Pre-consolidation snapshot: backends, steganalysis, platform presets, and WIP changes

Snapshot of all uncommitted work before merging stegasoo into soosef monorepo.
Includes: pluggable backends registry, steganalysis detection, platform presets,
and various in-progress modifications across core modules.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

src/stegasoo/__init__.py

@@ -46,6 +46,16 @@ from .image_utils import (
     get_image_info,
 )
 
+# Backend registry
+from .backends import EmbeddingBackend, registry as backend_registry
+
+# Platform presets
+from .platform_presets import PLATFORMS, get_preset
+
+# Steganalysis
+from .steganalysis import check_image
+from .backends.registry import BackendNotFoundError
+
 # Steganography functions
 from .steganography import (
     calculate_capacity_by_mode,
@@ -273,6 +283,15 @@ __all__ = [
     "generate_filename",
     # Crypto
     "has_argon2",
+    # Backends
+    "EmbeddingBackend",
+    "backend_registry",
+    "BackendNotFoundError",
+    # Platform presets
+    "get_preset",
+    "PLATFORMS",
+    # Steganalysis
+    "check_image",
     # Steganography
     "has_dct_support",
     "calculate_capacity_by_mode",

src/stegasoo/backends/__init__.py

@@ -0,0 +1,31 @@
+"""
+Stegasoo embedding backends.
+
+Provides a typed plugin interface for all embedding algorithms.
+Backends register with the module-level ``registry`` on import.
+
+Usage::
+
+    from stegasoo.backends import registry
+
+    backend = registry.get("lsb")
+    stego, stats = backend.embed(data, carrier, key)
+"""
+
+from .dct import DCTBackend
+from .lsb import LSBBackend
+from .protocol import EmbeddingBackend
+from .registry import BackendNotFoundError, BackendRegistry, registry
+
+# Auto-register built-in backends
+registry.register(LSBBackend())
+registry.register(DCTBackend())
+
+__all__ = [
+    "EmbeddingBackend",
+    "BackendRegistry",
+    "BackendNotFoundError",
+    "registry",
+    "LSBBackend",
+    "DCTBackend",
+]

src/stegasoo/backends/dct.py

@@ -0,0 +1,69 @@
+"""
+DCT (Discrete Cosine Transform) image embedding backend.
+
+Wraps the existing frequency-domain DCT functions in dct_steganography.py.
+"""
+
+from __future__ import annotations
+
+from typing import Any
+
+
+class DCTBackend:
+    """Frequency-domain DCT embedding for JPEG-resilient steganography."""
+
+    @property
+    def mode(self) -> str:
+        return "dct"
+
+    @property
+    def carrier_type(self) -> str:
+        return "image"
+
+    def is_available(self) -> bool:
+        from ..dct_steganography import HAS_SCIPY
+
+        return HAS_SCIPY
+
+    def embed(
+        self,
+        data: bytes,
+        carrier: bytes,
+        key: bytes,
+        *,
+        progress_file: str | None = None,
+        **options: Any,
+    ) -> tuple[bytes, Any]:
+        from ..dct_steganography import embed_in_dct
+
+        output_format = options.get("dct_output_format", "png")
+        color_mode = options.get("dct_color_mode", "color")
+        quant_step = options.get("quant_step")
+        jpeg_quality = options.get("jpeg_quality")
+        max_dimension = options.get("max_dimension")
+        return embed_in_dct(
+            data, carrier, key, output_format, color_mode, progress_file,
+            quant_step=quant_step, jpeg_quality=jpeg_quality, max_dimension=max_dimension,
+        )
+
+    def extract(
+        self,
+        carrier: bytes,
+        key: bytes,
+        *,
+        progress_file: str | None = None,
+        **options: Any,
+    ) -> bytes | None:
+        from ..dct_steganography import extract_from_dct
+
+        quant_step = options.get("quant_step")
+        try:
+            return extract_from_dct(carrier, key, progress_file, quant_step=quant_step)
+        except Exception:
+            return None
+
+    def calculate_capacity(self, carrier: bytes, **options: Any) -> int:
+        from ..dct_steganography import calculate_dct_capacity
+
+        info = calculate_dct_capacity(carrier)
+        return info.usable_capacity_bytes

src/stegasoo/backends/lsb.py

@@ -0,0 +1,63 @@
+"""
+LSB (Least Significant Bit) image embedding backend.
+
+Wraps the existing spatial-domain LSB functions in steganography.py.
+"""
+
+from __future__ import annotations
+
+from typing import Any
+
+
+class LSBBackend:
+    """Spatial-domain LSB embedding for lossless image formats."""
+
+    @property
+    def mode(self) -> str:
+        return "lsb"
+
+    @property
+    def carrier_type(self) -> str:
+        return "image"
+
+    def is_available(self) -> bool:
+        return True  # Only needs Pillow, which is always present
+
+    def embed(
+        self,
+        data: bytes,
+        carrier: bytes,
+        key: bytes,
+        *,
+        progress_file: str | None = None,
+        **options: Any,
+    ) -> tuple[bytes, Any]:
+        from ..steganography import _embed_lsb
+
+        bits_per_channel = options.get("bits_per_channel", 1)
+        output_format = options.get("output_format", None)
+        stego_bytes, stats, ext = _embed_lsb(
+            data, carrier, key, bits_per_channel, output_format, progress_file
+        )
+        # Attach output extension to stats for callers that need it
+        stats.output_extension = ext  # type: ignore[attr-defined]
+        return stego_bytes, stats
+
+    def extract(
+        self,
+        carrier: bytes,
+        key: bytes,
+        *,
+        progress_file: str | None = None,
+        **options: Any,
+    ) -> bytes | None:
+        from ..steganography import _extract_lsb
+
+        bits_per_channel = options.get("bits_per_channel", 1)
+        return _extract_lsb(carrier, key, bits_per_channel)
+
+    def calculate_capacity(self, carrier: bytes, **options: Any) -> int:
+        from ..steganography import calculate_capacity
+
+        bits_per_channel = options.get("bits_per_channel", 1)
+        return calculate_capacity(carrier, bits_per_channel)

