2.1.782

2026-08-16 13:13:16 +02:00 · 2023-11-11 01:43:01 -08:00
parent a9bb1079cf
commit 4fe08161a5
48 changed files with 1041 additions and 2639 deletions
@@ -188,7 +188,7 @@ class SamplerCustom:
                    {"model": ("MODEL",),
                    "add_noise": ("BOOLEAN", {"default": True}),
                    "noise_seed": ("INT", {"default": 0, "min": 0, "max": 0xffffffffffffffff}),
-                    "cfg": ("FLOAT", {"default": 8.0, "min": 0.0, "max": 100.0, "step":0.5, "round": 0.01}),
+                    "cfg": ("FLOAT", {"default": 8.0, "min": 0.0, "max": 100.0, "step":0.1, "round": 0.01}),
                    "positive": ("CONDITIONING", ),
                    "negative": ("CONDITIONING", ),
                    "sampler": ("SAMPLER", ),
@@ -0,0 +1,168 @@
+import folder_paths
+import fcbh.sd
+import fcbh.model_sampling
+import torch
+
+class LCM(fcbh.model_sampling.EPS):
+    def calculate_denoised(self, sigma, model_output, model_input):
+        timestep = self.timestep(sigma).view(sigma.shape[:1] + (1,) * (model_output.ndim - 1))
+        sigma = sigma.view(sigma.shape[:1] + (1,) * (model_output.ndim - 1))
+        x0 = model_input - model_output * sigma
+
+        sigma_data = 0.5
+        scaled_timestep = timestep * 10.0 #timestep_scaling
+
+        c_skip = sigma_data**2 / (scaled_timestep**2 + sigma_data**2)
+        c_out = scaled_timestep / (scaled_timestep**2 + sigma_data**2) ** 0.5
+
+        return c_out * x0 + c_skip * model_input
+
+class ModelSamplingDiscreteLCM(torch.nn.Module):
+    def __init__(self):
+        super().__init__()
+        self.sigma_data = 1.0
+        timesteps = 1000
+        beta_start = 0.00085
+        beta_end = 0.012
+
+        betas = torch.linspace(beta_start**0.5, beta_end**0.5, timesteps, dtype=torch.float32) ** 2
+        alphas = 1.0 - betas
+        alphas_cumprod = torch.cumprod(alphas, dim=0)
+
+        original_timesteps = 50
+        self.skip_steps = timesteps // original_timesteps
+
+
+        alphas_cumprod_valid = torch.zeros((original_timesteps), dtype=torch.float32)
+        for x in range(original_timesteps):
+            alphas_cumprod_valid[original_timesteps - 1 - x] = alphas_cumprod[timesteps - 1 - x * self.skip_steps]
+
+        sigmas = ((1 - alphas_cumprod_valid) / alphas_cumprod_valid) ** 0.5
+        self.set_sigmas(sigmas)
+
+    def set_sigmas(self, sigmas):
+        self.register_buffer('sigmas', sigmas)
+        self.register_buffer('log_sigmas', sigmas.log())
+
+    @property
+    def sigma_min(self):
+        return self.sigmas[0]
+
+    @property
+    def sigma_max(self):
+        return self.sigmas[-1]
+
+    def timestep(self, sigma):
+        log_sigma = sigma.log()
+        dists = log_sigma.to(self.log_sigmas.device) - self.log_sigmas[:, None]
+        return dists.abs().argmin(dim=0).view(sigma.shape) * self.skip_steps + (self.skip_steps - 1)
+
+    def sigma(self, timestep):
+        t = torch.clamp(((timestep - (self.skip_steps - 1)) / self.skip_steps).float(), min=0, max=(len(self.sigmas) - 1))
+        low_idx = t.floor().long()
+        high_idx = t.ceil().long()
+        w = t.frac()
+        log_sigma = (1 - w) * self.log_sigmas[low_idx] + w * self.log_sigmas[high_idx]
+        return log_sigma.exp()
+
+    def percent_to_sigma(self, percent):
+        return self.sigma(torch.tensor(percent * 999.0))
+
+
+def rescale_zero_terminal_snr_sigmas(sigmas):
+    alphas_cumprod = 1 / ((sigmas * sigmas) + 1)
+    alphas_bar_sqrt = alphas_cumprod.sqrt()
+
+    # Store old values.
+    alphas_bar_sqrt_0 = alphas_bar_sqrt[0].clone()
