Remove RAM check from colorspace detection

The detection uses the already-loaded model with minimal resources
(128x128 test frame, 5 steps), so it works even with low free RAM.
The previous 8GB RAM threshold was incorrectly skipping detection
when models were already loaded in memory.

videogen.py

@@ -1221,13 +1221,8 @@ def detect_model_colorspace(pipe, model_name, m_info, args):
     if existing_colorspace in ["RGB", "BGR"]:
         return existing_colorspace
     
-    # Check available system RAM before attempting detection
-    # Large models can use massive amounts of RAM during inference
-    available_ram_gb = get_available_ram_gb()
-    if available_ram_gb < 8.0:  # Need at least 8GB free RAM
-        print(f"  ⚠️  Low system RAM ({available_ram_gb:.1f}GB), skipping colorspace detection")
-        print(f"     Defaulting to RGB (use --swap_bgr if colors look wrong)")
-        return "RGB"
+    # Note: Detection uses the already-loaded model with minimal resources
+    # (128x128 test frame, 5 inference steps), so it works even with low free RAM
     
     print(f"  🔍 Detecting colorspace for {model_name}...")
     print(f"     (Using minimal resources - 128x128, 5 steps)")