src/stegasoo/backends/protocol.py

@@ -0,0 +1,91 @@
+"""
+Embedding backend protocol definition.
+
+All embedding backends (LSB, DCT, audio, video, etc.) implement this protocol,
+enabling registry-based dispatch instead of if/elif chains.
+"""
+
+from __future__ import annotations
+
+from typing import Any, Protocol, runtime_checkable
+
+
+@runtime_checkable
+class EmbeddingBackend(Protocol):
+    """Protocol that all embedding backends must satisfy.
+
+    Each backend handles a specific embedding mode (e.g. 'lsb', 'dct',
+    'audio_lsb', 'audio_spread') for a specific carrier type ('image',
+    'audio', 'video').
+    """
+
+    @property
+    def mode(self) -> str:
+        """The embedding mode identifier (e.g. 'lsb', 'dct')."""
+        ...
+
+    @property
+    def carrier_type(self) -> str:
+        """The carrier media type: 'image', 'audio', or 'video'."""
+        ...
+
+    def is_available(self) -> bool:
+        """Whether this backend's dependencies are installed."""
+        ...
+
+    def embed(
+        self,
+        data: bytes,
+        carrier: bytes,
+        key: bytes,
+        *,
+        progress_file: str | None = None,
+        **options: Any,
+    ) -> tuple[bytes, Any]:
+        """Embed data into a carrier.
+
+        Args:
+            data: Encrypted payload bytes.
+            carrier: Raw carrier file bytes (image, audio, etc.).
+            key: Derived key for pixel/sample selection.
+            progress_file: Optional progress file path.
+            **options: Backend-specific options (bits_per_channel,
+                       output_format, color_mode, chip_tier, etc.).
+
+        Returns:
+            Tuple of (stego carrier bytes, embed stats).
+        """
+        ...
+
+    def extract(
+        self,
+        carrier: bytes,
+        key: bytes,
+        *,
+        progress_file: str | None = None,
+        **options: Any,
+    ) -> bytes | None:
+        """Extract data from a carrier.
+
+        Args:
+            carrier: Stego carrier file bytes.
+            key: Derived key for pixel/sample selection.
+            progress_file: Optional progress file path.
+            **options: Backend-specific options.
+
+        Returns:
+            Extracted payload bytes, or None if no payload found.
+        """
+        ...
+
+    def calculate_capacity(self, carrier: bytes, **options: Any) -> int:
+        """Calculate maximum embeddable payload size in bytes.
+
+        Args:
+            carrier: Raw carrier file bytes.
+            **options: Backend-specific options (e.g. bits_per_channel).
+
+        Returns:
+            Maximum payload capacity in bytes.
+        """
+        ...

src/stegasoo/backends/registry.py

@@ -0,0 +1,63 @@
+"""
+Backend registry for embedding mode dispatch.
+
+Backends register themselves by mode string. The registry replaces
+if/elif dispatch in steganography.py with a lookup table.
+"""
+
+from __future__ import annotations
+
+from ..exceptions import StegasooError
+from .protocol import EmbeddingBackend
+
+
+class BackendNotFoundError(StegasooError):
+    """Raised when a requested backend mode is not registered."""
+
+
+class BackendRegistry:
+    """Registry mapping mode strings to embedding backends."""
+
+    def __init__(self) -> None:
+        self._backends: dict[str, EmbeddingBackend] = {}
+
+    def register(self, backend: EmbeddingBackend) -> None:
+        """Register a backend for its mode string."""
+        self._backends[backend.mode] = backend
+
+    def get(self, mode: str) -> EmbeddingBackend:
+        """Look up a backend by mode. Raises BackendNotFoundError if not found."""
+        if mode not in self._backends:
+            available = ", ".join(sorted(self._backends.keys())) or "(none)"
+            raise BackendNotFoundError(
+                f"No backend registered for mode '{mode}'. Available: {available}"
+            )
+        return self._backends[mode]
+
+    def has(self, mode: str) -> bool:
+        """Check if a backend is registered for the given mode."""
+        return mode in self._backends
+
+    def available_modes(self, carrier_type: str | None = None) -> list[str]:
+        """List registered mode strings, optionally filtered by carrier type.
+
+        Only includes modes whose backend reports is_available() == True.
+        """
+        return sorted(
+            mode
+            for mode, backend in self._backends.items()
+            if backend.is_available()
+            and (carrier_type is None or backend.carrier_type == carrier_type)
+        )
+
+    def all_modes(self, carrier_type: str | None = None) -> list[str]:
+        """List all registered mode strings (including unavailable ones)."""
+        return sorted(
+            mode
+            for mode, backend in self._backends.items()
+            if carrier_type is None or backend.carrier_type == carrier_type
+        )
+
+
+# Module-level singleton
+registry = BackendRegistry()

src/stegasoo/cli.py

@@ -184,8 +184,14 @@ def cli(ctx, json_output, debug_mode):
 )
 @click.option("--pin", prompt=True, hide_input=True, confirmation_prompt=True, help="PIN code")
 @click.option("--dry-run", is_flag=True, help="Show capacity usage without encoding")
+@click.option(
+    "--platform",
+    type=click.Choice(["telegram", "discord", "signal", "whatsapp"], case_sensitive=False),
+    help="DCT preset for social media platform (implies DCT+JPEG mode)",
+)
+@click.option("--verify/--no-verify", default=True, help="Pre-verify payload survives platform recompression")
 @click.pass_context
-def encode(ctx, carrier, reference, message, file_payload, output, passphrase, pin, dry_run):
+def encode(ctx, carrier, reference, message, file_payload, output, passphrase, pin, dry_run, platform, verify):
     """
     Encode a message or file into an image.
 
@@ -260,29 +266,48 @@ def encode(ctx, carrier, reference, message, file_payload, output, passphrase, p
 
     from .steganography import EMBED_MODE_DCT, EMBED_MODE_LSB
 
+    # Platform preset overrides
+    preset = None
+    if platform:
+        from .platform_presets import get_preset
+
+        preset = get_preset(platform)
+        use_dct = True  # Platform mode implies DCT+JPEG
+        if output_ext not in (".jpg", ".jpeg"):
+            output = str(Path(output).with_suffix(".jpg"))
+            click.echo(f"  Platform mode: output changed to {output}")
+
     try:
+        encode_kwargs = {
+            "reference_photo": reference_data,
+            "carrier_image": carrier_data,
+            "passphrase": passphrase,
+            "pin": pin,
+            "embed_mode": EMBED_MODE_DCT if use_dct else EMBED_MODE_LSB,
+            "dct_output_format": "jpeg" if use_dct else "png",
+        }
+
+        if preset:
+            encode_kwargs["platform"] = platform
+
         if file_payload:
-            # Encode file
-            result = stegasoo_encode_file(
-                filepath=file_payload,
-                reference_photo=reference_data,
-                carrier_image=carrier_data,
-                passphrase=passphrase,
-                pin=pin,
-                embed_mode=EMBED_MODE_DCT if use_dct else EMBED_MODE_LSB,
-                dct_output_format="jpeg" if use_dct else "png",
-            )
+            result = stegasoo_encode_file(filepath=file_payload, **encode_kwargs)
         else:
-            # Encode message
-            result = stegasoo_encode(
-                message=message,
-                reference_photo=reference_data,
-                carrier_image=carrier_data,
-                passphrase=passphrase,
-                pin=pin,
-                embed_mode=EMBED_MODE_DCT if use_dct else EMBED_MODE_LSB,
-                dct_output_format="jpeg" if use_dct else "png",
-            )
+            result = stegasoo_encode(message=message, **encode_kwargs)
+
+        # Pre-verify survival if platform mode
+        if preset and verify:
+            from .crypto import derive_pixel_key
+            from .platform_presets import pre_verify_survival
+
+            pixel_key = derive_pixel_key(reference_data, passphrase, pin)
+            survived = pre_verify_survival(result.stego_image, pixel_key, preset)
+            if not survived:
+                click.echo(
+                    f"  ⚠ Warning: Payload may not survive {preset.name} recompression. "
+                    "Try a larger carrier image or shorter message.",
+                    err=True,
+                )
 
         # Write output
         with open(output, "wb") as f:
@@ -325,8 +350,13 @@ def encode(ctx, carrier, reference, message, file_payload, output, passphrase, p
 @click.option("--passphrase", prompt=True, hide_input=True, help="Passphrase")
 @click.option("--pin", prompt=True, hide_input=True, help="PIN code")
 @click.option("-o", "--output", type=click.Path(), help="Output path for file payloads")
+@click.option(
+    "--platform",
+    type=click.Choice(["telegram", "discord", "signal", "whatsapp"], case_sensitive=False),
+    help="Platform preset (must match encoding platform)",
+)
 @click.pass_context
-def decode(ctx, image, reference, passphrase, pin, output):
+def decode(ctx, image, reference, passphrase, pin, output, platform):
     """
     Decode a message or file from an image.
 
