Spaces:
Running
on
L40S
Running
on
L40S
from io import BytesIO | |
import cv2 | |
import numpy as np | |
import torch | |
from PIL import Image | |
from ..log import log | |
from ..utils import EASINGS, apply_easing, pil2tensor | |
from .transform import MTB_TransformImage | |
def hex_to_rgb(hex_color: str, bgr: bool = False): | |
hex_color = hex_color.lstrip("#") | |
if bgr: | |
return tuple(int(hex_color[i : i + 2], 16) for i in (4, 2, 0)) | |
return tuple(int(hex_color[i : i + 2], 16) for i in (0, 2, 4)) | |
class MTB_BatchFloatMath: | |
def INPUT_TYPES(cls): | |
return { | |
"required": { | |
"reverse": ("BOOLEAN", {"default": False}), | |
"operation": ( | |
["add", "sub", "mul", "div", "pow", "abs"], | |
{"default": "add"}, | |
), | |
} | |
} | |
RETURN_TYPES = ("FLOATS",) | |
CATEGORY = "mtb/utils" | |
FUNCTION = "execute" | |
def execute(self, reverse: bool, operation: str, **kwargs: list[float]): | |
res: list[float] = [] | |
vals = list(kwargs.values()) | |
if reverse: | |
vals = vals[::-1] | |
ref_count = len(vals[0]) | |
for v in vals: | |
if len(v) != ref_count: | |
raise ValueError( | |
f"All values must have the same length (current: {len(v)}, ref: {ref_count}" | |
) | |
match operation: | |
case "add": | |
for i in range(ref_count): | |
result = sum(v[i] for v in vals) | |
res.append(result) | |
case "sub": | |
for i in range(ref_count): | |
result = vals[0][i] - sum(v[i] for v in vals[1:]) | |
res.append(result) | |
case "mul": | |
for i in range(ref_count): | |
result = vals[0][i] * vals[1][i] | |
res.append(result) | |
case "div": | |
for i in range(ref_count): | |
result = vals[0][i] / vals[1][i] | |
res.append(result) | |
case "pow": | |
for i in range(ref_count): | |
result: float = vals[0][i] ** vals[1][i] | |
res.append(result) | |
case "abs": | |
for i in range(ref_count): | |
result = abs(vals[0][i]) | |
res.append(result) | |
case _: | |
log.info(f"For now this mode ({operation}) is not implemented") | |
return (res,) | |
class MTB_BatchFloatNormalize: | |
"""Normalize the values in the list of floats""" | |
def INPUT_TYPES(cls): | |
return { | |
"required": {"floats": ("FLOATS",)}, | |
} | |
RETURN_TYPES = ("FLOATS",) | |
RETURN_NAMES = ("normalized_floats",) | |
CATEGORY = "mtb/batch" | |
FUNCTION = "execute" | |
def execute( | |
self, | |
floats: list[float], | |
): | |
min_value = min(floats) | |
max_value = max(floats) | |
normalized_floats = [ | |
(x - min_value) / (max_value - min_value) for x in floats | |
] | |
log.debug(f"Floats: {floats}") | |
log.debug(f"Normalized Floats: {normalized_floats}") | |
return (normalized_floats,) | |
class MTB_BatchTimeWrap: | |
"""Remap a batch using a time curve (FLOATS)""" | |
def INPUT_TYPES(cls): | |
return { | |
"required": { | |
"target_count": ("INT", {"default": 25, "min": 2}), | |
"frames": ("IMAGE",), | |
"curve": ("FLOATS",), | |
}, | |
} | |
RETURN_TYPES = ("IMAGE", "FLOATS") | |
RETURN_NAMES = ("image", "interpolated_floats") | |
CATEGORY = "mtb/batch" | |
FUNCTION = "execute" | |
def execute( | |
self, target_count: int, frames: torch.Tensor, curve: list[float] | |
): | |
"""Apply time warping to a list of video frames based on a curve.""" | |
log.debug(f"Input frames shape: {frames.shape}") | |
log.debug(f"Curve: {curve}") | |
total_duration = sum(curve) | |
log.debug(f"Total duration: {total_duration}") | |
B, H, W, C = frames.shape | |
log.debug(f"Batch Size: {B}") | |
