class MMVSkiaImage:
def __init__(self,
mmvskia_main,
depth=LOG_NO_DEPTH,
from_generator=False) -> None:
debug_prefix = "[MMVSkiaImage.__init__]"
ndepth = depth + LOG_NEXT_DEPTH
self.mmvskia_main = mmvskia_main
self.preludec = self.mmvskia_main.prelude["mmvimage"]
# Log the creation of this class
if self.preludec["log_creation"] and not from_generator:
logging.info(
f"{depth}{debug_prefix} Created new MMVSkiaImage object, getting unique identifier for it"
)
# Get an unique identifier for this MMVSkiaImage object
self.identifier = self.mmvskia_main.utils.get_unique_id(
purpose="MMVSkiaImage object",
depth=ndepth,
silent=self.preludec["log_get_unique_id"] and from_generator)
# The "animation" and path this object will follow
self.animation = {}
# Create classes
self.configure = MMVSkiaImageConfigure(mmvskia_main=self.mmvskia_main,
mmvimage_object=self)
self.image = Frame()
self.x = 0
self.y = 0
self.size = 1
self.rotate_value = 0
self.current_animation = 0
self.current_step = -1
self.is_deletable = False
self.is_vectorial = False
self.type = "mmvimage"
# If we want to get the images from a video, be sure to match the fps!!
self.video = None
# Offset is the animations and motions this frame offset
self.offset = [0, 0]
self.ROUND = 3
self._reset_effects_variables(depth=ndepth)
# Clean this MMVSkiaImage's todo processing or applied
def _reset_effects_variables(self, depth=LOG_NO_DEPTH):
debug_prefix = "[MMVSkiaImage._reset_effects_variables]"
ndepth = depth + LOG_NEXT_DEPTH
# Log action
if self.preludec["_reset_effects_variables"]["log_action"]:
logging.debug(
f"{ndepth}{debug_prefix} [{self.identifier}] Resetting effects variables (filters on image, mask, shaders, paint)"
)
self.image_filters = []
self.mask_filters = []
# self.color_filters = []
self.shaders = []
self.paint_dict = {"AntiAlias": True}
# Our Canvas is an MMVSkiaImage object so we reset it, initialize the animation layers automatically, bla bla
# we don't need the actual configuration from the user apart from post processing accesses by this
# MMVSkiaImage's MMVSkiaImageConfigure class
def create_canvas(self, depth=LOG_NO_DEPTH) -> None:
debug_prefix = "[MMVSkiaImage.create_canvas]"
ndepth = depth + LOG_NEXT_DEPTH
# Log action
if self.preludec["create_canvas"]["log_action"]:
logging.info(
f"{depth}{debug_prefix} [{self.identifier}] Create empty canvas (this ought be the video canvas?)"
)
# Will we be logging the steps?
log_steps = self.preludec["create_canvas"]["log_steps"]
# Initialize blank animation layer
if log_steps:
logging.debug(
f"{ndepth}{debug_prefix} [{self.identifier}] Init animation layer"
)
self.configure.init_animation_layer(depth=ndepth)
# Reset the canvas, create new image of Contex's width and height
if log_steps:
logging.debug(
f"{ndepth}{debug_prefix} [{self.identifier}] Reset canvas")
self.reset_canvas(depth=ndepth)
# Add Path Point at (0, 0)
if log_steps:
logging.debug(
f"{ndepth}{debug_prefix} [{self.identifier}] Add required static path of type Point at (x, y) = (0, 0)"
)
self.configure.add_path_point(x=0, y=0, depth=ndepth)