@@ -334,7 +364,7 @@ def decode(ctx, image, reference, passphrase, pin, output):
 
         stegasoo decode encoded.png -r ref.jpg --passphrase --pin
 
-        stegasoo decode encoded.png -r ref.jpg -o ./extracted/
+        stegasoo decode encoded.png -r ref.jpg --platform telegram
     """
     from .decode import decode as stegasoo_decode
 
@@ -344,12 +374,21 @@ def decode(ctx, image, reference, passphrase, pin, output):
     with open(reference, "rb") as f:
         reference_data = f.read()
 
+    # Resolve platform preset for DCT decoding
+    decode_kwargs = {}
+    if platform:
+        from .platform_presets import get_preset
+
+        preset = get_preset(platform)
+        decode_kwargs["platform"] = platform
+
     try:
         result = stegasoo_decode(
             stego_image=stego_data,
             reference_photo=reference_data,
             passphrase=passphrase,
             pin=pin,
+            **decode_kwargs,
         )
 
         if result.is_file:
@@ -1550,9 +1589,9 @@ def info(ctx, full):
 
     # Check for DCT support
     try:
-        from .dct_steganography import HAS_JPEGIO, HAS_SCIPY
+        from .dct_steganography import HAS_JPEGLIB, HAS_SCIPY
 
-        has_dct = HAS_SCIPY and HAS_JPEGIO
+        has_dct = HAS_SCIPY and HAS_JPEGLIB
     except ImportError:
         has_dct = False
 
@@ -2402,6 +2441,66 @@ def tools_convert(image, fmt, quality, output):
     click.echo(f"Converted to: {output}")
 
 
+# =============================================================================
+# STEGANALYSIS COMMANDS
+# =============================================================================
+
+
+@cli.command()
+@click.argument("image", type=click.Path(exists=True))
+@click.option("--json", "as_json", is_flag=True, help="Output as JSON")
+@click.option(
+    "--mode",
+    type=click.Choice(["lsb", "auto"]),
+    default="lsb",
+    help="Analysis mode (default: lsb)",
+)
+def check(image, as_json, mode):
+    """Analyze an image for steganographic detectability.
+
+    Runs chi-square and RS (Regular-Singular) statistical tests to estimate
+    how detectable any hidden data might be. Outputs a risk level.
+
+    Examples:
+
+        stegasoo check carrier.png
+
+        stegasoo check stego.png --json
+
+        stegasoo check suspicious.bmp --mode lsb
+    """
+    from .steganalysis import check_image
+
+    with open(image, "rb") as f:
+        image_data = f.read()
+
+    result = check_image(image_data, mode=mode)
+    result["filename"] = Path(image).name
+
+    if as_json:
+        click.echo(json.dumps(result, indent=2))
+    else:
+        risk = result["risk"]
+        risk_colors = {"low": "green", "medium": "yellow", "high": "red"}
+        risk_display = click.style(risk.upper(), fg=risk_colors.get(risk, "white"), bold=True)
+
+        click.echo(f"\n  Steganalysis: {result['filename']}")
+        click.echo(f"  Image: {result['width']}x{result['height']}, {result['channels']} channels")
+        click.echo(f"  Detectability risk: {risk_display}")
+
+        click.echo("\n  Chi-square (p-values):")
+        for ch, p in result["chi_square"].items():
+            indicator = "!" if p < 0.05 else " "
+            click.echo(f"    {indicator} {ch}: {p:.6f}")
+
+        click.echo("\n  RS embedding estimate:")
+        for ch, est in result["rs"].items():
+            indicator = "!" if est > 0.1 else " "
+            click.echo(f"    {indicator} {ch}: {est:.4f} ({est * 100:.1f}%)")
+
+        click.echo()
+
+
 # =============================================================================
 # ADMIN COMMANDS (Web UI administration)
 # =============================================================================

src/stegasoo/constants.py

@@ -44,7 +44,9 @@ MAGIC_HEADER = b"\x89ST3"
 # Version 1-3: Date-dependent encryption (v3.0.x - v3.1.x)
 # Version 4: Date-independent encryption (v3.2.0)
 # Version 5: Channel key support (v4.0.0) - adds flags byte to header
-FORMAT_VERSION = 5
+# Version 6: HKDF per-message key derivation (v4.4.0) - adds message nonce to header
+FORMAT_VERSION = 6
+FORMAT_VERSION_LEGACY = 5  # For backward-compatible decryption
 
 # Payload type markers
 PAYLOAD_TEXT = 0x01
@@ -66,6 +68,11 @@ ARGON2_PARALLELISM = 4
 # PBKDF2 fallback parameters
 PBKDF2_ITERATIONS = 600000
 
+# HKDF per-message key derivation (v4.4.0 / FORMAT_VERSION 6)
+MESSAGE_NONCE_SIZE = 16  # 128-bit random nonce per message
+HKDF_INFO_ENCRYPT = b"stegasoo-v6-encrypt"  # HKDF info for encryption key
+HKDF_INFO_PIXEL = b"stegasoo-v6-pixel"  # HKDF info for pixel selection key (reserved)
+
 # ============================================================================
 # INPUT LIMITS
 # ============================================================================
@@ -244,6 +251,17 @@ def get_wordlist() -> list[str]:
     return _bip39_words
 
 
+# =============================================================================
+# STEGANALYSIS (v4.4.0)
+# =============================================================================
+
+# Chi-square p-value threshold: HIGH p-value = equalized PoV pairs = suspicious
+STEGANALYSIS_CHI_SUSPICIOUS_THRESHOLD = 0.95  # p > 0.95 → pairs suspiciously equalized
+
+# RS embedding rate thresholds (primary metric): higher = more likely embedded
+STEGANALYSIS_RS_HIGH_THRESHOLD = 0.3  # > 30% estimated embedding → high risk
+STEGANALYSIS_RS_MEDIUM_THRESHOLD = 0.1  # > 10% estimated embedding → medium risk
+
 # =============================================================================
 # DCT STEGANOGRAPHY (v3.0+)
 # =============================================================================

src/stegasoo/crypto.py

@@ -29,7 +29,9 @@ import secrets
 import struct
 
 from cryptography.hazmat.backends import default_backend
+from cryptography.hazmat.primitives import hashes as _hashes
 from cryptography.hazmat.primitives.ciphers import Cipher, algorithms, modes
+from cryptography.hazmat.primitives.kdf.hkdf import HKDFExpand
 from PIL import Image
 
 from .constants import (
@@ -37,9 +39,12 @@ from .constants import (
     ARGON2_PARALLELISM,
     ARGON2_TIME_COST,
     FORMAT_VERSION,
+    FORMAT_VERSION_LEGACY,
+    HKDF_INFO_ENCRYPT,
     IV_SIZE,
     MAGIC_HEADER,
     MAX_FILENAME_LENGTH,
+    MESSAGE_NONCE_SIZE,
     PAYLOAD_FILE,
     PAYLOAD_TEXT,
     PBKDF2_ITERATIONS,
@@ -63,6 +68,7 @@ except ImportError:
     from cryptography.hazmat.primitives.kdf.pbkdf2 import PBKDF2HMAC
 
