Files
SOLRICKS-comfyui-solricks/VideoAdaptiveAA.py
T

70 lines
2.7 KiB
Python
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
# Created By MSXYZ and Claude 3 Opus
# Optimized and Fixed
import torch
import torch.nn.functional as F
class VideoAdaptiveAA:
def __init__(self):
pass
@classmethod
def INPUT_TYPES(s):
return {
"required": {
"images": ("IMAGE",),
"strength": ("FLOAT", {"default": 1.0, "min": 0.0, "max": 2.0, "step": 0.1}),
"edge_threshold": ("FLOAT", {"default": 0.1, "min": 0.0, "max": 1.0, "step": 0.01}),
"blur_radius": ("INT", {"default": 1, "min": 1, "max": 5, "step": 1}),
},
}
RETURN_TYPES = ("IMAGE",)
FUNCTION = "apply_aa"
CATEGORY = "CustomPostProcess"
def apply_aa(self, images, strength, edge_threshold, blur_radius):
# ComfyUI Image format: [B, H, W, C] -> Torch format: [B, C, H, W]
img = images.permute(0, 3, 1, 2)
# 1. Grayscale dönüşümü (Kenar tespiti için)
grayscale = 0.2989 * img[:, 0:1, :, :] + 0.5870 * img[:, 1:2, :, :] + 0.1140 * img[:, 2:3, :, :]
# 2. Sobel Filtresi ile kenar tespiti
sobel_x = torch.tensor([[-1, 0, 1], [-2, 0, 2], [-1, 0, 1]], dtype=torch.float32).view(1, 1, 3, 3).to(img.device)
sobel_y = torch.tensor([[-1, -2, -1], [0, 0, 0], [1, 2, 1]], dtype=torch.float32).view(1, 1, 3, 3).to(img.device)
edges_x = F.conv2d(grayscale, sobel_x, padding=1)
edges_y = F.conv2d(grayscale, sobel_y, padding=1)
# Sınır aşımını önlemek için ufak bir epsilon (1e-6) ekliyoruz (NaN hatalarını önler)
edges = torch.sqrt(edges_x**2 + edges_y**2 + 1e-6)
# 3. Kenar maskesi oluşturma
mask = (edges > edge_threshold).float()
mask = mask * strength
# KRİTİK DÜZELTME: Maske değerini 0.0 ile 1.0 arasına hapsediyoruz.
# Bu sayede strength > 1 olsa bile negatif renk patlamaları yaşanmaz.
mask = torch.clamp(mask, min=0.0, max=1.0)
# 4. Bulanıklaştırma (Anti-aliasing etkisi)
kernel_size = blur_radius * 2 + 1
blurred = F.avg_pool2d(img, kernel_size=kernel_size, stride=1, padding=blur_radius)
# 5. Adaptive Blending: Sadece kenarları bulanık olanla değiştir
# Formül: Orijinal * (1 - mask) + Bulanık * mask
result = img * (1.0 - mask) + blurred * mask
# Formatı geri çevir [B, C, H, W] -> [B, H, W, C]
result = result.permute(0, 2, 3, 1)
return (result,)
NODE_CLASS_MAPPINGS = {
"VideoAdaptiveAA": VideoAdaptiveAA
}
NODE_DISPLAY_NAME_MAPPINGS = {
"VideoAdaptiveAA": "🚀 Video Adaptive Anti-Aliasing"
}