# Create empty zeros canvas IMAGE, not CONTENTS.
# If we ever wanna mirror the contents and apply post processing
def reset_canvas(self, depth=LOG_NO_DEPTH) -> None:
debug_prefix = "[MMVSkiaImage.reset_canvas]"
ndepth = depth + LOG_NEXT_DEPTH
# Hard debug, this should be executed a lot and we don't wanna clutter the log file or stdout
if self.preludec["reset_canvas"]["log_action"]:
logging.debug(
f"{ndepth}{debug_prefix} [{self.identifier}] Reset canvas, create new image of Context's width and height in size"
)
# Actually create the new canvas
self.image.new(self.mmvskia_main.context.width,
self.mmvskia_main.context.height)
# Next step of animation
def next(self, depth=LOG_NO_DEPTH) -> None:
debug_prefix = "[MMVSkiaImage.next]"
ndepth = depth + LOG_NEXT_DEPTH
# Next step
self.current_step += 1
if self.preludec["next"]["log_current_step"]:
logging.debug(
f"{ndepth}{debug_prefix} [{self.identifier}] Next step, current step = [{self.current_step}]"
)
# Animation has ended, this current_animation isn't present on path.keys
if self.current_animation not in list(self.animation.keys()):
self.is_deletable = True
# Log we are marked to be deleted
if self.preludec["next"]["log_became_deletable"]:
logging.debug(
f"{ndepth}{debug_prefix} [{self.identifier}] Object is out of animation, marking to be deleted"
)
return
# The animation we're currently playing
this_animation = self.animation[self.current_animation]
animation = this_animation["animation"]
steps = animation[
"steps"] * self.mmvskia_main.context.fps_ratio_multiplier # Scale the interpolations
# The current step is one above the steps we've been told, next animation
if self.current_step >= steps + 1:
self.current_animation += 1
self.current_step = 0
return
# Reset offset, pending
self.offset = [0, 0]
self.image.pending = {}
sg = time.time()
self.image.reset_to_original_image()
self._reset_effects_variables()
position = this_animation["position"]
path = position["path"]
if "modules" in this_animation:
modules = this_animation["modules"]
self.is_vectorial = "vectorial" in modules
# The video module must be before everything as it gets the new frame
if "video" in modules:
s = time.time()
this_module = modules["video"]
# We haven't set a video capture or it has ended
if self.video is None:
self.video = cv2.VideoCapture(this_module["path"])
# Can we read next frame? if not, go back to frame 0 for a loop
ok, frame = self.video.read()
if not ok: # cry
self.video.set(cv2.CAP_PROP_POS_FRAMES, 0)
ok, frame = self.video.read()
# CV2 utilizes BGR matrix, but we need RGB
frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGBA)
self.image.load_from_array(frame)
self.image.resize_to_resolution(width=this_module["width"],
height=this_module["height"],
override=True)
if self.preludec["next"]["debug_timings"]:
logging.debug(
f"{depth}{debug_prefix} [{self.identifier}] Video module .next() took [{time.time() - s:.010f}]"
)
if "rotate" in modules:
s = time.time()
this_module = modules["rotate"]
rotate = this_module["object"]
amount = rotate.next()
amount = round(amount, self.ROUND)
if not self.is_vectorial:
self.image.rotate(amount, from_current_frame=True)
else:
self.rotate_value = amount
if self.preludec["next"]["debug_timings"]:
logging.debug(
f"{depth}{debug_prefix} [{self.identifier}] Rotate module .next() took [{time.time() - s:.010f}]"
)
if "resize" in modules:
s = time.time()
this_module = modules["resize"]
resize = this_module["object"]
# Where the vignetting intensity is pointing to according to our
resize.next()
self.size = resize.get_value()
if not self.is_vectorial:
# If we're going to rotate, resize the rotated frame which is not the original image
offset = self.image.resize_by_ratio(
self.size, from_current_frame=True)
if this_module["keep_center"]:
self.offset[0] += offset[0]
self.offset[1] += offset[1]
if self.preludec["next"]["debug_timings"]:
logging.debug(
f"{depth}{debug_prefix} [{self.identifier}] Resize module .next() took [{time.time() - s:.010f}]"
)
# DONE
if "blur" in modules:
s = time.time()
this_module = modules["blur"]
blur = this_module["object"]
blur.next()
amount = blur.get_value()
self.image_filters.append(
skia.ImageFilters.Blur(amount, amount))
if self.preludec["next"]["debug_timings"]:
logging.debug(