normalized_times = np.linspace(0, 1, target_count) | |
interpolated_curve = np.interp( | |
normalized_times, np.linspace(0, 1, len(curve)), curve | |
).tolist() | |
log.debug(f"Interpolated curve: {interpolated_curve}") | |
interpolated_frame_indices = [ | |
(B - 1) * value for value in interpolated_curve | |
] | |
log.debug(f"Interpolated frame indices: {interpolated_frame_indices}") | |
rounded_indices = [ | |
int(round(idx)) for idx in interpolated_frame_indices | |
] | |
rounded_indices = np.clip(rounded_indices, 0, B - 1) | |
# Gather frames based on interpolated indices | |
warped_frames = [] | |
for index in rounded_indices: | |
warped_frames.append(frames[index].unsqueeze(0)) | |
warped_tensor = torch.cat(warped_frames, dim=0) | |
log.debug(f"Warped frames shape: {warped_tensor.shape}") | |
return (warped_tensor, interpolated_curve) | |
class MTB_BatchMake: | |
"""Simply duplicates the input frame as a batch""" | |
def INPUT_TYPES(cls): | |
return { | |
"required": { | |
"image": ("IMAGE",), | |
"count": ("INT", {"default": 1}), | |
} | |
} | |
RETURN_TYPES = ("IMAGE",) | |
FUNCTION = "generate_batch" | |
CATEGORY = "mtb/batch" | |
def generate_batch(self, image: torch.Tensor, count): | |
if len(image.shape) == 3: | |
image = image.unsqueeze(0) | |
return (image.repeat(count, 1, 1, 1),) | |
class MTB_BatchShape: | |
"""Generates a batch of 2D shapes with optional shading (experimental)""" | |
def INPUT_TYPES(cls): | |
return { | |
"required": { | |
"count": ("INT", {"default": 1}), | |
"shape": ( | |
["Box", "Circle", "Diamond", "Tube"], | |
{"default": "Circle"}, | |
), | |
"image_width": ("INT", {"default": 512}), | |
"image_height": ("INT", {"default": 512}), | |
"shape_size": ("INT", {"default": 100}), | |
"color": ("COLOR", {"default": "#ffffff"}), | |
"bg_color": ("COLOR", {"default": "#000000"}), | |
"shade_color": ("COLOR", {"default": "#000000"}), | |
"thickness": ("INT", {"default": 5}), | |
"shadex": ("FLOAT", {"default": 0.0}), | |
"shadey": ("FLOAT", {"default": 0.0}), | |
}, | |
} | |
RETURN_TYPES = ("IMAGE",) | |
FUNCTION = "generate_shapes" | |
CATEGORY = "mtb/batch" | |
def generate_shapes( | |
self, | |
count, | |
shape, | |
image_width, | |
image_height, | |
shape_size, | |
color, | |
bg_color, | |
shade_color, | |
thickness, | |
shadex, | |
shadey, | |
): | |
log.debug(f"COLOR: {color}") | |
log.debug(f"BG_COLOR: {bg_color}") | |
log.debug(f"SHADE_COLOR: {shade_color}") | |
# Parse color input to BGR tuple for OpenCV | |
color = hex_to_rgb(color) | |
bg_color = hex_to_rgb(bg_color) | |
shade_color = hex_to_rgb(shade_color) | |
res = [] | |
for x in range(count): | |
# Initialize an image canvas | |
canvas = np.full( | |
(image_height, image_width, 3), bg_color, dtype=np.uint8 | |
) | |
mask = np.zeros((image_height, image_width), dtype=np.uint8) | |
# Compute the center point of the shape | |
center = (image_width // 2, image_height // 2) | |
if shape == "Box": | |
half_size = shape_size // 2 | |
top_left = (center[0] - half_size, center[1] - half_size) | |
bottom_right = (center[0] + half_size, center[1] + half_size) | |
cv2.rectangle(mask, top_left, bottom_right, 255, -1) | |
elif shape == "Circle": | |
cv2.circle(mask, center, shape_size // 2, 255, -1) | |
elif shape == "Diamond": | |
pts = np.array( | |
[ | |
[center[0], center[1] - shape_size // 2], | |
[center[0] + shape_size // 2, center[1]], | |
[center[0], center[1] + shape_size // 2], | |
