From 1d024be9dab6d0289f72a86e3b3845edee21efc2 Mon Sep 17 00:00:00 2001
From: facok <128763816+facok@users.noreply.github.com>
Date: Fri, 20 Mar 2026 17:06:17 +0800
Subject: [PATCH] Revert to 512px default, remove diagnostic logging
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

256 vs 512 difference (97.5% vs 97.3%) is within random variation —
different image_size consumes different random numbers from the same
seed, producing different images rather than a controlled comparison.
Keep 512 consistent with LCS color calibration. Remove per-blur-level
mean logging that was used during development.
---
 core/sharpness.py | 11 ++---------
 nodes/sharpen.py  |  2 +-
 2 files changed, 3 insertions(+), 10 deletions(-)

diff --git a/core/sharpness.py b/core/sharpness.py
index 17a940f..265cdf1 100644
--- a/core/sharpness.py
+++ b/core/sharpness.py
@@ -100,7 +100,7 @@ def _apply_gaussian_blur(images: torch.Tensor, blur_sigma: float) -> torch.Tenso
         return blurred
 
 
-def calibrate_sharpness(vae, num_samples: int = 64, image_size: int = 256,
+def calibrate_sharpness(vae, num_samples: int = 64, image_size: int = 512,
                         blur_levels: Tuple[float, ...] = (0, 0.5, 1, 2, 4, 8),
                         batch_size: int = 8,
                         lcs_data: LCSData = None) -> SharpnessData:
@@ -179,19 +179,12 @@ def calibrate_sharpness(vae, num_samples: int = 64, image_size: int = 256,
     blur_labels_t = torch.tensor(blur_labels, dtype=torch.float32)
     print(f"[LCS Sharpness Calibration] Collected {X.shape[0]} vectors of dimension {X.shape[1]}")
 
-    # Check per-blur-level mean to see if blur affects brightness in latent space
-    for bl in blur_levels:
-        mask_bl = blur_labels_t == bl
-        level_mean = X[mask_bl].mean().item()
-        print(f"[LCS Sharpness Calibration] blur σ={bl}: latent mean={level_mean:.4f} (n={mask_bl.sum().item()})")
-
     # Remove per-vector mean BEFORE PCA.
     # VAE encoding of blurred images shifts the latent mean (non-linear VAE effect).
     # Without this, PCA captures brightness drift as the dominant component.
     # Per-vector zero-mean forces PCA to find patterns in the relative channel
     # structure, not in the absolute level — isolating true sharpness.
-    X_per_mean = X.mean(dim=1, keepdim=True)  # [N, 1]
-    X = X - X_per_mean
+    X = X - X.mean(dim=1, keepdim=True)
 
     # Optionally remove LCS color component to ensure sharpness PC1 is orthogonal to color
     if lcs_data is not None:
diff --git a/nodes/sharpen.py b/nodes/sharpen.py
index 6381c61..7fd0b65 100644
--- a/nodes/sharpen.py
+++ b/nodes/sharpen.py
@@ -61,7 +61,7 @@ class LCSSharpnessCalibrate(io.ComfyNode):
             inputs=[
                 io.Vae.Input("vae", tooltip="VAE model (calibration is cached per-VAE)"),
                 LCS_DATA.Input("lcs_data", optional=True, tooltip="Optional: remove color component to prevent color shifts"),
-                io.Int.Input("image_size", default=256, min=256, max=1024, step=128,
+                io.Int.Input("image_size", default=512, min=256, max=1024, step=128,
                              tooltip="Size of calibration images (higher = more patches averaged, may improve PCA)"),
             ],
             outputs=[