+    alphas_bar_sqrt_T = alphas_bar_sqrt[-1].clone()
+
+    # Shift so the last timestep is zero.
+    alphas_bar_sqrt -= (alphas_bar_sqrt_T)
+
+    # Scale so the first timestep is back to the old value.
+    alphas_bar_sqrt *= alphas_bar_sqrt_0 / (alphas_bar_sqrt_0 - alphas_bar_sqrt_T)
+
+    # Convert alphas_bar_sqrt to betas
+    alphas_bar = alphas_bar_sqrt**2  # Revert sqrt
+    alphas_bar[-1] = 4.8973451890853435e-08
+    return ((1 - alphas_bar) / alphas_bar) ** 0.5
+
+class ModelSamplingDiscrete:
+    @classmethod
+    def INPUT_TYPES(s):
+        return {"required": { "model": ("MODEL",),
+                              "sampling": (["eps", "v_prediction", "lcm"],),
+                              "zsnr": ("BOOLEAN", {"default": False}),
+                              }}
+
+    RETURN_TYPES = ("MODEL",)
+    FUNCTION = "patch"
+
+    CATEGORY = "advanced/model"
+
+    def patch(self, model, sampling, zsnr):
+        m = model.clone()
+
+        sampling_base = fcbh.model_sampling.ModelSamplingDiscrete
+        if sampling == "eps":
+            sampling_type = fcbh.model_sampling.EPS
+        elif sampling == "v_prediction":
+            sampling_type = fcbh.model_sampling.V_PREDICTION
+        elif sampling == "lcm":
+            sampling_type = LCM
+            sampling_base = ModelSamplingDiscreteLCM
+
+        class ModelSamplingAdvanced(sampling_base, sampling_type):
+            pass
+
+        model_sampling = ModelSamplingAdvanced()
+        if zsnr:
+            model_sampling.set_sigmas(rescale_zero_terminal_snr_sigmas(model_sampling.sigmas))
+
+        m.add_object_patch("model_sampling", model_sampling)
+        return (m, )
+
+class RescaleCFG:
+    @classmethod
+    def INPUT_TYPES(s):
+        return {"required": { "model": ("MODEL",),
+                              "multiplier": ("FLOAT", {"default": 0.7, "min": 0.0, "max": 1.0, "step": 0.01}),
+                              }}
+    RETURN_TYPES = ("MODEL",)
+    FUNCTION = "patch"
+
+    CATEGORY = "advanced/model"
+
+    def patch(self, model, multiplier):
+        def rescale_cfg(args):
+            cond = args["cond"]
+            uncond = args["uncond"]
+            cond_scale = args["cond_scale"]
+            sigma = args["sigma"]
+            sigma = sigma.view(sigma.shape[:1] + (1,) * (cond.ndim - 1))
+            x_orig = args["input"]
+
+            #rescale cfg has to be done on v-pred model output
+            x = x_orig / (sigma * sigma + 1.0)
+            cond = ((x - (x_orig - cond)) * (sigma ** 2 + 1.0) ** 0.5) / (sigma)
+            uncond = ((x - (x_orig - uncond)) * (sigma ** 2 + 1.0) ** 0.5) / (sigma)
+
+            #rescalecfg
+            x_cfg = uncond + cond_scale * (cond - uncond)
+            ro_pos = torch.std(cond, dim=(1,2,3), keepdim=True)
+            ro_cfg = torch.std(x_cfg, dim=(1,2,3), keepdim=True)
+
+            x_rescaled = x_cfg * (ro_pos / ro_cfg)
+            x_final = multiplier * x_rescaled + (1.0 - multiplier) * x_cfg
+
+            return x_orig - (x - x_final * sigma / (sigma * sigma + 1.0) ** 0.5)
+
+        m = model.clone()
+        m.set_model_sampler_cfg_function(rescale_cfg)
+        return (m, )
+
+NODE_CLASS_MAPPINGS = {
+    "ModelSamplingDiscrete": ModelSamplingDiscrete,
+    "RescaleCFG": RescaleCFG,
+}
@@ -23,7 +23,7 @@ class Blend:
                    "max": 1.0,
                    "step": 0.01
                }),
-                "blend_mode": (["normal", "multiply", "screen", "overlay", "soft_light"],),
+                "blend_mode": (["normal", "multiply", "screen", "overlay", "soft_light", "difference"],),
            },
        }