[center[0] - shape_size // 2, center[1]], | |
] | |
) | |
cv2.fillPoly(mask, [pts], 255) | |
elif shape == "Tube": | |
cv2.ellipse( | |
mask, | |
center, | |
(shape_size // 2, shape_size // 2), | |
0, | |
0, | |
360, | |
255, | |
thickness, | |
) | |
# Color the shape | |
canvas[mask == 255] = color | |
# Apply shading effects to a separate shading canvas | |
shading = np.zeros_like(canvas, dtype=np.float32) | |
shading[:, :, 0] = shadex * np.linspace(0, 1, image_width) | |
shading[:, :, 1] = shadey * np.linspace( | |
0, 1, image_height | |
).reshape(-1, 1) | |
shading_canvas = cv2.addWeighted( | |
canvas.astype(np.float32), 1, shading, 1, 0 | |
).astype(np.uint8) | |
# Apply shading only to the shape area using the mask | |
canvas[mask == 255] = shading_canvas[mask == 255] | |
res.append(canvas) | |
return (pil2tensor(res),) | |
class MTB_BatchFloatFill: | |
"""Fills a batch float with a single value until it reaches the target length""" | |
def INPUT_TYPES(cls): | |
return { | |
"required": { | |
"floats": ("FLOATS",), | |
"direction": (["head", "tail"], {"default": "tail"}), | |
"value": ("FLOAT", {"default": 0.0}), | |
"count": ("INT", {"default": 1}), | |
} | |
} | |
FUNCTION = "fill_floats" | |
RETURN_TYPES = ("FLOATS",) | |
CATEGORY = "mtb/batch" | |
def fill_floats(self, floats, direction, value, count): | |
size = len(floats) | |
if size > count: | |
raise ValueError( | |
f"Size ({size}) is less then target count ({count})" | |
) | |
rem = count - size | |
if direction == "tail": | |
floats = floats + [value] * rem | |
else: | |
floats = [value] * rem + floats | |
return (floats,) | |
class MTB_BatchFloatAssemble: | |
"""Assembles mutiple batches of floats into a single stream (batch)""" | |
def INPUT_TYPES(cls): | |
return {"required": {"reverse": ("BOOLEAN", {"default": False})}} | |
RETURN_TYPES = ("FLOATS",) | |
CATEGORY = "mtb/batch" | |
FUNCTION = "assemble_floats" | |
def assemble_floats(self, reverse: bool, **kwargs: list[float]): | |
res: list[float] = [] | |
if reverse: | |
for x in reversed(kwargs.values()): | |
if x: | |
res += x | |
else: | |
for x in kwargs.values(): | |
if x: | |
res += x | |
return (res,) | |
class MTB_BatchFloat: | |
"""Generates a batch of float values with interpolation""" | |
def INPUT_TYPES(cls): | |
return { | |
"required": { | |
"mode": ( | |
["Single", "Steps"], | |
{"default": "Steps"}, | |
), | |
"count": ("INT", {"default": 2}), | |
"min": ("FLOAT", {"default": 0.0, "step": 0.001}), | |
"max": ("FLOAT", {"default": 1.0, "step": 0.001}), | |
"easing": ( | |
[ | |
"Linear", | |
"Sine In", | |
"Sine Out", | |
"Sine In/Out", | |
"Quart In", | |
"Quart Out", | |
"Quart In/Out", | |
"Cubic In", | |
"Cubic Out", | |
"Cubic In/Out", | |
"Circ In", | |
"Circ Out", | |
"Circ In/Out", | |
"Back In", | |
"Back Out", | |
"Back In/Out", | |
"Elastic In", | |
"Elastic Out", | |
"Elastic In/Out", | |
"Bounce In", | |
"Bounce Out", | |
"Bounce In/Out", | |
], | |
{"default": "Linear"}, | |
), | |
} | |
} | |
FUNCTION = "set_floats" | |
RETURN_TYPES = ("FLOATS",) | |
CATEGORY = "mtb/batch" | |
def set_floats(self, mode, count, min, max, easing): | |
if mode == "Steps" and count == 1: | |
raise ValueError( | |
"Steps mode requires at least a count of 2 values" | |
) | |
keyframes = [] | |
if mode == "Single": | |
keyframes = [min] * count | |
return (keyframes,) | |
for i in range(count): | |
normalized_step = i / (count - 1) | |