 
+
 # =============================================================================
 # CHANNEL KEY RESOLUTION
 # =============================================================================
@@ -314,6 +320,30 @@ def derive_pixel_key(
     return hashlib.sha256(material + b"pixel_selection").digest()
 
 
+def derive_message_key(root_key: bytes, nonce: bytes) -> bytes:
+    """
+    Derive a per-message encryption key via HKDF-Expand.
+
+    Each message gets a unique encryption key even with identical credentials,
+    because the nonce is random per message. This provides key diversification:
+    compromising the ciphertext of one message doesn't help with another.
+
+    Args:
+        root_key: 32-byte root key from Argon2id/PBKDF2
+        nonce: 16-byte random nonce (unique per message)
+
+    Returns:
+        32-byte per-message encryption key
+    """
+    hkdf = HKDFExpand(
+        algorithm=_hashes.SHA256(),
+        length=32,
+        info=HKDF_INFO_ENCRYPT + nonce,
+        backend=default_backend(),
+    )
+    return hkdf.derive(root_key)
+
+
 def _pack_payload(
     content: str | bytes | FilePayload,
 ) -> tuple[bytes, int]:
@@ -472,7 +502,12 @@ def encrypt_message(
     """
     try:
         salt = secrets.token_bytes(SALT_SIZE)
-        key = derive_hybrid_key(photo_data, passphrase, salt, pin, rsa_key_data, channel_key)
+        root_key = derive_hybrid_key(photo_data, passphrase, salt, pin, rsa_key_data, channel_key)
+
+        # v6: Per-message key via HKDF — each message gets a unique encryption key
+        message_nonce = secrets.token_bytes(MESSAGE_NONCE_SIZE)
+        key = derive_message_key(root_key, message_nonce)
+
         iv = secrets.token_bytes(IV_SIZE)
 
         # Determine flags
@@ -502,28 +537,36 @@ def encrypt_message(
             "Padded message: %d bytes (payload + %d padding)", len(padded_message), padding_needed
         )
 
-        # Build header for AAD
+        # Build header for AAD (v6: includes nonce in authenticated data)
         header = MAGIC_HEADER + bytes([FORMAT_VERSION, flags])
 
         # Encrypt with AES-256-GCM
         cipher = Cipher(algorithms.AES(key), modes.GCM(iv), backend=default_backend())
         encryptor = cipher.encryptor()
-        encryptor.authenticate_additional_data(header)
+        encryptor.authenticate_additional_data(header + message_nonce)
         ciphertext = encryptor.update(padded_message) + encryptor.finalize()
 
-        total_size = len(header) + len(salt) + len(iv) + len(encryptor.tag) + len(ciphertext)
+        total_size = (
+            len(header)
+            + MESSAGE_NONCE_SIZE
+            + len(salt)
+            + len(iv)
+            + len(encryptor.tag)
+            + len(ciphertext)
+        )
         logger.debug(
-            "Encrypted output: %d bytes (header=%d, salt=%d, iv=%d, tag=%d, ciphertext=%d)",
+            "Encrypted output: %d bytes (header=%d, nonce=%d, salt=%d, iv=%d, tag=%d, ct=%d)",
             total_size,
             len(header),
+            MESSAGE_NONCE_SIZE,
             len(salt),
             len(iv),
             len(encryptor.tag),
             len(ciphertext),
         )
 
-        # v4.0.0: Header with flags byte
-        return header + salt + iv + encryptor.tag + ciphertext
+        # v6: [magic|version|flags|nonce|salt|iv|tag|ciphertext]
+        return header + message_nonce + salt + iv + encryptor.tag + ciphertext
 
     except Exception as e:
         logger.error("Encryption failed: %s", e)
@@ -534,43 +577,78 @@ def parse_header(encrypted_data: bytes) -> dict | None:
     """
     Parse the header from encrypted data.
 
-    v4.0.0: Includes flags byte for channel key indicator.
+    Supports both v5 (legacy) and v6 (HKDF) header formats.
+
+    v5: [magic:4][ver:1][flags:1][salt:32][iv:12][tag:16][ciphertext]  (66+ bytes)
+    v6: [magic:4][ver:1][flags:1][nonce:16][salt:32][iv:12][tag:16][ciphertext]  (82+ bytes)
 
     Args:
         encrypted_data: Raw encrypted bytes
 
     Returns:
-        Dict with salt, iv, tag, ciphertext, flags or None if invalid
+        Dict with version, salt, iv, tag, ciphertext, flags, and optionally
+        message_nonce (v6). Returns None if invalid.
     """
-    # Min size: Magic(4) + Version(1) + Flags(1) + Salt(32) + IV(12) + Tag(16) = 66 bytes
+    # Min v5 size: 4+1+1+32+12+16 = 66 bytes
     if len(encrypted_data) < 66 or encrypted_data[:4] != MAGIC_HEADER:
         return None
 
     try:
         version = encrypted_data[4]
-        if version != FORMAT_VERSION:
+
+        if version == FORMAT_VERSION:
+            # v6: has message nonce
+            if len(encrypted_data) < 82:
+                return None
+            flags = encrypted_data[5]
+            offset = 6
+            message_nonce = encrypted_data[offset : offset + MESSAGE_NONCE_SIZE]
+            offset += MESSAGE_NONCE_SIZE
+            salt = encrypted_data[offset : offset + SALT_SIZE]
+            offset += SALT_SIZE
+            iv = encrypted_data[offset : offset + IV_SIZE]
+            offset += IV_SIZE
+            tag = encrypted_data[offset : offset + TAG_SIZE]
+            offset += TAG_SIZE
+            ciphertext = encrypted_data[offset:]
+
+            return {
+                "version": version,
+                "flags": flags,
+                "has_channel_key": bool(flags & FLAG_CHANNEL_KEY),
+                "message_nonce": message_nonce,
+                "salt": salt,
+                "iv": iv,
+                "tag": tag,
+                "ciphertext": ciphertext,
+            }
+
+        elif version == FORMAT_VERSION_LEGACY:
+            # v5: no nonce
+            flags = encrypted_data[5]
+            offset = 6
+            salt = encrypted_data[offset : offset + SALT_SIZE]
+            offset += SALT_SIZE
+            iv = encrypted_data[offset : offset + IV_SIZE]
+            offset += IV_SIZE
+            tag = encrypted_data[offset : offset + TAG_SIZE]
+            offset += TAG_SIZE
+            ciphertext = encrypted_data[offset:]
+
+            return {
+                "version": version,
+                "flags": flags,
+                "has_channel_key": bool(flags & FLAG_CHANNEL_KEY),
+                "message_nonce": None,
+                "salt": salt,
+                "iv": iv,
+                "tag": tag,
+                "ciphertext": ciphertext,
+            }
+
+        else:
             return None
 