f"{depth}{debug_prefix} [{self.identifier}] Blur module .next() took [{time.time() - s:.010f}]"
)
if "fade" in modules:
s = time.time()
this_module = modules["fade"]
fade = this_module["object"]
fade.next()
self.image.transparency(fade.get_value())
if self.preludec["next"]["debug_timings"]:
logging.debug(
f"{depth}{debug_prefix} [{self.identifier}] Fade module .next() took [{time.time() - s:.010f}]"
)
# Apply vignetting
if "vignetting" in modules:
s = time.time()
this_module = modules["vignetting"]
vignetting = this_module["object"]
# Where the vignetting intensity is pointing to according to our
vignetting.next()
vignetting.get_center()
next_vignetting = vignetting.get_value()
# This is a somewhat fake vignetting, we just start a black point with full transparency
# at the center and make a radial gradient that is black with no transparency at the radius
self.mmvskia_main.skia.canvas.drawPaint({
'Shader':
skia.GradientShader.MakeRadial(
center=(vignetting.center_x, vignetting.center_y),
radius=next_vignetting,
colors=[
skia.Color4f(0, 0, 0, 0),
skia.Color4f(0, 0, 0, 1)
])
})
if self.preludec["next"]["debug_timings"]:
logging.debug(
f"{depth}{debug_prefix} [{self.identifier}] Vignetting module .next() took [{time.time() - s:.010f}]"
)
if "vectorial" in modules:
s = time.time()
this_module = modules["vectorial"]
vectorial = this_module["object"]
effects = {
"size": self.size,
"rotate": self.rotate_value,
"image_filters": self.image_filters,
}
# Visualizer blit itself into the canvas automatically
vectorial.next(effects)
if self.preludec["next"]["debug_timings"]:
logging.debug(
f"{depth}{debug_prefix} [{self.identifier}] Vectorial module .next() took [{time.time() - s:.010f}]"
)
if self.preludec["next"]["debug_timings"]:
logging.debug(
f"{depth}{debug_prefix} [{self.identifier}] Global .next() took [{time.time() - sg:.010f}]"
)
# Iterate through every position module
for modifier in path:
# # Override modules
argument = [self.x, self.y] + self.offset
# Move according to a Point (be stationary)
if self.mmvskia_main.utils.is_matching_type(
[modifier], [MMVSkiaModifierPoint]):
# Attribute (x, y) to Point's x and y
[self.x, self.y], self.offset = modifier.next(*argument)
# Move according to a Line (interpolate current steps)
if self.mmvskia_main.utils.is_matching_type([modifier],
[MMVSkiaModifierLine]):
# Interpolate and unpack next coordinate
[self.x, self.y], self.offset = modifier.next(*argument)
# # Offset modules
# Get next shake offset value
if self.mmvskia_main.utils.is_matching_type(
[modifier], [MMVSkiaModifierShake]):
[self.x, self.y], self.offset = modifier.next(*argument)
# Blit this item on the canvas
def blit(self) -> None:
# Vectorial objects blit themselves on a .next() function
if self.is_vectorial:
return
# Invert x and y because top left is (0, 0), y first
y = int(self.x + self.offset[1])
x = int(self.y + self.offset[0])
if self.mask_filters:
self.paint_dict["MaskFilter"] = self.mask_filters
if self.image_filters:
self.paint_dict["ImageFilter"] = skia.ImageFilters.Merge(
self.image_filters)
# Get a paint with the options, image filters (if any) for skia to draw
paint = skia.Paint(self.paint_dict)
# Blit this image
self.mmvskia_main.skia.canvas.drawImage(
self.image.image,
x,
y,
paint=paint,
)
class MMVImage:
def __init__(self, mmv) -> None:
self.mmv = mmv
# The "animation" and path this object will follow
self.animation = {}
# Create classes
self.configure = MMVImageConfigure(self, self.mmv)
self.image = Frame()
self.x = 0
self.y = 0
self.size = 1
self.rotate_value = 0
self.current_animation = 0
self.current_step = -1
self.is_deletable = False
self.is_vectorial = False
self.type = "mmvimage"
self.overlay_mode = "composite" # composite and copy
# If we want to get the images from a video, be sure to match the fps!!
self.video = None
# Offset is the animations and motions this frame offset
self.offset = [0, 0]
self.ROUND = 3
self._reset_effects_variables()
def _reset_effects_variables(self):
self.image_filters = []
self.mask_filters = []
# self.color_filters = []
self.shaders = []
self.paint_dict = {"AntiAlias": True}
# Our Canvas is an MMVImage object
def create_canvas(self) -> None:
self.configure.init_animation_layer()
self.reset_canvas()
self.configure.add_path_point(0, 0)