eased_step = apply_easing(normalized_step, easing) | |
eased_value = min + (max - min) * eased_step | |
keyframes.append(eased_value) | |
return (keyframes,) | |
class MTB_BatchMerge: | |
"""Merges multiple image batches with different frame counts""" | |
def INPUT_TYPES(cls): | |
return { | |
"required": { | |
"fusion_mode": ( | |
["add", "multiply", "average"], | |
{"default": "average"}, | |
), | |
"fill": (["head", "tail"], {"default": "tail"}), | |
} | |
} | |
RETURN_TYPES = ("IMAGE",) | |
FUNCTION = "merge_batches" | |
CATEGORY = "mtb/batch" | |
def merge_batches(self, fusion_mode: str, fill: str, **kwargs): | |
images = kwargs.values() | |
max_frames = max(img.shape[0] for img in images) | |
adjusted_images = [] | |
for img in images: | |
frame_count = img.shape[0] | |
if frame_count < max_frames: | |
fill_frame = img[0] if fill == "head" else img[-1] | |
fill_frames = fill_frame.repeat( | |
max_frames - frame_count, 1, 1, 1 | |
) | |
adjusted_batch = ( | |
torch.cat((fill_frames, img), dim=0) | |
if fill == "head" | |
else torch.cat((img, fill_frames), dim=0) | |
) | |
else: | |
adjusted_batch = img | |
adjusted_images.append(adjusted_batch) | |
# Merge the adjusted batches | |
merged_image = None | |
for img in adjusted_images: | |
if merged_image is None: | |
merged_image = img | |
else: | |
if fusion_mode == "add": | |
merged_image += img | |
elif fusion_mode == "multiply": | |
merged_image *= img | |
elif fusion_mode == "average": | |
merged_image = (merged_image + img) / 2 | |
return (merged_image,) | |
class MTB_Batch2dTransform: | |
"""Transform a batch of images using a batch of keyframes""" | |
def INPUT_TYPES(cls): | |
return { | |
"required": { | |
"image": ("IMAGE",), | |
"border_handling": ( | |
["edge", "constant", "reflect", "symmetric"], | |
{"default": "edge"}, | |
), | |
"constant_color": ("COLOR", {"default": "#000000"}), | |
}, | |
"optional": { | |
"x": ("FLOATS",), | |
"y": ("FLOATS",), | |
"zoom": ("FLOATS",), | |
"angle": ("FLOATS",), | |
"shear": ("FLOATS",), | |
}, | |
} | |
RETURN_TYPES = ("IMAGE",) | |
FUNCTION = "transform_batch" | |
CATEGORY = "mtb/batch" | |
def get_num_elements( | |
self, param: None | torch.Tensor | list[torch.Tensor] | list[float] | |
) -> int: | |
if isinstance(param, torch.Tensor): | |
return torch.numel(param) | |
elif isinstance(param, list): | |
return len(param) | |
return 0 | |
def transform_batch( | |
self, | |
image: torch.Tensor, | |
border_handling: str, | |
constant_color: str, | |
x: list[float] | None = None, | |
y: list[float] | None = None, | |
zoom: list[float] | None = None, | |
angle: list[float] | None = None, | |
shear: list[float] | None = None, | |
): | |
if all( | |
self.get_num_elements(param) <= 0 | |
for param in [x, y, zoom, angle, shear] | |
): | |
raise ValueError( | |
"At least one transform parameter must be provided" | |
) | |
keyframes: dict[str, list[float]] = { | |
"x": [], | |
"y": [], | |
"zoom": [], | |
"angle": [], | |
"shear": [], | |
} | |
default_vals = {"x": 0, "y": 0, "zoom": 1.0, "angle": 0, "shear": 0} | |
if x and self.get_num_elements(x) > 0: | |
keyframes["x"] = x | |
if y and self.get_num_elements(y) > 0: | |
keyframes["y"] = y | |
if zoom and self.get_num_elements(zoom) > 0: | |
# some easing types like elastic can pull back... maybe it should abs the value? | |
keyframes["zoom"] = [max(x, 0.00001) for x in zoom] | |
if angle and self.get_num_elements(angle) > 0: | |
