fix: More conservative VRAM estimate for base models with weights

Calculate: base_vram + 2GB + 50%
This ensures a 14B model estimated at 18GB will require ~29GB instead of 22.5GB.

videogen.py

@@ -3909,9 +3909,10 @@ def select_best_model(gen_type, models, vram_gb=24, prefer_quality=True, return_
                 }
             
             # Check VRAM compatibility using base model requirements
-            # LoRAs and fine-tuned weights add significant overhead (25% more)
+            # LoRAs and fine-tuned weights add significant overhead
+            # Conservative estimate: base VRAM + 2GB + 50% for weights/tensors/loras
             base_vram = parse_vram_estimate(base_model_info.get("vram", "~10 GB"))
-            vram_est = base_vram * 1.25  # 25% more for weights/tensors/loras
+            vram_est = base_vram + 2 + (base_vram * 0.50)  # Base + 2GB + 50%
             if allow_bigger_models:
                 # If allowing bigger models, check if VRAM + 75% of available RAM is sufficient
                 available_ram = get_available_ram_gb()