-        flags = encrypted_data[5]
-
-        offset = 6
-        salt = encrypted_data[offset : offset + SALT_SIZE]
-        offset += SALT_SIZE
-        iv = encrypted_data[offset : offset + IV_SIZE]
-        offset += IV_SIZE
-        tag = encrypted_data[offset : offset + TAG_SIZE]
-        offset += TAG_SIZE
-        ciphertext = encrypted_data[offset:]
-
-        return {
-            "version": version,
-            "flags": flags,
-            "has_channel_key": bool(flags & FLAG_CHANNEL_KEY),
-            "salt": salt,
-            "iv": iv,
-            "tag": tag,
-            "ciphertext": ciphertext,
-        }
     except Exception:
         return None
 
@@ -622,12 +700,21 @@ def decrypt_message(
     message_has_key = header["has_channel_key"]
 
     try:
-        key = derive_hybrid_key(
+        root_key = derive_hybrid_key(
             photo_data, passphrase, header["salt"], pin, rsa_key_data, channel_key
         )
 
-        # Reconstruct header for AAD verification
-        aad_header = MAGIC_HEADER + bytes([FORMAT_VERSION, header["flags"]])
+        version = header["version"]
+        message_nonce = header["message_nonce"]
+
+        if version == FORMAT_VERSION and message_nonce is not None:
+            # v6: Derive per-message key via HKDF
+            key = derive_message_key(root_key, message_nonce)
+            aad_header = MAGIC_HEADER + bytes([FORMAT_VERSION, header["flags"]]) + message_nonce
+        else:
+            # v5 (legacy): Root key used directly
+            key = root_key
+            aad_header = MAGIC_HEADER + bytes([FORMAT_VERSION_LEGACY, header["flags"]])
 
         cipher = Cipher(
             algorithms.AES(key), modes.GCM(header["iv"], header["tag"]), backend=default_backend()
@@ -647,7 +734,7 @@ def decrypt_message(
         payload_data = padded_plaintext[:original_length]
         result = _unpack_payload(payload_data)
 
-        logger.debug("Decryption successful: %s", result.payload_type)
+        logger.debug("Decryption successful: %s (v%d)", result.payload_type, version)
         return result
 
     except Exception as e:

src/stegasoo/dct_steganography.py

@@ -12,7 +12,7 @@ Why is this cool?
 
 Two approaches depending on what you want:
 1. PNG output: We do our own DCT math via scipy (works on any image)
-2. JPEG output: We use jpeglib to directly tweak the coefficients (chef's kiss)
+2. JPEG output: We use jpeglib to directly modify the coefficients (chef's kiss)
 
 v4.1.0 - The "please stop corrupting my data" release:
 - Reed-Solomon error correction (can fix up to 16 byte errors per chunk)
@@ -56,13 +56,12 @@ except ImportError:
         idctn = None
 
 # Check for jpeglib availability (for proper JPEG mode)
-# jpeglib replaces jpegio for Python 3.13+ compatibility
 try:
     import jpeglib
 
-    HAS_JPEGIO = True  # Keep variable name for compatibility
+    HAS_JPEGLIB = True
 except ImportError:
-    HAS_JPEGIO = False
+    HAS_JPEGLIB = False
     jpeglib = None
 
 # Import custom exceptions
@@ -170,20 +169,20 @@ QUANT_STEP = 25
 
 # Magic bytes so we can identify our own images
 DCT_MAGIC = b"DCTS"  # scipy DCT mode marker
-JPEGIO_MAGIC = b"JPGS"  # jpegio native JPEG mode marker
+JPEGLIB_MAGIC = b"JPGS"  # jpeglib native JPEG mode marker
 HEADER_SIZE = 10  # Magic (4) + version (1) + flags (1) + length (4)
 
 OUTPUT_FORMAT_PNG = "png"
 OUTPUT_FORMAT_JPEG = "jpeg"
 JPEG_OUTPUT_QUALITY = 95  # High quality but not 100 (100 causes issues, see below)
 
-# For jpegio mode: we only embed in coefficients with magnitude >= 2
+# For jpeglib mode: we only embed in coefficients with magnitude >= 2
 # Coefficients of 0 or 1 are usually quantized noise - unreliable
-JPEGIO_MIN_COEF_MAGNITUDE = 2
+JPEGLIB_MIN_COEF_MAGNITUDE = 2
 
 # We embed in the Y (luminance) channel only - it has the most capacity
 # Cb/Cr are often subsampled 4:2:0 anyway
-JPEGIO_EMBED_CHANNEL = 0
+JPEGLIB_EMBED_CHANNEL = 0
 
 # Header flags
 FLAG_COLOR_MODE = 0x01  # Set if we preserved color (YCbCr mode)
@@ -204,10 +203,10 @@ RS_LENGTH_PREFIX_SIZE = RS_LENGTH_HEADER_SIZE * RS_LENGTH_COPIES  # 24 bytes tot
 MAX_CHUNK_HEIGHT = 512  # Process in strips to keep memory sane
 
 # Fun bug: JPEGs saved with quality=100 have quantization tables full of 1s
-# This makes the DCT coefficients HUGE and jpegio crashes spectacularly
+# This makes the DCT coefficients HUGE and jpeglib crashes spectacularly
 # Solution: detect and re-save at quality 95 first
-JPEGIO_NORMALIZE_QUALITY = 95
-JPEGIO_MAX_QUANT_VALUE_THRESHOLD = 1  # All 1s in quant table = bad news
+JPEGLIB_NORMALIZE_QUALITY = 95
+JPEGLIB_MAX_QUANT_VALUE_THRESHOLD = 1  # All 1s in quant table = bad news
 
 
 # ============================================================================
@@ -261,8 +260,8 @@ def has_dct_support() -> bool:
     return HAS_SCIPY
 
 
-def has_jpegio_support() -> bool:
-    return HAS_JPEGIO
+def has_jpeglib_support() -> bool:
+    return HAS_JPEGLIB
 
 
 # ============================================================================
@@ -654,11 +653,11 @@ def _parse_header(header_bits: list) -> tuple[int, int, int]:
 
 
 # ============================================================================
-# JPEGIO HELPERS
+# JPEGLIB HELPERS
 # ============================================================================
 
 
-def _jpegio_bytes_to_file(data: bytes, suffix: str = ".jpg") -> str:
+def _jpeglib_bytes_to_file(data: bytes, suffix: str = ".jpg") -> str:
     import os
     import tempfile
 
@@ -670,19 +669,19 @@ def _jpegio_bytes_to_file(data: bytes, suffix: str = ".jpg") -> str:
     return path
 
