From f533c4755a487cd30d0adb118f2aaaeca1a6510e Mon Sep 17 00:00:00 2001 From: Jason Hoku Date: Sat, 25 Apr 2026 19:26:53 -0700 Subject: [PATCH] fix: skip pre-encoding stage for ltx_video runs (uses dual-clip internally) --- generation_orchestrator.py | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/generation_orchestrator.py b/generation_orchestrator.py index aa1675b..cf4f82c 100644 --- a/generation_orchestrator.py +++ b/generation_orchestrator.py @@ -982,7 +982,11 @@ def run_generation_loop( # ==== PRE-ENCODING STAGE ==== unique_model_keys = set(get_model_cache_key(conf) for conf in expanded) - if not (optional_positive and optional_negative) and expanded and len(unique_model_keys) == 1: + # LTX video runs use their own dual-CLIP encoder inside ltx_video_generate(), + # so skip the standard image-gen pre-encoding entirely when every config is LTX. + all_ltx = expanded and all(c.get("model_type") == "ltx_video" for c in expanded) + + if not all_ltx and not (optional_positive and optional_negative) and expanded and len(unique_model_keys) == 1: first_conf = expanded[0] target_model_name = first_conf["model"]