image.open(str(img_path)) original_size = img.size log_debug(f" image [{img_idx+1}/{len(images_path[phase_idx])}] processing ".center(80, "-")) log_debug(f"image path {img_path}") for action_idx, action in enumerate(actions): suffix = action.action_type.name if action_idx == len(actions) - 1: suffix += "_final" img_save_folder = working_dir / f"{action_idx+2}_phase{phase_idx+1}_{suffix}" img_save_folder.mkdir(exist_ok=True) if action_idx == len(actions) - 1: res_folder[phase_idx] = img_save_folder value = action.values[img_idx] msg = f"phase_{phase_idx+1} - img [{img_idx+1}/{len(images_path[phase_idx])}]" if isinstance(value, tuple): msg += f" - action [{action.action_type.name} => ({value[0]:.1%}, {value[1]:.1%})]" else: msg += f" - action [{action.action_type.name} => {value:g}]" msg += f" - folder [{img_save_folder.name}]" log_debug(msg) if action.action_type == FramesActions.Type.mirror: img = AnimationImages.mirror_image_effect(img, value) elif action.action_type == FramesActions.Type.zoom: img = AnimationImages.zoom_effect(img, value) elif action.action_type == FramesActions.Type.crop: img = AnimationImages.crop_effect(img, value, original_size) elif action.action_type == FramesActions.Type.rotation: img = AnimationImages.rotation_effect(img, value) elif action.action_type == FramesActions.Type.blur: img = AnimationImages.blur_effect(img, value) elif action.action_type == FramesActions.Type.distortion: img = AnimationImages.distortion_effect(img, value) if value > peak_distortion_value: peak_distortion_msg = AnimationImages.PincushionDeformation(value, 1.0).get_debug_info(img) peak_distortion_value = value peak_distortion_img = img_path elif action.action_type == FramesActions.Type.brightness: img = AnimationImages.brightness_effect(img, value) if debug or action_idx == len(actions) - 1: img.save(str(img_save_folder / img_path.name)) log_debug("") if peak_distortion_img is not None: log_debug(f"peak distortion effect: value [{peak_distortion_value}:.1%], img path: [{peak_distortion_img}]") for line in peak_distortion_msg: log_debug(line) return res_folder @staticmethod def mirror_image_effect(in_img, mirror_direction): images = [in_img, in_img.transpose(0), in_img.transpose(1), in_img.transpose(0).transpose(1)] w, h = in_img.width, in_img.height if mirror_direction == FramesActions.MirrorDirection.all_directions_1: new_size = (3 * w, 3 * h) pos_dict = {(0, 0): 3, (1, 0): 2, (2, 0): 3, (0, 1): 1, (1, 1): 0, (2, 1): 1, (0, 2): 3, (1, 2): 2, (2, 2): 3} elif mirror_direction == FramesActions.MirrorDirection.left_1: new_size = (2 * w, 1 * h) pos_dict = {(0, 0): 1, (1, 0): 0} elif mirror_direction == FramesActions.MirrorDirection.right_1: new_size = (2 * w, 1 * h) pos_dict = {(0, 0): 0, (1, 0): 1} elif mirror_direction == FramesActions.MirrorDirection.left_3: new_size = (4 * w, 1 * h) pos_dict = {(0, 0): 1, (1, 0): 0, (2, 0): 1, (3, 0): 0} elif mirror_direction == FramesActions.MirrorDirection.right_3: new_size = (4 * w, 1 * h) pos_dict = {(0, 0): 0, (1, 0): 1, (2, 0): 0, (3, 0): 1} else: return in_img res = Image.new('RGB', new_size) [res.paste(images[idx], (w * xy[0], h * xy[1])) for xy, idx in pos_dict.items()] return res @staticmethod def zoom_effect(in_img, zoom_value): w, h = in_img.size zoom2 = zoom_value * 2 new_img = in_img.crop((w/2 - w / zoom2, h/2 - h / zoom2, w/2 + w / zoom2, h/2 + h / zoom2)) try: return new_img.resize((w, h), Image.Resampling.BICUBIC) except AttributeError: return new_img.resize((w, h), Image.BICUBIC) @staticmethod def crop_effect(in_img, top_left_corner, original_img_size): h, w = original_img_size[0], original_img_size[1] tfc_x, tfc_y = int(round(top_left_corner[0] * h, 0)), int(round(top_left_corner[1] * w, 0)) return in_img.crop((tfc_x, tfc_y, tfc_x + h, tfc_y + w)) @staticmethod def rotation_effect(in_img, rot_angle): return in_img.rotate(rot_angle) @staticmethod def blur_effect(in_img, blur_value): blue_strength = min(in_img.size[0], in_img.size[1]) * blur_value * 0.1 # print(blue_strength) return in_img.filter(ImageFilter.GaussianBlur(blue_strength)) @staticmethod def distortion_effect(in_img, distortion_strength): return ImageOps.deform(in_img, AnimationImages.PincushionDeformation(distortion_strength, 1.0)) @staticmethod def brightness_effect(in_img, brightness_value): enhancer = ImageEnhance.Brightness(in_img) return enhancer.enhance(brightness_value) class DataHandler: def __init__(self): self.start_time = datetime.datetime.now() self.tmp_path = None self.output = None self.input_vid1 = None self.input_vid2 = None self.phase1_vid = None self.phase2_vid = None self.merged_vid = None self.fps = 30 self.vid1_raw_images_folder = None self.vid2_raw_images_folder = None self.phase1_images = [] self.phase2_images = [] self.animation = None def verify_arguments(self, in_args, in_tmp_path): self.tmp_path = in_tmp_path self.output = pathlib.Path(in_args.output) if in_args.output == "": self._suggest_output(in_args.output) if in_args.debug: self.tmp_path = self.output.parent / (self.output.stem + "_debug") if self.tmp_path.is_dir(): shutil.rmtree(str(self.tmp_path)) self.tmp_path.mkdir() self._setup_logging(in_args.debug, self.tmp_path / f"{__package__}.log") intro_print(in_args.art) if not self._verify_critical_info(in_args): return False self.phase1_vid = self.output.parent / (self.output.stem + "_phase1" + _OUTPUT_VIDEO_TYPE) self.phase2_vid = self.output.parent / (self.output.stem + "_phase2" + _OUTPUT_VIDEO_TYPE) self.merged_vid = self.output.parent / (self.output.stem + "_merged" + _OUTPUT_VIDEO_TYPE) log_info(f"first input video: {self.input_vid1}") log_info(f"second input video: {self.input_vid2}") if in_args.merge: log_debug(f"transition phase1 video: {self.phase1_vid}") log_debug(f"transition phase2 video: {self.phase2_vid}") log_info(f"output transition merged video: {self.merged_vid}") else: log_info(f"output transition phase1 video: {self.phase1_vid}") log_info(f"output transition phase2 video: {self.phase2_vid}") self._get_fps_from_video() log_info(f"frames per second (FPS): {self.fps}") self.vid1_raw_images_folder = self.tmp_path / "1_phase1_raw" self.vid2_raw_images_folder = self.tmp_path / "1_phase2_raw" self.vid1_raw_images_folder.mkdir() log_debug(f"created vid1_raw_images_folder: {self.vid1_raw_images_folder}") self.vid2_raw_images_folder.mkdir() log_debug(f"created vid2_raw_images_folder: {self.vid2_raw_images_folder}") if not self._extract_phase1_images(in_args.num_frames): return False num_frames_for_vid2 = in_args.num_frames if self.animation == Animations.long_translation or self.animation == Animations.long_translation_inv: num_frames_for_vid2 = 2 * in_args.num_frames if not self._extract_phase2_images(num_frames_for_vid2): return False log_info(f"number of frames for phase1: [{len(self.phase1_images)}], for phase2: [{len(self.phase2_images)}]") return True def final_images_to_video(self, res_folders): output_videos = [self.phase1_vid, self.phase2_vid] fps = str(self.fps) for idx in range(2): log_info(f"merging phase_{idx} images into a video ...") img_names = [f.stem for f in res_folders[idx].glob("*.png")] img_names.sort() start_idx = f"{max(int(img_names[0]) - 1, 0):04d}" cmd = ["ffmpeg", "-hide_banner", "-start_number", start_idx, "-framerate", fps, "-y", "-r", fps, "-i", str(res_folders[idx] / "%04d.png"), "-r", fps, "-vcodec", _OUTPUT_VIDEO_CODEC, str(output_videos[idx])] self._exec_command(cmd, f"command used for merging phase_{idx} images into a video ...") if not output_videos[idx].is_file(): log_error(f"ffmpeg failed to convert images to: {output_videos[idx]}") return False return True def _verify_critical_info(self, in_args): if shutil.which("ffmpeg") is None: log_error("'ffmpeg' is not installed, please install it before use") return False if len(in_args.input) < 2 or len(in_args.input) > 2: log_error(f"2 input videos needed, [{len(in_args.input)}] provided") return False self.input_vid1 = pathlib.Path(in_args.input[0]) if not self.input_vid1.is_file(): log_error(f"could not find first video under: {self.input_vid1}") return False self.input_vid2 = pathlib.Path(in_args.input[1]) if not self.input_vid2.is_file(): log_error(f"could not find second video under: {self.input_vid2}") return False if in_args.num_frames < 2 or in_args.num_frames > 100: log_error(f"number of frames per phase should be in the range [2, 100] (provided: [{in_args.num_frames}])") return False for animation_enum in Animations: # print(f"{in_args.animation.lower().strip()} <-> {animation_enum.name}") if in_args.animation.lower().strip() == animation_enum.name: self.animation = animation_enum break if self.animation is None: log_error(f"animation provided [{in_args.animation}] not recognized, please use one of the " f"following animations:") log_info(_ANIMATION_HELP) return False return True def _extract_phase1_images(self, in_num_frames): duration_ms = int(math.ceil(1000 * (in_num_frames + 2) / self.fps)) cmd = ["ffmpeg", "-hide_banner", "-sseof", f"-{duration_ms}ms", "-i", str(self.input_vid1), str(self.vid1_raw_images_folder / "%04d.png")] self._exec_command(cmd, "command used for extracting images from video num 1:") for img_f in self.vid1_raw_images_folder.glob("*.png"): self.phase1_images.append(img_f) self.phase1_images.sort() if len(self.phase1_images) < in_num_frames: log_error(f"could not extract [{in_num_frames}] images from the first video " f"({len(self.phase1_images)} extracted)") return False if len(self.phase1_images) > in_num_frames: self.phase1_images = self.phase1_images[-in_num_frames:] return True def _extract_phase2_images(self, in_num_frames): duration_ms = int(math.ceil(1000 * (in_num_frames + 2) / self.fps)) cmd = ["ffmpeg", "-hide_banner", "-to", f"{duration_ms}ms", "-i", str(self.input_vid2), str(self.vid2_raw_images_folder / "%04d.png")] self._exec_command(cmd, "command used for extracting images from video num 2:") for img_f in self.vid2_raw_images_folder.glob("*.png"): self.phase2_images.append(img_f) self.phase2_images.sort() if len(self.phase2_images) < in_num_frames: log_error(f"could not extract [{in_num_frames}] images from the second video " f"({len(self.phase2_images)} extracted)") return False if len(self.phase2_images) > in_num_frames: self.phase2_images = self.phase2_images[:in_num_frames] return True @staticmethod def _exec_command(in_cmd, in_presentation): log_debug("") log_debug(in_presentation) log_debug("") log_debug(" ".join(in_cmd)) res = subprocess.run(in_cmd, capture_output=True, text=True) log_debug("") log_debug("") log_debug("stdout:") log_debug("") log_debug(res.stdout) log_debug("") log_debug("stderr:") log_debug("") log_debug(res.stderr) log_debug("") log_debug("") return res.stdout, res.stderr def _get_fps_from_video(self): cmd = ["ffmpeg", "-hide_banner", "-i", str(self.input_vid1)] stdout, stderr = self._exec_command(cmd, "command used for extracting FPS") res = (stdout.lower() + " " + stderr.lower()).split(" ") log_debug("searching for FPS value in ffmpeg video information") for fps_word in ["fps", "fps,"]: for idx, word in enumerate(res): if word == fps_word: try: self.fps = int(res[idx - 1]) log_debug(f"FPS extracted from video [{self.fps}]") return except ValueError: log_debug(f"failed extract FPS, value [{res[idx - 1]}], error :[{ValueError}]") log_warning(f"cloud not retrieve FPS value from video (using ffmpeg): {self.input_vid1}") log_warning("falling back to FPS value of [30]") self.fps = 30 def get_duration_msg(self): end_time = datetime.datetime.now() t_delta = end_time - self.start_time seconds = int(t_delta.total_seconds()) days, seconds = divmod(seconds, 86400) hours, seconds = divmod(seconds, 3600) minutes, seconds = divmod(seconds, 60) if days > 0: return '%d d %d h %d m %d s' % (days, hours, minutes, seconds) elif hours > 0: return '%d h %d m %d s' % (hours, minutes, seconds) elif minutes > 0: return '%d m %d s' % (minutes, seconds) else: return '%d s' % (seconds,) def _suggest_output(self, in_output): # self.output = pathlib.Path().cwd() / "vt_debug" # return if in_output != "": out_path = pathlib.Path(in_output) self.output = out_path.parent / out_path.stem return cur_dir = pathlib.Path().cwd() video_files = [f.name for f in cur_dir.glob("*" + _OUTPUT_VIDEO_TYPE)] num = 1 previous_num = 0 while previous_num != num: previous_num = num for vn in video_files: if vn.startswith(f"vt{num}_"): num += 1 self.output = cur_dir / f"vt{num}" def merge_video_chunks(self): # TODO implement log_info("merging the two phases video chunks into one transition video") cmd = ["ffmpeg", "-hide_banner", "-i", str(self.phase1_vid), "-i", str(self.phase2_vid), "-filter_complex", "[0:v] [1:v] concat=n=2:v=1 [v]", "-map", "[v]", str(self.merged_vid)] self._exec_command(cmd, f"command used for merging phase video chunks into the output video ...") if not self.merged_vid.is_file(): log_error(f"ffmpeg failed to merge video phases into: {self.merged_vid}") return False log_debug(f"remove output video phase1: {self.phase1_vid}") self.phase1_vid.unlink(missing_ok=True) log_debug(f"remove output video phase2: {self.phase2_vid}") self.phase2_vid.unlink(missing_ok=True) return True @staticmethod def _setup_logging(debug, log_file_path): init_logger = logging.getLogger(__package__) if debug: init_logger.setLevel(logging.DEBUG) else: init_logger.setLevel(logging.INFO) i_formatter = logging.Formatter('%(message)s') ch = logging.StreamHandler() ch.setFormatter(i_formatter) init_logger.addHandler(ch) if debug: handler = logging.FileHandler(str(log_file_path), encoding='utf8') n_formatter = logging.Formatter('[%(levelname)s] [%(asctime)s] - %(message)s', "%H:%M:%S") handler.setFormatter(n_formatter) handler.setLevel(logging.DEBUG) init_logger.addHandler(handler) def str2bool(v): if isinstance(v, bool): return v if v.lower() in ('yes', 'true', 't', 'y', '1'): return True elif v.lower() in ('no', 'false', 'f', 'n', '0'): return False else: raise argparse.ArgumentTypeError('Boolean value expected, possible values: yes, y, true, 1, no, n, false, 0.') if __name__ == "__main__": all_animation_names = ", ".join([animation_enum.name for animation_enum in Animations]) parser = argparse.ArgumentParser(formatter_class=argparse.ArgumentDefaultsHelpFormatter, description='make a transition animation between two videos, using the last part ' 'of the first video, and the first part of the second