# Create empty zeros canvas IMAGE, not CONTENTS.
# If we ever wanna mirror the contents and apply post processing
def reset_canvas(self) -> None:
self.image.new(self.mmv.context.width, self.mmv.context.height)
# Don't pickle cv2 video
def __getstate__(self):
state = self.__dict__.copy()
if "video" in state:
del state["video"]
return state
# Next step of animation
def next(self, fftinfo: dict, this_step: int, skia_canvas=None) -> None:
self.current_step += 1
# Animation has ended, this current_animation isn't present on path.keys
if self.current_animation not in list(self.animation.keys()):
self.is_deletable = True
return
# The animation we're currently playing
this_animation = self.animation[self.current_animation]
animation = this_animation["animation"]
steps = animation["steps"]
# The current step is one above the steps we've been told, next animation
if self.current_step == steps + 1:
self.current_animation += 1
self.current_step = 0
return
# Reset offset, pending
self.offset = [0, 0]
self.image.pending = {}
self.image.reset_to_original_image()
self._reset_effects_variables()
position = this_animation["position"]
path = position["path"]
if "modules" in this_animation:
modules = this_animation["modules"]
self.is_vectorial = "vectorial" in modules
# The video module must be before everything as it gets the new frame
if "video" in modules:
this_module = modules["video"]
# We haven't set a video capture or it has ended
if self.video is None:
self.video = cv2.VideoCapture(this_module["path"])
# Can we read next frame? if not, go back to frame 0 for a loop
ok, frame = self.video.read()
if not ok: # cry
self.video.set(cv2.CAP_PROP_POS_FRAMES, 0)
ok, frame = self.video.read()
# CV2 utilizes BGR matrix, but we need RGB
frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGBA)
shake = 0
for modifier in position:
if self.mmv.utils.is_matching_type([modifier],
[MMVModifierShake]):
shake = modifier.distance
width = self.mmv.context.width + (4 * shake)
height = self.mmv.context.height + (4 * shake)
self.image.load_from_array(frame)
self.image.resize_to_resolution(width, height, override=True)
if "rotate" in modules:
this_module = modules["rotate"]
rotate = this_module["object"]
amount = rotate.next()
amount = round(amount, self.ROUND)
if not self.is_vectorial:
self.image.rotate(amount, from_current_frame=True)
else:
self.rotate_value = amount
if "resize" in modules:
this_module = modules["resize"]
resize = this_module["object"]
# Where the vignetting intensity is pointing to according to our
resize.next(fftinfo["average_value"])
self.size = resize.get_value()
if not self.is_vectorial:
# If we're going to rotate, resize the rotated frame which is not the original image
offset = self.image.resize_by_ratio(
self.size, from_current_frame=True)
if this_module["keep_center"]:
self.offset[0] += offset[0]
self.offset[1] += offset[1]
# DONE
if "blur" in modules:
this_module = modules["blur"]
blur = this_module["object"]
blur.next(fftinfo["average_value"])
amount = blur.get_value()
self.image_filters.append(
skia.ImageFilters.Blur(amount, amount))
if "fade" in modules:
this_module = modules["fade"]
fade = this_module["object"]
fade.next(fftinfo["average_value"])
self.image.transparency(fade.get_value())
# Apply vignetting
if "vignetting" in modules:
this_module = modules["vignetting"]
vignetting = this_module["object"]
# Where the vignetting intensity is pointing to according to our
vignetting.next(fftinfo["average_value"])
vignetting.get_center()
next_vignetting = vignetting.get_value()
# This is a somewhat fake vignetting, we just start a black point with full transparency
# at the center and make a radial gradient that is black with no transparency at the radius
skia_canvas.canvas.drawPaint({
'Shader':
skia.GradientShader.MakeRadial(
center=(vignetting.center_x, vignetting.center_y),
radius=next_vignetting,
colors=[
skia.Color4f(0, 0, 0, 0),
skia.Color4f(0, 0, 0, 1)
])
})
if "vectorial" in modules:
this_module = modules["vectorial"]
vectorial = this_module["object"]
effects = {
"size": self.size,
"rotate": self.rotate_value,
"image_filters": self.image_filters,
}
# Visualizer blit itself into the canvas automatically
vectorial.next(fftinfo, this_step, effects)
# Iterate through every position module
for modifier in path:
# # Override modules
argument = [self.x, self.y] + self.offset
# Move according to a Point (be stationary)
if self.mmv.utils.is_matching_type([modifier], [MMVModifierPoint]):
# Attribute (x, y) to Point's x and y
[self.x, self.y], self.offset = modifier.next(*argument)
# Move according to a Line (interpolate current steps)
if self.mmv.utils.is_matching_type([modifier], [MMVModifierLine]):
# Interpolate and unpack next coordinate
[self.x, self.y], self.offset = modifier.next(*argument)
# # Offset modules
# Get next shake offset value
if self.mmv.utils.is_matching_type([modifier], [MMVModifierShake]):
[self.x, self.y], self.offset = modifier.next(*argument)
# Blit this item on the canvas
def blit(self, blit_to_skia) -> None:
# Vectorial objects blit themselves on a .next() function
if self.is_vectorial:
return
# Invert x and y because top left is (0, 0), y first
y = int(self.x + self.offset[1])
x = int(self.y + self.offset[0])
if self.mask_filters:
self.paint_dict["MaskFilter"] = self.mask_filters
if self.image_filters:
self.paint_dict["ImageFilter"] = skia.ImageFilters.Merge(
self.image_filters)
# Get a paint with the options, image filters (if any) for skia to draw
paint = skia.Paint(self.paint_dict)
# Blit this image
blit_to_skia.canvas.drawImage(
self.image.image,
x,
y,
paint=paint,
)