keyframes["angle"] = angle | |
if shear and self.get_num_elements(shear) > 0: | |
keyframes["shear"] = shear | |
for name, values in keyframes.items(): | |
count = len(values) | |
if count > 0 and count != image.shape[0]: | |
raise ValueError( | |
f"Length of {name} values ({count}) must match number of images ({image.shape[0]})" | |
) | |
if count == 0: | |
keyframes[name] = [default_vals[name]] * image.shape[0] | |
transformer = MTB_TransformImage() | |
res = [ | |
transformer.transform( | |
image[i].unsqueeze(0), | |
keyframes["x"][i], | |
keyframes["y"][i], | |
keyframes["zoom"][i], | |
keyframes["angle"][i], | |
keyframes["shear"][i], | |
border_handling, | |
constant_color, | |
)[0] | |
for i in range(image.shape[0]) | |
] | |
return (torch.cat(res, dim=0),) | |
class MTB_BatchFloatFit: | |
"""Fit a list of floats using a source and target range""" | |
def INPUT_TYPES(cls): | |
return { | |
"required": { | |
"values": ("FLOATS", {"forceInput": True}), | |
"clamp": ("BOOLEAN", {"default": False}), | |
"auto_compute_source": ("BOOLEAN", {"default": False}), | |
"source_min": ("FLOAT", {"default": 0.0, "step": 0.01}), | |
"source_max": ("FLOAT", {"default": 1.0, "step": 0.01}), | |
"target_min": ("FLOAT", {"default": 0.0, "step": 0.01}), | |
"target_max": ("FLOAT", {"default": 1.0, "step": 0.01}), | |
"easing": ( | |
EASINGS, | |
{"default": "Linear"}, | |
), | |
} | |
} | |
FUNCTION = "fit_range" | |
RETURN_TYPES = ("FLOATS",) | |
CATEGORY = "mtb/batch" | |
DESCRIPTION = "Fit a list of floats using a source and target range" | |
def fit_range( | |
self, | |
values: list[float], | |
clamp: bool, | |
auto_compute_source: bool, | |
source_min: float, | |
source_max: float, | |
target_min: float, | |
target_max: float, | |
easing: str, | |
): | |
if auto_compute_source: | |
source_min = min(values) | |
source_max = max(values) | |
from .graph_utils import MTB_FitNumber | |
res = [] | |
fit_number = MTB_FitNumber() | |
for value in values: | |
(transformed_value,) = fit_number.set_range( | |
value, | |
clamp, | |
source_min, | |
source_max, | |
target_min, | |
target_max, | |
easing, | |
) | |
res.append(transformed_value) | |
return (res,) | |
class MTB_PlotBatchFloat: | |
"""Plot floats""" | |
def INPUT_TYPES(cls): | |
return { | |
"required": { | |
"width": ("INT", {"default": 768}), | |
"height": ("INT", {"default": 768}), | |
"point_size": ("INT", {"default": 4}), | |
"seed": ("INT", {"default": 1}), | |
"start_at_zero": ("BOOLEAN", {"default": False}), | |
} | |
} | |
RETURN_TYPES = ("IMAGE",) | |
RETURN_NAMES = ("plot",) | |
FUNCTION = "plot" | |
CATEGORY = "mtb/batch" | |
def plot( | |
self, | |
width: int, | |
height: int, | |
point_size: int, | |
seed: int, | |
start_at_zero: bool, | |
interactive_backend: bool = False, | |
**kwargs, | |
): | |
import matplotlib | |
# NOTE: This is for notebook usage or tests, i.e not exposed to comfy that should always use Agg | |
if not interactive_backend: | |
matplotlib.use("Agg") | |
import matplotlib.pyplot as plt | |
fig, ax = plt.subplots(figsize=(width / 100, height / 100), dpi=100) | |
fig.set_edgecolor("black") | |
fig.patch.set_facecolor("#2e2e2e") | |
# Setting background color and grid | |
ax.set_facecolor("#2e2e2e") # Dark gray background | |
ax.grid(color="gray", linestyle="-", linewidth=0.5, alpha=0.5) | |
# Finding global min and max across all lists for scaling the plot | |
all_values = [value for values in kwargs.values() for value in values] | |