 
-def _jpegio_get_usable_positions(coef_array: np.ndarray) -> list:
+def _jpeglib_get_usable_positions(coef_array: np.ndarray) -> list:
     positions = []
     h, w = coef_array.shape
     for row in range(h):
         for col in range(w):
             if (row % BLOCK_SIZE == 0) and (col % BLOCK_SIZE == 0):
                 continue
-            if abs(coef_array[row, col]) >= JPEGIO_MIN_COEF_MAGNITUDE:
+            if abs(coef_array[row, col]) >= JPEGLIB_MIN_COEF_MAGNITUDE:
                 positions.append((row, col))
     return positions
 
 
-def _jpegio_generate_order(num_positions: int, seed: bytes) -> list:
+def _jpeglib_generate_order(num_positions: int, seed: bytes) -> list:
     hash_bytes = hashlib.sha256(seed + b"jpeg_coef_order").digest()
     rng = np.random.RandomState(int.from_bytes(hash_bytes[:4], "big"))
     order = list(range(num_positions))
@@ -690,15 +689,15 @@ def _jpegio_generate_order(num_positions: int, seed: bytes) -> list:
     return order
 
 
-def _jpegio_create_header(data_length: int, flags: int = 0) -> bytes:
-    return struct.pack(">4sBBI", JPEGIO_MAGIC, 1, flags, data_length)
+def _jpeglib_create_header(data_length: int, flags: int = 0) -> bytes:
+    return struct.pack(">4sBBI", JPEGLIB_MAGIC, 1, flags, data_length)
 
 
-def _jpegio_parse_header(header_bytes: bytes) -> tuple[int, int, int]:
+def _jpeglib_parse_header(header_bytes: bytes) -> tuple[int, int, int]:
     if len(header_bytes) < HEADER_SIZE:
         raise ValueError("Insufficient header data")
     magic, version, flags, length = struct.unpack(">4sBBI", header_bytes[:HEADER_SIZE])
-    if magic != JPEGIO_MAGIC:
+    if magic != JPEGLIB_MAGIC:
         raise InvalidMagicBytesError("Not a Stegasoo JPEG or wrong mode")
     return version, flags, length
 
@@ -782,7 +781,7 @@ def estimate_capacity_comparison(image_data: bytes) -> dict:
             "available": HAS_SCIPY,
         },
         "jpeg_native": {
-            "available": HAS_JPEGIO,
+            "available": HAS_JPEGLIB,
             "note": "Uses jpeglib for proper JPEG coefficient embedding",
         },
     }
@@ -795,24 +794,54 @@ def embed_in_dct(
     output_format: str = OUTPUT_FORMAT_PNG,
     color_mode: str = "color",
     progress_file: str | None = None,
+    quant_step: int | None = None,
+    jpeg_quality: int | None = None,
+    max_dimension: int | None = None,
 ) -> tuple[bytes, DCTEmbedStats]:
-    """Embed data using DCT coefficient modification."""
+    """Embed data using DCT coefficient modification.
+
+    Args:
+        data: Payload bytes to embed.
+        carrier_image: Carrier image bytes.
+        seed: Key for block selection.
+        output_format: 'png' or 'jpeg'.
+        color_mode: 'color' or 'grayscale'.
+        progress_file: Optional progress file.
+        quant_step: Override QIM quantization step (default: QUANT_STEP).
+            Higher = more robust to recompression, more visible.
+        jpeg_quality: Override JPEG output quality (default: JPEG_OUTPUT_QUALITY).
+        max_dimension: Resize carrier if larger than this.
+    """
     if output_format not in (OUTPUT_FORMAT_PNG, OUTPUT_FORMAT_JPEG):
         raise ValueError(f"Invalid output format: {output_format}")
 
     if color_mode not in ("color", "grayscale"):
         color_mode = "color"
 
+    qs = quant_step if quant_step is not None else QUANT_STEP
+
     # Apply EXIF orientation to carrier image before embedding
-    # This ensures portrait photos are embedded in their correct visual orientation
     carrier_image = _apply_exif_orientation(carrier_image)
 
-    if output_format == OUTPUT_FORMAT_JPEG and HAS_JPEGIO:
-        return _embed_jpegio(data, carrier_image, seed, color_mode, progress_file)
+    # Resize if max_dimension specified (for platform presets)
+    if max_dimension is not None:
+        img_check = Image.open(io.BytesIO(carrier_image))
+        w, h = img_check.size
+        if max(w, h) > max_dimension:
+            scale = max_dimension / max(w, h)
+            new_size = (int(w * scale), int(h * scale))
+            img_check = img_check.resize(new_size, Image.LANCZOS)
+            buf = io.BytesIO()
+            img_check.save(buf, format="PNG")
+            carrier_image = buf.getvalue()
+        img_check.close()
+
+    if output_format == OUTPUT_FORMAT_JPEG and HAS_JPEGLIB:
+        return _embed_jpeglib(data, carrier_image, seed, color_mode, progress_file)
 
     _check_scipy()
     return _embed_scipy_dct_safe(
-        data, carrier_image, seed, output_format, color_mode, progress_file
+        data, carrier_image, seed, output_format, color_mode, progress_file, quant_step=qs
     )
 
 
@@ -823,6 +852,7 @@ def _embed_scipy_dct_safe(
     output_format: str,
     color_mode: str = "color",
     progress_file: str | None = None,
+    quant_step: int = QUANT_STEP,
 ) -> tuple[bytes, DCTEmbedStats]:
     """
     Embed using scipy DCT with safe memory handling.
@@ -885,7 +915,9 @@ def _embed_scipy_dct_safe(
         gc.collect()
 
         # Embed in Y channel
-        Y_embedded = _embed_in_channel_safe(Y_padded, bits, block_order, blocks_x, progress_file)
+        Y_embedded = _embed_in_channel_safe(
+            Y_padded, bits, block_order, blocks_x, progress_file, quant_step=quant_step
+        )
         del Y_padded
         gc.collect()
 
@@ -909,7 +941,9 @@ def _embed_scipy_dct_safe(
         del image
         gc.collect()
 
-        embedded = _embed_in_channel_safe(padded, bits, block_order, blocks_x, progress_file)
+        embedded = _embed_in_channel_safe(
+            padded, bits, block_order, blocks_x, progress_file, quant_step=quant_step
+        )
         del padded
         gc.collect()
 
@@ -943,6 +977,7 @@ def _embed_in_channel_safe(
     block_order: list,
     blocks_x: int,
     progress_file: str | None = None,
+    quant_step: int = QUANT_STEP,
 ) -> np.ndarray:
     """
     Embed bits in channel using vectorized DCT operations.
@@ -1005,17 +1040,17 @@ def _embed_in_channel_safe(
                 coeffs = dct_blocks[i, embed_rows, embed_cols]
                 bit_array = np.array(block_bits)
                 # QIM embedding: round to grid, adjust for bit
-                quantized = np.round(coeffs / QUANT_STEP).astype(int)
+                quantized = np.round(coeffs / quant_step).astype(int)
                 # If quantized % 2 != bit, nudge coefficient
                 needs_adjust = (quantized % 2) != bit_array
                 # Determine direction to nudge
                 dct_blocks[i, embed_rows[needs_adjust], embed_cols[needs_adjust]] = (
                     (quantized[needs_adjust] + (1 - 2 * (quantized[needs_adjust] % 2 == 1)))
-                    * QUANT_STEP
+                    * quant_step
                 ).astype(np.float64)
                 # For bits that already match, just quantize
                 dct_blocks[i, embed_rows[~needs_adjust], embed_cols[~needs_adjust]] = (
-                    quantized[~needs_adjust] * QUANT_STEP
+                    quantized[~needs_adjust] * quant_step
                 ).astype(np.float64)
             else:
                 # Partial block - process remaining bits individually
@@ -1052,12 +1087,12 @@ def _embed_in_channel_safe(
     return result
 
 
-def _normalize_jpeg_for_jpegio(image_data: bytes) -> bytes:
+def _normalize_jpeg_for_jpeglib(image_data: bytes) -> bytes:
     """
-    Normalize a JPEG image to ensure jpegio can process it safely.
+    Normalize a JPEG image to ensure jpeglib can process it safely.
 