@@ -54,6 +54,8 @@ class Blend:
            return torch.where(img1 <= 0.5, 2 * img1 * img2, 1 - 2 * (1 - img1) * (1 - img2))
        elif mode == "soft_light":
            return torch.where(img2 <= 0.5, img1 - (1 - 2 * img2) * img1 * (1 - img1), img1 + (2 * img2 - 1) * (self.g(img1) - img1))
+        elif mode == "difference":
+            return img1 - img2
        else:
            raise ValueError(f"Unsupported blend mode: {mode}")

@@ -126,7 +128,7 @@ class Quantize:
                    "max": 256,
                    "step": 1
                }),
-                "dither": (["none", "floyd-steinberg"],),
+                "dither": (["none", "floyd-steinberg", "bayer-2", "bayer-4", "bayer-8", "bayer-16"],),
            },
        }

@@ -135,19 +137,47 @@ class Quantize:

    CATEGORY = "image/postprocessing"

-    def quantize(self, image: torch.Tensor, colors: int = 256, dither: str = "FLOYDSTEINBERG"):
+    def bayer(im, pal_im, order):
+        def normalized_bayer_matrix(n):
+            if n == 0:
+                return np.zeros((1,1), "float32")
+            else:
+                q = 4 ** n
+                m = q * normalized_bayer_matrix(n - 1)
+                return np.bmat(((m-1.5, m+0.5), (m+1.5, m-0.5))) / q
+
+        num_colors = len(pal_im.getpalette()) // 3
+        spread = 2 * 256 / num_colors
+        bayer_n = int(math.log2(order))
+        bayer_matrix = torch.from_numpy(spread * normalized_bayer_matrix(bayer_n) + 0.5)
+
+        result = torch.from_numpy(np.array(im).astype(np.float32))
+        tw = math.ceil(result.shape[0] / bayer_matrix.shape[0])
+        th = math.ceil(result.shape[1] / bayer_matrix.shape[1])
+        tiled_matrix = bayer_matrix.tile(tw, th).unsqueeze(-1)
+        result.add_(tiled_matrix[:result.shape[0],:result.shape[1]]).clamp_(0, 255)
+        result = result.to(dtype=torch.uint8)
+
+        im = Image.fromarray(result.cpu().numpy())
+        im = im.quantize(palette=pal_im, dither=Image.Dither.NONE)
+        return im
+
+    def quantize(self, image: torch.Tensor, colors: int, dither: str):
        batch_size, height, width, _ = image.shape
        result = torch.zeros_like(image)

-        dither_option = Image.Dither.FLOYDSTEINBERG if dither == "floyd-steinberg" else Image.Dither.NONE
-
        for b in range(batch_size):
-            tensor_image = image[b]
-            img = (tensor_image * 255).to(torch.uint8).numpy()
-            pil_image = Image.fromarray(img, mode='RGB')
+            im = Image.fromarray((image[b] * 255).to(torch.uint8).numpy(), mode='RGB')

-            palette = pil_image.quantize(colors=colors) # Required as described in https://github.com/python-pillow/Pillow/issues/5836
-            quantized_image = pil_image.quantize(colors=colors, palette=palette, dither=dither_option)
+            pal_im = im.quantize(colors=colors) # Required as described in https://github.com/python-pillow/Pillow/issues/5836
+
+            if dither == "none":
+                quantized_image = im.quantize(palette=pal_im, dither=Image.Dither.NONE)
+            elif dither == "floyd-steinberg":
+                quantized_image = im.quantize(palette=pal_im, dither=Image.Dither.FLOYDSTEINBERG)
+            elif dither.startswith("bayer"):
+                order = int(dither.split('-')[-1])
+                quantized_image = Quantize.bayer(im, pal_im, order)

            quantized_array = torch.tensor(np.array(quantized_image.convert("RGB"))).float() / 255
            result[b] = quantized_array
@@ -4,7 +4,7 @@ class LatentRebatch:
    @classmethod
    def INPUT_TYPES(s):
        return {"required": { "latents": ("LATENT",),
-                              "batch_size": ("INT", {"default": 1, "min": 1, "max": 64}),
+                              "batch_size": ("INT", {"default": 1, "min": 1, "max": 4096}),
                              }}
    RETURN_TYPES = ("LATENT",)
    INPUT_IS_LIST = True