global_min = min(all_values) | |
global_max = max(all_values) | |
y_padding = 0.05 * (global_max - global_min) | |
ax.set_ylim(global_min - y_padding, global_max + y_padding) | |
max_length = max(len(values) for values in kwargs.values()) | |
if start_at_zero: | |
x_values = np.linspace(0, max_length - 1, max_length) | |
else: | |
x_values = np.linspace(1, max_length, max_length) | |
ax.set_xlim(1, max_length) # Set X-axis limits | |
np.random.seed(seed) | |
colors = np.random.rand(len(kwargs), 3) # Generate random RGB values | |
for color, (label, values) in zip(colors, kwargs.items()): | |
ax.plot(x_values[: len(values)], values, label=label, color=color) | |
ax.legend( | |
title="Legend", | |
title_fontsize="large", | |
fontsize="medium", | |
edgecolor="black", | |
loc="best", | |
) | |
# Setting labels and title | |
ax.set_xlabel("Time", fontsize="large", color="white") | |
ax.set_ylabel("Value", fontsize="large", color="white") | |
ax.set_title( | |
"Plot of Values over Time", fontsize="x-large", color="white" | |
) | |
# Adjusting tick colors to be visible on dark background | |
ax.tick_params(colors="white") | |
# Changing color of the axes border | |
for _, spine in ax.spines.items(): | |
spine.set_edgecolor("white") | |
# Rendering the plot into a NumPy array | |
buf = BytesIO() | |
plt.savefig(buf, format="png", bbox_inches="tight") | |
buf.seek(0) | |
image = Image.open(buf) | |
plt.close(fig) # Closing the figure to free up memory | |
return (pil2tensor(image),) | |
def draw_point(self, image, point, color, point_size): | |
x, y = point | |
y = image.shape[0] - 1 - y # Invert Y-coordinate | |
half_size = point_size // 2 | |
x_start, x_end = ( | |
max(0, x - half_size), | |
min(image.shape[1], x + half_size + 1), | |
) | |
y_start, y_end = ( | |
max(0, y - half_size), | |
min(image.shape[0], y + half_size + 1), | |
) | |
image[y_start:y_end, x_start:x_end] = color | |
def draw_line(self, image, start, end, color): | |
x1, y1 = start | |
x2, y2 = end | |
# Invert Y-coordinate | |
y1 = image.shape[0] - 1 - y1 | |
y2 = image.shape[0] - 1 - y2 | |
dx = x2 - x1 | |
dy = y2 - y1 | |
is_steep = abs(dy) > abs(dx) | |
if is_steep: | |
x1, y1 = y1, x1 | |
x2, y2 = y2, x2 | |
swapped = False | |
if x1 > x2: | |
x1, x2 = x2, x1 | |
y1, y2 = y2, y1 | |
swapped = True | |
dx = x2 - x1 | |
dy = y2 - y1 | |
error = int(dx / 2.0) | |
y = y1 | |
ystep = None | |
if y1 < y2: | |
ystep = 1 | |
else: | |
ystep = -1 | |
for x in range(x1, x2 + 1): | |
coord = (y, x) if is_steep else (x, y) | |
image[coord] = color | |
error -= abs(dy) | |
if error < 0: | |
y += ystep | |
error += dx | |
if swapped: | |
image[(x1, y1)] = color | |
image[(x2, y2)] = color | |
DEFAULT_INTERPOLANT = lambda t: t * t * t * (t * (t * 6 - 15) + 10) | |
class MTB_BatchShake: | |
"""Applies a shaking effect to batches of images.""" | |
def INPUT_TYPES(cls): | |
return { | |
"required": { | |
"images": ("IMAGE",), | |
"position_amount_x": ("FLOAT", {"default": 1.0}), | |
"position_amount_y": ("FLOAT", {"default": 1.0}), | |
"rotation_amount": ("FLOAT", {"default": 10.0}), | |
"frequency": ("FLOAT", {"default": 1.0, "min": 0.005}), | |
"frequency_divider": ("FLOAT", {"default": 1.0, "min": 0.005}), | |
"octaves": ("INT", {"default": 1, "min": 1}), | |
"seed": ("INT", {"default": 0}), | |
}, | |
} | |
RETURN_TYPES = ("IMAGE", "FLOATS", "FLOATS", "FLOATS") | |
