"""투룸 옥탑방 도면 v3 + S5_Shot3 + S25_Shot13 두 케이스 검증. 시나리오 분석 결과: - L05는 단칸이 아닌 투룸 옥탑방 (안방=민숙 방 + 수리영 방 + 거실/주방 + 욕실 + 현관 + 옥상 외부). - 도면 v1, v2 모두 단칸으로 그려져 잘못. v3는 multi-room 정확히 표현. 테스트 두 케이스: Variant F: S5_Shot3 (낮 / 거실 / 민숙이 TV 보는 순간) - chain bg 노드 'interior_living_kitchen_day_normal' 재생성 - 도면 v3 + 새 chain bg + C04 + P06 + 기존 prompt - 검증: TV 이중 합성 회피 + 정확한 거실 layout Variant G: S25_Shot13 (해질 / 거실 disturbed / 수리영이 옥상으로 나가는 순간) - chain bg 노드 'interior_living_room_dusk_disturbed' 재생성 - 도면 v3 + 새 chain bg + C01 + P01 + P04 + 기존 prompt - 검증: front door 너머가 옥상(rooftop concrete)인지 vs 베란다인지 """ from __future__ import annotations import base64 import json import os import socket import sys import time import urllib.error import urllib.request from pathlib import Path try: from dotenv import load_dotenv load_dotenv(Path(__file__).parent.parent / ".env") except ImportError: pass from openai import OpenAI PID = "c00bbe19-a9b5-463f-acfc-806f2e820258" EID = "fe165e3a-19c2-4a0f-9acb-e0c9bab0ee5a" PROJECT_ROOT = Path("/Users/manta/Documents/Projects/TheRoad-I1") OUT_DIR = Path(__file__).parent / "experiment_results" # S5_Shot3 references C04_FACE = PROJECT_ROOT / "projects" / PID / "images" / EID / "reference" / "008b5d4c-5a02-4233-853e-ffc99225cfe1.png" P06_REF = PROJECT_ROOT / "projects" / PID / "images" / EID / "reference" / "48735792-33c0-415b-afdf-46146fe19ed4.png" # S25_Shot13 references C01_FACE = PROJECT_ROOT / "projects" / PID / "images" / EID / "reference" / "aa3ea6a8-2dd8-4e10-a520-4964146baa7f.png" P01_REF = PROJECT_ROOT / "projects" / PID / "images" / EID / "reference" / "2add2a95-eb2e-4365-8613-0ee619d2cf8b.png" P04_REF = PROJECT_ROOT / "projects" / PID / "images" / EID / "reference" / "d73172cc-2acb-40be-98f4-302de81060ad.png" # ────────── Step 1: 투룸 도면 v3 prompt ────────── FLOORPLAN_V3_PROMPT = ( "Top-down architectural floor plan of a Korean two-room rooftop apartment " "(투룸 옥탑방). Multi-room layout viewed directly from above. The whole " "apartment sits on top of a building roof — the front entry door opens " "outward to ROOFTOP CONCRETE FLOOR (NOT a balcony, NOT a street level, " "NOT a hallway). " "Layout (the apartment occupies most of the canvas; rooftop area is below):" "(1) LIVING/KITCHEN ROOM (largest, center): TV on a low stand against the " "left wall (label 'TV'), kitchen counter with sink along the upper " "(back) wall (label 'SINK'), small window with curtain to the right of " "the sink along the same back wall (label 'WINDOW'), wooden dining " "table with chairs in the center floor (label 'TABLE'). " "(2) MAIN BEDROOM (안방 — mother's room, top-right corner): bed inside " "(label 'MAIN BEDROOM'). Accessed via interior door from the living room. " "(3) DAUGHTER'S BEDROOM (수리영 방, right side, below main bedroom): bed " "inside, curtain over the back window of this room (label 'DAUGHTER " "BEDROOM'). Accessed via interior door from the living room. " "(4) BATHROOM (bottom-right corner of the apartment): mirror over sink " "(label 'BATHROOM'). Accessed via interior door from the living room. " "(5) ENTRYWAY (현관, lower-left of the apartment): a small space between " "the front entry door and the living room (label 'ENTRYWAY'). " "(6) FRONT ENTRY DOOR (lower wall of the apartment): a door swinging " "outward toward the rooftop, with arc symbol showing swing direction " "(label 'FRONT DOOR'). " "OUTSIDE THE APARTMENT (drawn at the very bottom of the canvas, below " "the entry door): a horizontal strip labeled 'ROOFTOP' showing concrete " "floor texture (parallel lines representing concrete tiles) and a thin " "metal railing along the bottom edge of the canvas (label 'ROOFTOP " "RAILING' or 'PARAPET'). This rooftop strip indicates the apartment is " "ON TOP OF A BUILDING and the entry door opens to outdoor rooftop, NOT " "a hallway or balcony. " "Floor surface inside the apartment: wooden floorboards indicated by " "light parallel hatching. " "Style: clean schematic architectural diagram, plain black ink lines on " "off-white paper, English labels in small caps placed clearly next to " "each room/feature, no perspective, no