     JPEGs saved with quality=100 have quantization tables with all values = 1,
-    which causes jpegio to crash due to huge coefficient magnitudes.
+    which causes jpeglib to crash due to huge coefficient magnitudes.
     This function detects such images and re-saves them at a safe quality level.
 
     Args:
@@ -1078,7 +1113,7 @@ def _normalize_jpeg_for_jpegio(image_data: bytes) -> bytes:
     if hasattr(img, "quantization") and img.quantization:
         for table_id, table in img.quantization.items():
             # If all values in any table are <= threshold, normalize
-            if max(table) <= JPEGIO_MAX_QUANT_VALUE_THRESHOLD:
+            if max(table) <= JPEGLIB_MAX_QUANT_VALUE_THRESHOLD:
                 needs_normalization = True
                 break
 
@@ -1091,25 +1126,25 @@ def _normalize_jpeg_for_jpegio(image_data: bytes) -> bytes:
         img = img.convert("RGB")
 
     buffer = io.BytesIO()
-    img.save(buffer, format="JPEG", quality=JPEGIO_NORMALIZE_QUALITY, subsampling=0)
+    img.save(buffer, format="JPEG", quality=JPEGLIB_NORMALIZE_QUALITY, subsampling=0)
     img.close()
 
     return buffer.getvalue()
 
 
-def _embed_jpegio(
+def _embed_jpeglib(
     data: bytes,
     carrier_image: bytes,
     seed: bytes,
     color_mode: str = "color",
     progress_file: str | None = None,
 ) -> tuple[bytes, DCTEmbedStats]:
-    """Embed using jpegio for proper JPEG coefficient modification."""
+    """Embed using jpeglib for proper JPEG coefficient modification."""
     import os
     import tempfile
 
     # Normalize JPEG to avoid crashes with quality=100 images
-    carrier_image = _normalize_jpeg_for_jpegio(carrier_image)
+    carrier_image = _normalize_jpeg_for_jpeglib(carrier_image)
 
     img = Image.open(io.BytesIO(carrier_image))
     width, height = img.size
@@ -1122,20 +1157,20 @@ def _embed_jpegio(
         carrier_image = buffer.getvalue()
     img.close()
 
-    input_path = _jpegio_bytes_to_file(carrier_image, suffix=".jpg")
+    input_path = _jpeglib_bytes_to_file(carrier_image, suffix=".jpg")
     output_path = tempfile.mktemp(suffix=".jpg")
 
     flags = FLAG_COLOR_MODE if color_mode == "color" else 0
 
     try:
         jpeg = jpeglib.to_jpegio(jpeglib.read_dct(input_path))
-        coef_array = jpeg.coef_arrays[JPEGIO_EMBED_CHANNEL]
+        coef_array = jpeg.coef_arrays[JPEGLIB_EMBED_CHANNEL]
 
-        all_positions = _jpegio_get_usable_positions(coef_array)
-        order = _jpegio_generate_order(len(all_positions), seed)
+        all_positions = _jpeglib_get_usable_positions(coef_array)
+        order = _jpeglib_generate_order(len(all_positions), seed)
 
         # Build raw payload (header + data)
-        header = _jpegio_create_header(len(data), flags)
+        header = _jpeglib_create_header(len(data), flags)
         raw_payload = header + data
 
         # Apply Reed-Solomon error correction to entire payload if available
@@ -1402,6 +1437,7 @@ def extract_from_dct(
     stego_image: bytes,
     seed: bytes,
     progress_file: str | None = None,
+    quant_step: int | None = None,
 ) -> bytes:
     """
     Extract data from DCT stego image.
@@ -1412,6 +1448,7 @@ def extract_from_dct(
 
     Uses quick header validation to skip obviously invalid rotations.
     """
+    qs = quant_step if quant_step is not None else QUANT_STEP
     rotations_to_try = [0, 90, 180, 270]
     last_error = None
     valid_rotations = []
@@ -1429,7 +1466,7 @@ def extract_from_dct(
     # If no rotations pass quick check, try all anyway (fallback)
     if not valid_rotations:
         # Must try all rotations - quick validation might have failed due to
-        # scipy vs jpegio differences or other edge cases
+        # scipy vs jpeglib differences or other edge cases
         for rotation in rotations_to_try:
             if rotation == 0:
                 valid_rotations.append((0, stego_image))
@@ -1443,9 +1480,9 @@ def extract_from_dct(
             fmt = img.format
             img.close()
 
-            if fmt == "JPEG" and HAS_JPEGIO:
+            if fmt == "JPEG" and HAS_JPEGLIB:
                 try:
-                    result = _extract_jpegio(image_to_decode, seed, progress_file)
+                    result = _extract_jpeglib(image_to_decode, seed, progress_file)
                     if rotation != 0:
                         try:
                             from . import debug
@@ -1459,7 +1496,7 @@ def extract_from_dct(
                     continue
 
             _check_scipy()
-            result = _extract_scipy_dct_safe(image_to_decode, seed, progress_file)
+            result = _extract_scipy_dct_safe(image_to_decode, seed, progress_file, quant_step=qs)
             if rotation != 0:
                 try:
                     from . import debug
@@ -1481,6 +1518,7 @@ def _extract_scipy_dct_safe(
     stego_image: bytes,
     seed: bytes,
     progress_file: str | None = None,
+    quant_step: int = QUANT_STEP,
 ) -> bytes:
     """Extract using safe DCT operations with vectorized processing."""
     # Progress starts at 25% (decode.py writes 20% for Argon2, 25% before extraction)
@@ -1542,7 +1580,7 @@ def _extract_scipy_dct_safe(
         coeffs = dct_blocks[:, embed_rows, embed_cols]
 
         # Quantize and extract bits (vectorized)
-        quantized = np.round(coeffs / QUANT_STEP).astype(int)
+        quantized = np.round(coeffs / quant_step).astype(int)
         bits = (quantized % 2).flatten().tolist()
         all_bits.extend(bits)
 
@@ -1660,28 +1698,28 @@ def _extract_scipy_dct_safe(
     return data
 
 
-def _extract_jpegio(
+def _extract_jpeglib(
     stego_image: bytes,
     seed: bytes,
     progress_file: str | None = None,
 ) -> bytes:
-    """Extract using jpegio for JPEG images."""
+    """Extract using jpeglib for JPEG images."""
     import os
 