RETURN_NAMES = ("image", "pos_x", "pos_y", "rot") | |
FUNCTION = "apply_shake" | |
CATEGORY = "mtb/batch" | |
# def interpolant(self, t): | |
# return t * t * t * (t * (t * 6 - 15) + 10) | |
def generate_perlin_noise_2d( | |
self, shape, res, tileable=(False, False), interpolant=None | |
): | |
"""Generate a 2D numpy array of perlin noise. | |
Args: | |
shape: The shape of the generated array (tuple of two ints). | |
This must be a multple of res. | |
res: The number of periods of noise to generate along each | |
axis (tuple of two ints). Note shape must be a multiple of | |
res. | |
tileable: If the noise should be tileable along each axis | |
(tuple of two bools). Defaults to (False, False). | |
interpolant: The interpolation function, defaults to | |
t*t*t*(t*(t*6 - 15) + 10). | |
Returns | |
------- | |
A numpy array of shape shape with the generated noise. | |
Raises | |
------ | |
ValueError: If shape is not a multiple of res. | |
""" | |
interpolant = interpolant or DEFAULT_INTERPOLANT | |
delta = (res[0] / shape[0], res[1] / shape[1]) | |
d = (shape[0] // res[0], shape[1] // res[1]) | |
grid = ( | |
np.mgrid[0 : res[0] : delta[0], 0 : res[1] : delta[1]].transpose( | |
1, 2, 0 | |
) | |
% 1 | |
) | |
# Gradients | |
angles = 2 * np.pi * np.random.rand(res[0] + 1, res[1] + 1) | |
gradients = np.dstack((np.cos(angles), np.sin(angles))) | |
if tileable[0]: | |
gradients[-1, :] = gradients[0, :] | |
if tileable[1]: | |
gradients[:, -1] = gradients[:, 0] | |
gradients = gradients.repeat(d[0], 0).repeat(d[1], 1) | |
g00 = gradients[: -d[0], : -d[1]] | |
g10 = gradients[d[0] :, : -d[1]] | |
g01 = gradients[: -d[0], d[1] :] | |
g11 = gradients[d[0] :, d[1] :] | |
# Ramps | |
n00 = np.sum(np.dstack((grid[:, :, 0], grid[:, :, 1])) * g00, 2) | |
n10 = np.sum(np.dstack((grid[:, :, 0] - 1, grid[:, :, 1])) * g10, 2) | |
n01 = np.sum(np.dstack((grid[:, :, 0], grid[:, :, 1] - 1)) * g01, 2) | |
n11 = np.sum( | |
np.dstack((grid[:, :, 0] - 1, grid[:, :, 1] - 1)) * g11, 2 | |
) | |
# Interpolation | |
t = interpolant(grid) | |
n0 = n00 * (1 - t[:, :, 0]) + t[:, :, 0] * n10 | |
n1 = n01 * (1 - t[:, :, 0]) + t[:, :, 0] * n11 | |
return np.sqrt(2) * ((1 - t[:, :, 1]) * n0 + t[:, :, 1] * n1) | |
def generate_fractal_noise_2d( | |
self, | |
shape, | |
res, | |
octaves=1, | |
persistence=0.5, | |
lacunarity=2, | |
tileable=(True, True), | |
interpolant=None, | |
): | |
"""Generate a 2D numpy array of fractal noise. | |
Args: | |
shape: The shape of the generated array (tuple of two ints). | |
This must be a multiple of lacunarity**(octaves-1)*res. | |
res: The number of periods of noise to generate along each | |
axis (tuple of two ints). Note shape must be a multiple of | |
(lacunarity**(octaves-1)*res). | |
octaves: The number of octaves in the noise. Defaults to 1. | |
persistence: The scaling factor between two octaves. | |
lacunarity: The frequency factor between two octaves. | |
tileable: If the noise should be tileable along each axis | |
(tuple of two bools). Defaults to (True,True). | |
interpolant: The, interpolation function, defaults to | |
t*t*t*(t*(t*6 - 15) + 10). | |
Returns | |
------- | |
A numpy array of fractal noise and of shape shape generated by | |
combining several octaves of perlin noise. | |
Raises | |
------ | |
ValueError: If shape is not a multiple of | |
(lacunarity**(octaves-1)*res). | |
""" | |
interpolant = interpolant or DEFAULT_INTERPOLANT | |
