Delete prepare_dataset.py

prepare_dataset.py

@@ -1,45 +0,0 @@
-#!/usr/bin/env python3
-"""Prepare the bundled character images for SDXL style LoRA training.
-
-This script validates the reference set, creates captions/metadata, and can
-optionally create a resized training cache. It never claims that training has
-already happened; it prepares the real dataset used by the trainer.
-"""
-from pathlib import Path
-from PIL import Image
-import json, hashlib
-
-ROOT = Path(__file__).parent
-SRC = ROOT / "training_data"
-OUT = ROOT / "dataset"
-OUT.mkdir(exist_ok=True)
-
-TRIGGER = "offex3dstyle"
-BASE_CAPTION = (
-    f"{TRIGGER}, high quality stylized 3D cartoon character, full body character art, "
-    "exaggerated expressive face, large readable eyes, detailed hands and clothing, "
-    "bold clean shapes, polished 3D render, smooth sculpted forms, strong silhouette, "
-    "centered isolated character, professional game/animation character asset"
-)
-
-images = sorted([p for p in SRC.iterdir() if p.suffix.lower() in {'.png','.jpg','.jpeg','.webp'}])
-if not images:
-    raise SystemExit("No training images found in training_data/")
-
-rows=[]
-for p in images:
-    try:
-        with Image.open(p) as im:
-            im.verify()
-        with Image.open(p) as im:
-            w,h=im.size
-    except Exception as e:
-        raise SystemExit(f"Invalid image {p.name}: {e}")
-    digest=hashlib.sha256(p.read_bytes()).hexdigest()
-    cap=OUT/(p.stem+'.txt')
-    cap.write_text(BASE_CAPTION, encoding='utf-8')
-    rows.append({"file_name": str(Path('..')/'training_data'/p.name), "caption": BASE_CAPTION, "width": w, "height": h, "sha256": digest})
-
-(OUT/'metadata.jsonl').write_text('\n'.join(json.dumps(x,ensure_ascii=False) for x in rows)+'\n', encoding='utf-8')
-(OUT/'dataset_info.json').write_text(json.dumps({"trigger":TRIGGER,"count":len(rows),"caption":BASE_CAPTION,"images":rows},indent=2),encoding='utf-8')
-print(f"Prepared {len(rows)} images. Trigger token: {TRIGGER}")