Update LightOnOCR target resolution to 1540px

Based on feedback from LightOn authors (

@staghado
), update default
image resolution from 1288px to 1540px to match training settings.

Changes:
- Default target_size: 1288px → 1540px
- Updated docstrings: trained at 200 DPI (not 300 DPI)
- Updated help text to note this matches training

The model was trained with max resolution of 1540px and 200 DPI,
so using these settings should provide optimal quality.

Reference: https://x.com/staghado/status/1849518069623726119

🤖 Generated with Claude Code

Co-Authored-By: Claude <noreply@anthropic.com>

lighton-ocr.py

@@ -81,11 +81,11 @@ def check_cuda_availability():
         logger.info(f"CUDA is available. GPU: {torch.cuda.get_device_name(0)}")
 
 
-def resize_image_to_target(image: Image.Image, target_size: int = 1288) -> Image.Image:
+def resize_image_to_target(image: Image.Image, target_size: int = 1540) -> Image.Image:
     """
     Resize image so longest dimension is target_size while maintaining aspect ratio.
 
-    LightOnOCR is optimized for 1280-1300px longest dimension at 300 DPI.
+    LightOnOCR was trained with images at 1540px max resolution and 200 DPI.
     """
     width, height = image.size
 
@@ -107,12 +107,12 @@ def resize_image_to_target(image: Image.Image, target_size: int = 1288) -> Image
 def make_ocr_message(
     image: Union[Image.Image, Dict[str, Any], str],
     resize: bool = True,
-    target_size: int = 1288,
+    target_size: int = 1540,
 ) -> List[Dict]:
     """
     Create chat message for OCR processing.
 
-    LightOnOCR expects images at 1280-1300px longest dimension for optimal results.
+    LightOnOCR was trained with 1540px max resolution at 200 DPI for optimal results.
     """
     # Convert to PIL Image if needed
     if isinstance(image, Image.Image):
@@ -280,7 +280,7 @@ def main(
     temperature: float = 0.2,
     top_p: float = 0.9,
     gpu_memory_utilization: float = 0.8,
-    target_size: int = 1288,
+    target_size: int = 1540,
     no_resize: bool = False,
     hf_token: str = None,
     split: str = "train",
@@ -577,8 +577,8 @@ Examples:
     parser.add_argument(
         "--target-size",
         type=int,
-        default=1288,
-        help="Target size for longest image dimension in pixels (default: 1288)",
+        default=1540,
+        help="Target size for longest image dimension in pixels (default: 1540, matching training)",
     )
     parser.add_argument(
         "--no-resize",