video') parser.add_argument('-i', '--input', help='input videos, must be two', type=str, nargs='+', metavar='\b', default=INPUT_VIDEOS) parser.add_argument('-n', '--num_frames', help='the number of frames used for each animation phase, ' 'most animations consists of two phases', type=int, default=NUM_FRAMES, metavar='\b') parser.add_argument('-a', '--animation', help=f'possible animations (use -a help to show more info): ' f'{all_animation_names} ', type=str, default=ANIMATION, metavar='\b') parser.add_argument('-o', '--output', help='the name of the output (determined automatically if left empty), ' 'FPS is copied from the first video.', type=str, default=OUTPUT, metavar='\b') parser.add_argument('-r', '--max_rotation', help=f'rotation (in degree) value at the midpoint of the animation, ' f'possible range {list(_LIMITS["rotation"])}', type=int, default=MAX_ROTATION, metavar='\b') parser.add_argument('-d', '--max_distortion', help=f'lens distortion value at the midpoint of the animation' f'(in percentage), possible range {list(_LIMITS["distortion"])}', type=float, default=MAX_DISTORTION, metavar='\b') parser.add_argument('-b', '--max_blur', help=f'gaussian blur value at the midpoint of the animation, ' f'(in percentage), possible range {list(_LIMITS["blur"])}', type=float, default=MAX_BLUR, metavar='\b') parser.add_argument('-s', '--max_brightness', help=f'brightness value at the midpoint of the animation, ' f'(in percentage), possible range {list(_LIMITS["brightness"])}', type=float, default=MAX_BRIGHTNESS, metavar='\b') parser.add_argument('-z', '--max_zoom', help=f'zoom value at the midpoint of the animation, ' f'(in percentage), possible range {list(_LIMITS["zoom"])}', type=float, default=MAX_ZOOM, metavar='\b') parser.add_argument('-g', '--debug', help='this will show more info, will create a logs file, ' 'and will create a folder which contains animation images', type=str2bool, default=DEBUG, metavar='\b') parser.add_argument('-t', '--art', help='Display ASCII art', type=str2bool, default=ART, metavar='\b') parser.add_argument('-e', '--remove', help='delete original videos after a successful animation creation', type=str2bool, default=REMOVE_ORIGINAL, metavar='\b') parser.add_argument('-m', '--merge', help='merge both phases video chunks into one transition video', type=str2bool, default=MERGE_PHASES, metavar='\b') args = parser.parse_args() if args.animation.lower() == "help": print(_ANIMATION_HELP) exit(0) with tempfile.TemporaryDirectory() as tmp_dir: tmp_path = pathlib.Path(tmp_dir) dh = DataHandler() if not dh.verify_arguments(args, tmp_path): exit(1) actions_determinator = AnimationActions(args.max_zoom, args.max_brightness, args.max_rotation, args.max_blur, args.max_distortion, args.num_frames) phase1_actions, phase2_actions = actions_determinator.get_actions_values(dh.animation) final_phase_folder = AnimationImages.make_transition(dh.tmp_path, dh.phase1_images, dh.phase2_images, phase1_actions, phase2_actions, args.debug) if not dh.final_images_to_video(final_phase_folder): exit(1) if args.merge: if not dh.merge_video_chunks(): exit(1) log_info(f"output transition video: {dh.merged_vid}") else: log_info(f"output transition phase1 video: {dh.phase1_vid}") log_info(f"output transition phase2 video: {dh.phase2_vid}") if args.remove: log_debug(f"remove original video1: {dh.input_vid1}") dh.input_vid1.unlink() log_debug(f"remove original video2: {dh.input_vid2}") dh.input_vid2.unlink() log_info("") log_info((f" Transition finished. Duration = {dh.get_duration_msg()} ".center(80, "="))) log_info("") end_print(args.art) exit(0)