Download rendering.py from MortalSnow/OTC-Bench: direct link, hf CLI and curl.
- Browser
- Download file 2.43 kB
-
https://huggingface.co/datasets/MortalSnow/OTC-Bench/resolve/main/rendering.py
- Command line
-
hf download hf://datasets/MortalSnow/OTC-Bench/rendering.py
-
curl -L -o rendering.py https://huggingface.co/datasets/MortalSnow/OTC-Bench/resolve/main/rendering.py
2.43 kB
| """Frozen renderer extracted without behavioral changes.""" | |
| from PIL import Image, ImageDraw | |
| def _fit_size(source_size, maximum_size): | |
| source_width, source_height = source_size | |
| max_width, max_height = maximum_size | |
| scale = min(max_width / source_width, max_height / source_height) | |
| return max(1, round(source_width * scale)), max(1, round(source_height * scale)) | |
| def render_donor_patch(crop, size): | |
| width, height = size | |
| patch = Image.new("RGB", size, (128, 128, 128)) | |
| maximum = max(1, int(width * 0.90)), max(1, int(height * 0.90)) | |
| resized = crop.resize(_fit_size(crop.size, maximum), Image.Resampling.LANCZOS) | |
| patch.paste(resized, ((width - resized.width) // 2, (height - resized.height) // 2)) | |
| return patch | |
| def _gray_slot(image, slot): | |
| ImageDraw.Draw(image).rectangle( | |
| (slot[0], slot[1], slot[2] - 1, slot[3] - 1), fill=(128, 128, 128) | |
| ) | |
| def _marked_canvas(original, canvas_size, target_box): | |
| canvas = Image.new("RGB", canvas_size, (128, 128, 128)) | |
| canvas.paste(original, (0, 0)) | |
| ImageDraw.Draw(canvas).rectangle( | |
| (target_box[0], target_box[1], target_box[2] - 1, target_box[3] - 1), | |
| outline=(220, 20, 60), | |
| width=4, | |
| ) | |
| return canvas | |
| def render_target_donor_pair(original, canvas_size, target_box, r, p, s, donor_crop, condition): | |
| """Render one T or N-near condition using an identical donor patch.""" | |
| if condition not in {"T", "N-near"}: | |
| raise ValueError(f"unsupported pair condition: {condition}") | |
| image = _marked_canvas(original, canvas_size, target_box) | |
| for slot in (r, p, s): | |
| _gray_slot(image, slot) | |
| patch = render_donor_patch(donor_crop, (r[2] - r[0], r[3] - r[1])) | |
| destination = r if condition == "T" else p | |
| image.paste(patch, (destination[0], destination[1])) | |
| return image, patch | |
| def render_baseline(original, canvas_size, target_box, r, p, s, condition): | |
| if condition not in {"O", "G"}: | |
| raise ValueError(f"unsupported baseline condition: {condition}") | |
| image = _marked_canvas(original, canvas_size, target_box) | |
| _gray_slot(image, p) | |
| _gray_slot(image, s) | |
| if condition == "G": | |
| _gray_slot(image, r) | |
| return image | |
| def render_donor_check(canvas_size, r, patch): | |
| image = Image.new("RGB", canvas_size, (128, 128, 128)) | |
| image.paste(patch, (r[0], r[1])) | |
| return image | |