shading, no people." ) # ────────── Step 2: chain bg 노드 1 (S5용 — 거실 day normal) ────────── CHAIN_BG_LIVING_DAY_PROMPT = ( "Photorealistic cinematic interior background of a Korean two-room " "rooftop apartment (투룸 옥탑방), pale daylight, no people, " "single static frame, 35mm film grain, eye-level wide angle. Match the " "architectural floor plan reference (top-down) EXACTLY for furniture " "and door positions. " "Camera is positioned in the LIVING/KITCHEN room near the front entry " "side, looking diagonally toward the back where the kitchen counter " "and the two interior bedroom doors are visible. " "Visible in this view: ONE television on a low wooden stand against " "the left wall (single TV — do NOT add a second TV); kitchen counter " "with stainless sink in the middle of the back wall; small window with " "thin curtain to the right of the sink; wooden dining table with two or " "three chairs in the center floor; the closed interior door to the main " "bedroom (안방) at the upper-right; the closed interior door to the " "daughter's bedroom (수리영 방) on the right wall further forward. " "The front entry door is offscreen behind the camera, ignore it. " "DO NOT show interior of bedrooms or bathroom — only the living/kitchen " "room is in this frame. The apartment is one rooftop apartment, NOT a " "multi-floor or multi-unit building. " "Material/lighting: dirty plaster walls in muted yellow-gray, worn " "wooden floorboards, low ceiling with a single bare bulb fixture, pale " "daylight from the small back window, soft warm tungsten ambient, dust " "in the air, damp domestic shadows, uneasy stillness, muted blue-gray " "and washed yellow palette." ) # ────────── Step 3: chain bg 노드 2 (S25용 — 거실 dusk disturbed + 옥상 외부) ────────── CHAIN_BG_LIVING_DUSK_DISTURBED_PROMPT = ( "Photorealistic cinematic interior background of the same Korean " "two-room rooftop apartment (투룸 옥탑방), now at DUSK and in a " "disturbed/ransacked state. No people. Single static frame, 35mm film " "grain, eye-level wide angle. Match the architectural floor plan " "reference EXACTLY for furniture and door positions. " "Camera is positioned in the LIVING/KITCHEN room facing the FRONT " "ENTRY DOOR (which is now wide open). The view looks across the " "ransacked living room and OUT through the open front door to the " "ROOFTOP CONCRETE FLOOR with a thin metal railing in the far " "background — the door does NOT open to a balcony, hallway, or " "city view; it opens to a flat building rooftop with concrete tiles. " "Visible in this view: the SAME television on a low wooden stand " "against the left wall (single TV); the kitchen counter with sink " "in the back; the dining table now slightly askew with overturned " "chairs; scattered cushions, papers, and debris on the wooden " "floorboards; the open front entry door framing a rectangle of " "fading dusk sky over the rooftop concrete; the interior bedroom " "doors slightly visible in the background. " "DO NOT depict any balcony, balcony railing, distant cityscape with " "buildings far below, or street-level scenery. The exterior view " "through the door must be: flat rooftop concrete + simple parapet/" "railing + dusk sky. " "Material/lighting: muted plaster walls in cold blue-gray, worn " "wooden floorboards, fading dusk light leaking in through the open " "front door and a partially open back window, hollow underlit feel, " "uneasy stillness, muted blue-gray and washed yellow palette." ) # ────────── Step 4: scene image prompts (DB에서 가져온 그대로) ────────── S5_SHOT3_PROMPT = ( "Photorealistic cinematic still. " "[L05: A cramped modern Korean 옥탑방 (rooftop room) interior, dusty small " "window, worn sink, low table, pale daylight, damp muted domestic shadows.] " "C04O06 in a plain apron occupies the right-center of the frame, at " "three-quarter angle from the left, one hand gripping the apron hem, eyes " "fixed on the television at the far left. The lens emphasizes her serious " "expression, tightened jaw, and the tense hand at the apron hem through a " "thin veil of white kitchen steam in the left foreground. A