From a46c4c7146ff6197adca19d77a741ea001795fb7 Mon Sep 17 00:00:00 2001
From: digital-garbage <133010426+digital-garbage@users.noreply.github.com>
Date: Mon, 25 May 2026 18:03:41 +0300
Subject: [PATCH] Fix transition decode: flatten [B, T, H, W, C] to [N, H, W,
 C]

LTXAV video VAE returns 5D tensor; _blur_pixel_frame expected 3D frames.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
---
 samplers.py | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/samplers.py b/samplers.py
index a58a000f..171d0d18 100644
--- a/samplers.py
+++ b/samplers.py
@@ -1274,11 +1274,16 @@ class FunPackLTXAVSceneChainSampler:
 
             window = self._time_slice(video_tensor, start_l, end_l)
             try:
-                decoded = vae.decode(window)  # [N, H, W, C]
+                decoded = vae.decode(window)
             except Exception as exc:
                 print(f"[FunPackSceneChain] {effect} decode failed: {exc}")
                 continue
 
+            # Video VAEs may return [B, T, H, W, C] — flatten to [N, H, W, C]
+            if decoded.dim() == 5:
+                b, t, h, w, c = decoded.shape
+                decoded = decoded.reshape(b * t, h, w, c)
+
             decoded = decoded.clone().float()
             n = decoded.shape[0]
             center = n // 2