noise = np.zeros(shape) | |
frequency = 1 | |
amplitude = 1 | |
for _ in range(octaves): | |
noise += amplitude * self.generate_perlin_noise_2d( | |
shape, | |
(frequency * res[0], frequency * res[1]), | |
tileable, | |
interpolant, | |
) | |
frequency *= lacunarity | |
amplitude *= persistence | |
return noise | |
def fbm(self, x, y, octaves): | |
# noise_2d = self.generate_fractal_noise_2d((256, 256), (8, 8), octaves) | |
# Now, extract a single noise value based on x and y, wrapping indices if necessary | |
x_idx = int(x) % 256 | |
y_idx = int(y) % 256 | |
return self.noise_pattern[x_idx, y_idx] | |
def apply_shake( | |
self, | |
images, | |
position_amount_x, | |
position_amount_y, | |
rotation_amount, | |
frequency, | |
frequency_divider, | |
octaves, | |
seed, | |
): | |
# Rehash | |
np.random.seed(seed) | |
self.position_offset = np.random.uniform(-1e3, 1e3, 3) | |
self.rotation_offset = np.random.uniform(-1e3, 1e3, 3) | |
self.noise_pattern = self.generate_perlin_noise_2d( | |
(512, 512), (32, 32), (True, True) | |
) | |
# Assuming frame count is derived from the first dimension of images tensor | |
frame_count = images.shape[0] | |
frequency = frequency / frequency_divider | |
# Generate shaking parameters for each frame | |
x_translations = [] | |
y_translations = [] | |
rotations = [] | |
for frame_num in range(frame_count): | |
time = frame_num * frequency | |
x_idx = (self.position_offset[0] + frame_num) % 256 | |
y_idx = (self.position_offset[1] + frame_num) % 256 | |
np_position = np.array( | |
[ | |
self.fbm(x_idx, time, octaves), | |
self.fbm(y_idx, time, octaves), | |
] | |
) | |
# np_position = np.array( | |
# [ | |
# self.fbm(self.position_offset[0] + frame_num, time, octaves), | |
# self.fbm(self.position_offset[1] + frame_num, time, octaves), | |
# ] | |
# ) | |
# np_rotation = self.fbm(self.rotation_offset[2] + frame_num, time, octaves) | |
rot_idx = (self.rotation_offset[2] + frame_num) % 256 | |
np_rotation = self.fbm(rot_idx, time, octaves) | |
x_translations.append(np_position[0] * position_amount_x) | |
y_translations.append(np_position[1] * position_amount_y) | |
rotations.append(np_rotation * rotation_amount) | |
# Convert lists to tensors | |
# x_translations = torch.tensor(x_translations, dtype=torch.float32) | |
# y_translations = torch.tensor(y_translations, dtype=torch.float32) | |
# rotations = torch.tensor(rotations, dtype=torch.float32) | |
# Create an instance of Batch2dTransform | |
transform = MTB_Batch2dTransform() | |
log.debug( | |
f"Applying shaking with parameters: \nposition {position_amount_x}, {position_amount_y}\nrotation {rotation_amount}\nfrequency {frequency}\noctaves {octaves}" | |
) | |
# Apply shaking transformations to images | |
shaken_images = transform.transform_batch( | |
images, | |
border_handling="edge", # Assuming edge handling as default | |
constant_color="#000000", # Assuming black as default constant color | |
x=x_translations, | |
y=y_translations, | |
angle=rotations, | |
)[0] | |
return (shaken_images, x_translations, y_translations, rotations) | |
__nodes__ = [ | |
MTB_BatchFloat, | |
MTB_Batch2dTransform, | |
MTB_BatchShape, | |
MTB_BatchMake, | |
MTB_BatchFloatAssemble, | |
MTB_BatchFloatFill, | |
MTB_BatchFloatNormalize, | |
MTB_BatchMerge, | |
MTB_BatchShake, | |
MTB_PlotBatchFloat, | |
MTB_BatchTimeWrap, | |
MTB_BatchFloatFit, | |
MTB_BatchFloatMath, | |
] | |