blurred " "television edge, screen angled away, sits at the far left, its cold " "blue-gray glow brushing her cheek; pale daylight from the closed dusty " "window adds washed yellow highlights. A steaming pot with a slightly " "lifted lid sits in the foreground, P06 rests rinsed beside the sink, and " "a thin stream from the faucet falls into the basin in the background. " "Muted blue-gray and washed yellow palette, damp domestic surfaces, uneasy " "stillness." ) S25_SHOT13_PROMPT = ( "Photorealistic cinematic still. From just inside the Oktapbang threshold " "at shoulder height, a young Korean woman in active casual clothes stands " "centered with back to camera, one foot beyond the open front door and " "the other still inside, P04 strapped to her shoulders. Her head is " "angled toward the doorway, gaze directed outside. Focus on the small " "creased P01 clenched low at her side near her thigh, its bloodstained " "partial front and edge visible, smaller than her hand. The front door is " "open away from the camera toward the exterior, slightly wind-tugged, " "forming a dark boundary around her silhouette. [L05: ransacked Oktapbang " "interior in fading dusk, open window, overturned chairs, murky blue-gray " "shadows, cramped domestic clutter receding toward the doorway] Cold " "evening light from the open window and doorway outlines her shoulders " "and backpack, preserving body volume against heavy dim shadows; muted " "blue-gray tones dominate the abandoned living area in the lower frame." ) GEMINI_API_URL_TEMPLATE = ( "https://generativelanguage.googleapis.com/v1beta/models/{model}:" "generateContent?key={api_key}" ) GEMINI_MODEL = "gemini-3.1-flash-image-preview" ASPECT_RATIO = "16:9" TIMEOUT = 240 MAX_RETRIES = 2 def gemini_generate(api_key: str, prompt: str, input_images: list) -> bytes: parts: list = [{"text": prompt}] for label, img_bytes in input_images: if label: parts.append({"text": label}) parts.append({ "inline_data": { "mime_type": "image/png", "data": base64.b64encode(img_bytes).decode("ascii"), } }) body = { "contents": [{"parts": parts}], "generationConfig": { "responseModalities": ["TEXT", "IMAGE"], "imageConfig": {"aspectRatio": ASPECT_RATIO, "imageSize": "2K"}, }, } url = GEMINI_API_URL_TEMPLATE.format(model=GEMINI_MODEL, api_key=api_key) req = urllib.request.Request( url, data=json.dumps(body).encode("utf-8"), headers={"Content-Type": "application/json"}, method="POST", ) last = None for attempt in range(1, MAX_RETRIES + 2): try: with urllib.request.urlopen(req, timeout=TIMEOUT) as resp: payload = json.loads(resp.read().decode("utf-8")) break except urllib.error.HTTPError as exc: text = exc.read().decode("utf-8", errors="replace") last = RuntimeError(f"HTTP {exc.code}: {text}") if exc.code in {429, 500, 502, 503, 504} and attempt <= MAX_RETRIES: time.sleep(2 * attempt); continue raise last from exc except (urllib.error.URLError, socket.timeout) as exc: last = exc if attempt <= MAX_RETRIES: time.sleep(2 * attempt); continue raise RuntimeError(f"URL error: {exc}") from exc else: raise RuntimeError(f"Failed: {last}") pf = payload.get("promptFeedback", {}) if pf.get("blockReason"): raise RuntimeError(f"Moderation: {pf['blockReason']}") cands = payload.get("candidates", []) if cands and cands[0].get("finishReason") == "SAFETY": raise RuntimeError("Safety filter") for cand in cands: for part in cand.get("content", {}).get("parts", []): inline = part.get("inlineData") or part.get("inline_data") if isinstance(inline, dict) and inline.get("data"): return base64.b64decode(inline["data"]) raise RuntimeError(f"No image: {json.dumps(payload)[:300]}") def gpt_generate(client: OpenAI, prompt: str, ref_path: Path = None) -> bytes: if ref_path: with open(ref_path, "rb") as f: resp = client.images.edit(model="gpt-image-2", image=f, prompt=prompt, size="1536x1024", quality="high", n=1) else: resp = client.images.generate(model="gpt-image-2", prompt=prompt, size="1024x1024", quality="high", n=1) return base64.b64decode(resp.data[0].b64_json) def main(): ok = os.environ.get("OPENAI_API_KEY", "") gk = os.environ.get("GEMINI_API_KEY", "") if not ok or not gk: print("ERROR: OPENAI_API_KEY + GEMINI_API_KEY required", file=sys.stderr) sys.exit(1) OUT_DIR.mkdir(exist_ok=True) client = OpenAI(api_key=ok) # Step 1: 도면 v3 print("="*60); print("Step 1: 투룸 도면 v3"); print("="*60) fp_path = OUT_DIR / "30_floorplan_v3_tworoom.png" if fp_path.exists(): print(f"reuse: {fp_path}") else: print(f"prompt: {len(FLOORPLAN_V3_PROMPT)} chars") fp_path.write_bytes(gpt_generate(client, FLOORPLAN_V3_PROMPT)) print(f"OK: {fp_path} ({fp_path.stat().st_size:,} bytes)") print() # Step 2: chain bg 노드 1 (거실 day normal) print("="*60); print("Step 2: chain bg #1 (거실 day normal)"); print("="*60) bg1_path = OUT_DIR / "31_chain_bg_v3_living_day.png" if bg1_path.exists(): print(f"reuse: {bg1_path}") else: print(f"prompt: {len(CHAIN_BG_LIVING_DAY_PROMPT)} chars") bg1_path.write_bytes(gpt_generate(client, CHAIN_BG_LIVING_DAY_PROMPT, fp_path)) print(f"OK: {bg1_path} ({bg1_path.stat().st_size:,} bytes)") print() # Step 3: chain bg 노드 2 (거실 dusk disturbed + 옥상 외부) print("="*60); print("Step 3: chain bg #2 (거실 dusk disturbed + 옥상 외부)"); print("="*60) bg2_path = OUT_DIR / "32_chain_bg_v3_living_dusk_disturbed.png" if bg2_path.exists(): print(f"reuse: {bg2_path}") else: print(f"prompt: {len(CHAIN_BG_LIVING_DUSK_DISTURBED_PROMPT)} chars") bg2_path.write_bytes(gpt_generate(client, CHAIN_BG_LIVING_DUSK_DISTURBED_PROMPT, fp_path)) print(f"OK: {bg2_path} ({bg2_path.stat().st_size:,} bytes)") print() # Variant F: S5_Shot3 print("="*60); print("Variant F: S5_Shot3 (낮 / 거실 / TV 시청)"); print("="*60) f_input = [ ("spatial layout reference (floor plan, top-down) — two-room rooftop apartment", fp_path.read_bytes()), ("background reference (rebuilt from floor plan) — match all furniture and door positions", bg1_path.read_bytes()), ("character C04 identity", C04_FACE.read_bytes()), ("object P06", P06_REF.read_bytes()), ] f_path = OUT_DIR / "s5shot3_F_tworoom_v3.png" print(f"refs: {len(f_input)}, prompt: {len(S5_SHOT3_PROMPT)} chars") try: f_path.write_bytes(gemini_generate(gk, S5_SHOT3_PROMPT, f_input)) print(f"OK: {f_path} ({f_path.stat().st_size:,} bytes)") except Exception as exc: print(f"FAIL: {exc}") f_path = None print() # Variant G: S25_Shot13 print("="*60); print("Variant G: S25_Shot13 (해질 / 옥상 외부 / 수리영 나가는 순간)"); print("="*60) g_input = [ ("spatial layout reference (floor plan, top-down) — two-room rooftop apartment " "with rooftop exterior beyond front door", fp_path.read_bytes()), ("background reference (rebuilt for dusk disturbed state) — front door opens to " "ROOFTOP CONCRETE FLOOR (not balcony, not street); reuse all interior furniture", bg2_path.read_bytes()), ("character C01 identity", C01_FACE.read_bytes()), ("object P01 (small bloodstained photograph)", P01_REF.read_bytes()), ("object P04 (backpack with doll)", P04_REF.read_bytes()), ] g_path = OUT_DIR / "s25shot13_G_tworoom_v3.png" print(f"refs: {len(g_input)}, prompt: {len(S25_SHOT13_PROMPT)} chars") try: g_path.write_bytes(gemini_generate(gk, S25_SHOT13_PROMPT, g_input)) print(f"OK: {g_path} ({g_path.stat().st_size:,} bytes)") except Exception as exc: print(f"FAIL: {exc}") g_path = None print() # metadata meta = { "test": "투룸 옥탑방 도면 v3 + S5_Shot3 + S25_Shot13 검증", "step1_floorplan_v3": {"path": str(fp_path), "prompt_chars": len(FLOORPLAN_V3_PROMPT)}, "step2_chain_bg_living_day": {"path": str(bg1_path), "prompt_chars": len(CHAIN_BG_LIVING_DAY_PROMPT)}, "step3_chain_bg_living_dusk_disturbed": {"path": str(bg2_path), "prompt_chars": len(CHAIN_BG_LIVING_DUSK_DISTURBED_PROMPT)}, "variant_F_S5_Shot3": { "result": str(f_path) if f_path else None, "refs": ["floor plan v3", "chain bg living day", "C04", "P06"], "prompt": S5_SHOT3_PROMPT, }, "variant_G_S25_Shot13": { "result": str(g_path) if g_path else None, "refs": ["floor plan v3", "chain bg dusk disturbed", "C01", "P01", "P04"], "prompt": S25_SHOT13_PROMPT, }, } (OUT_DIR / "tworoom_v3_metadata.json").write_text( json.dumps(meta, ensure_ascii=False, indent=2), encoding="utf-8" ) print(f"metadata: {OUT_DIR / 'tworoom_v3_metadata.json'}") if __name__ == "__main__": main()