From b58f0ae282ef2ba2ce12e87404cb0331e4212f34 Mon Sep 17 00:00:00 2001
From: CalamitousFelicitousness <iowasovereign@gmail.com>
Date: Tue, 21 Apr 2026 02:13:58 +0100
Subject: [PATCH] fix(ltx): exclude shared VAE from upsample pipe
 balanced_offload

The upsample pipes receive shared.sd_model.vae as a constructor formality;
the forward pass is pure latent to latent. Main pipe already owns that
VAE's accelerate hook lifecycle, so walking it again from the upsample
pipe raises "Cannot copy out of meta tensor" when the main pipe has
offloaded params to meta. Skip it in the walk.
---
 modules/ltx/ltx_process.py | 14 ++++++++++----
 1 file changed, 10 insertions(+), 4 deletions(-)

diff --git a/modules/ltx/ltx_process.py b/modules/ltx/ltx_process.py
index 33de05531..3fde6df45 100644
--- a/modules/ltx/ltx_process.py
+++ b/modules/ltx/ltx_process.py
@@ -358,10 +358,16 @@ def run_ltx(task_id,
                 t4 = time.time()
                 upsamplejob = shared.state.begin('Upsample')
                 try:
+                    # Shared-VAE exclude: both upsample pipes receive shared.sd_model.vae as a
+                    # constructor formality (pure latent -> latent forward). The main pipe already
+                    # owns the VAE's hook lifecycle, so walking it again here hits meta tensors
+                    # from the prior offload pass. Excluding also shortens the walk to the one
+                    # module that actually belongs to this pipe: latent_upsampler.
+                    upsample_exclude = ['vae']
                     if caps.family == '0.9':
                         global upsample_pipe # pylint: disable=global-statement
                         upsample_pipe = load_upsample(upsample_pipe, upsample_repo_id_09)
-                        upsample_pipe = sd_models.apply_balanced_offload(upsample_pipe)
+                        upsample_pipe = sd_models.apply_balanced_offload(upsample_pipe, exclude=upsample_exclude)
                         up_args = {
                             'width': final_w,
                             'height': final_h,
@@ -373,12 +379,12 @@ def run_ltx(task_id,
                         log.debug(f'Video: op=upsample family=0.9 latents={latents.shape} {up_args}')
                         yield None, 'LTX: Upsample in progress...'
                         latents = upsample_pipe(latents=latents, **up_args).frames[0]
-                        upsample_pipe = sd_models.apply_balanced_offload(upsample_pipe)
+                        upsample_pipe = sd_models.apply_balanced_offload(upsample_pipe, exclude=upsample_exclude)
                     else:
                         global upsample_pipe_2x # pylint: disable=global-statement
                         upsample_repo = upsample_repo_id_23 if caps.variant == '2.3' else upsample_repo_id_20
                         upsample_pipe_2x = load_upsample_2x(upsample_pipe_2x, upsample_repo)
-                        upsample_pipe_2x = sd_models.apply_balanced_offload(upsample_pipe_2x)
+                        upsample_pipe_2x = sd_models.apply_balanced_offload(upsample_pipe_2x, exclude=upsample_exclude)
                         # 2.x base pass returns denormalized latents; latents_normalized=False tells the
                         # upsampler "already raw, do not denormalize again".
                         up_args = {
@@ -394,7 +400,7 @@ def run_ltx(task_id,
                         log.debug(f'Video: op=upsample family=2.x latents={latents.shape} auto={auto_refine_upsample} {up_args}')
                         yield None, 'LTX: Upsample in progress...'
                         latents = upsample_pipe_2x(latents=latents, **up_args).frames[0]
-                        upsample_pipe_2x = sd_models.apply_balanced_offload(upsample_pipe_2x)
+                        upsample_pipe_2x = sd_models.apply_balanced_offload(upsample_pipe_2x, exclude=upsample_exclude)
                 except AssertionError as e:
                     yield from abort(e, ok=True, p=p)
                     return