     # Progress starts at 25% (decode.py writes 20% for Argon2, 25% before extraction)
 
     # Normalize JPEG to avoid crashes with quality=100 images
     # (shouldn't happen with stego images, but be defensive)
-    stego_image = _normalize_jpeg_for_jpegio(stego_image)
+    stego_image = _normalize_jpeg_for_jpeglib(stego_image)
 
-    temp_path = _jpegio_bytes_to_file(stego_image, suffix=".jpg")
+    temp_path = _jpeglib_bytes_to_file(stego_image, suffix=".jpg")
 
     try:
         jpeg = jpeglib.to_jpegio(jpeglib.read_dct(temp_path))
-        coef_array = jpeg.coef_arrays[JPEGIO_EMBED_CHANNEL]
+        coef_array = jpeg.coef_arrays[JPEGLIB_EMBED_CHANNEL]
 
-        all_positions = _jpegio_get_usable_positions(coef_array)
-        order = _jpegio_generate_order(len(all_positions), seed)
+        all_positions = _jpeglib_get_usable_positions(coef_array)
+        order = _jpeglib_generate_order(len(all_positions), seed)
 
         _write_progress(progress_file, 30, 100, "extracting")
 
@@ -1751,7 +1789,7 @@ def _extract_jpegio(
                         _write_progress(progress_file, 75, 100, "decoding")
                         raw_payload = _rs_decode(rs_encoded)
                         _write_progress(progress_file, 95, 100, "decoding")
-                        _, flags, data_length = _jpegio_parse_header(raw_payload[:HEADER_SIZE])
+                        _, flags, data_length = _jpeglib_parse_header(raw_payload[:HEADER_SIZE])
                         data = raw_payload[HEADER_SIZE : HEADER_SIZE + data_length]
                         _write_progress(progress_file, 100, 100, "complete")
                         return data
@@ -1772,7 +1810,7 @@ def _extract_jpegio(
             ]
         )
 
-        _, flags, data_length = _jpegio_parse_header(header_bytes)
+        _, flags, data_length = _jpeglib_parse_header(header_bytes)
         total_bits_needed = (HEADER_SIZE + data_length) * 8
 
         all_bits = []

src/stegasoo/decode.py

@@ -54,6 +54,7 @@ def decode(
     embed_mode: str = EMBED_MODE_AUTO,
     channel_key: str | bool | None = None,
     progress_file: str | None = None,
+    platform: str | None = None,
 ) -> DecodeResult:
     """
     Decode a message or file from a stego image.
@@ -124,12 +125,21 @@ def decode(
     # Progress: key derivation done, starting extraction
     _write_progress(progress_file, 25, 100, "extracting")
 
+    # Resolve platform preset for DCT extraction
+    extract_kwargs = {}
+    if platform:
+        from .platform_presets import get_preset
+
+        preset = get_preset(platform)
+        extract_kwargs["quant_step"] = preset.quant_step
+
     # Extract encrypted data
     encrypted = extract_from_image(
         stego_image,
         pixel_key,
         embed_mode=embed_mode,
         progress_file=progress_file,
+        **extract_kwargs,
     )
 
     if not encrypted:

src/stegasoo/encode.py

@@ -51,6 +51,7 @@ def encode(
     dct_color_mode: str = "color",
     channel_key: str | bool | None = None,
     progress_file: str | None = None,
+    platform: str | None = None,
 ) -> EncodeResult:
     """
     Encode a message or file into an image.
@@ -123,6 +124,18 @@ def encode(
     # Derive pixel/coefficient selection key (with channel key)
     pixel_key = derive_pixel_key(reference_photo, passphrase, pin, rsa_key_data, channel_key)
 
+    # Resolve platform preset for DCT encoding
+    platform_kwargs = {}
+    if platform:
+        from .platform_presets import get_preset
+
+        preset = get_preset(platform)
+        platform_kwargs = {
+            "quant_step": preset.quant_step,
+            "max_dimension": preset.max_dimension,
+            "jpeg_quality": preset.jpeg_quality,
+        }
+
     # Embed in image
     stego_data, stats, extension = embed_in_image(
         encrypted,
@@ -133,6 +146,7 @@ def encode(
         dct_output_format=dct_output_format,
         dct_color_mode=dct_color_mode,
         progress_file=progress_file,
+        **platform_kwargs,
     )
 
     # Generate filename

src/stegasoo/platform_presets.py

@@ -0,0 +1,169 @@
+"""
+Platform-Calibrated DCT Presets (v4.4.0)
+
+Pre-tuned DCT embedding parameters for social media platforms. Each platform
+recompresses uploaded images differently — these presets bake in the known
+parameters so payloads survive the round-trip.
+
+Usage::
+
+    from stegasoo.platform_presets import get_preset, PLATFORMS
+
+    preset = get_preset("telegram")
+    # Use preset.quant_step, preset.jpeg_quality, etc. in DCT encode
+
+Preset parameters were derived from empirical testing. Platform compression
+behavior can change without notice — use ``pre_verify_survival()`` to confirm
+payloads survive before relying on a preset.
+"""
+
+from __future__ import annotations
+
+from dataclasses import dataclass
+
+
+@dataclass(frozen=True)
+class PlatformPreset:
+    """Tuned DCT parameters for a specific platform."""
+
+    name: str
+    jpeg_quality: int  # Platform's recompression quality
+    max_dimension: int  # Max width/height before platform resizes
+    quant_step: int  # QIM quantization step (higher = more robust)
+    embed_start: int  # Start index into EMBED_POSITIONS (skip low-freq)
+    embed_end: int  # End index into EMBED_POSITIONS (skip high-freq)
+    recompress_quality: int  # Quality to simulate platform recompression for pre-verify
+    notes: str = ""
+
+
+# Platform presets — derived from empirical testing of each platform's
+# image processing pipeline. These WILL change as platforms update.
+# Last verified: 2026-03-25
+
+PRESETS: dict[str, PlatformPreset] = {
+    "telegram": PlatformPreset(
+        name="Telegram",
+        jpeg_quality=82,
+        max_dimension=2560,
+        quant_step=35,
+        embed_start=4,
+        embed_end=16,
+        recompress_quality=80,
+        notes="~81KB max embeddable. Moderate recompression.",
+    ),
+    "discord": PlatformPreset(
+        name="Discord",
+        jpeg_quality=85,
+        max_dimension=4096,
+        quant_step=30,
+        embed_start=4,
+        embed_end=18,
+        recompress_quality=83,
+        notes="Varies with Nitro. Non-Nitro users get more aggressive compression.",
+    ),
+    "signal": PlatformPreset(
+        name="Signal",
+        jpeg_quality=80,
+        max_dimension=2048,
+        quant_step=40,
+        embed_start=5,
+        embed_end=15,
+        recompress_quality=78,
+        notes="Aggressive recompression. Use smaller payloads for reliability.",
+    ),
+    "whatsapp": PlatformPreset(
+        name="WhatsApp",
+        jpeg_quality=70,
+        max_dimension=1600,
+        quant_step=50,
+        embed_start=5,
+        embed_end=14,
+        recompress_quality=68,
+        notes="Most lossy. Capacity is significantly reduced.",
+    ),
+}
+
+PLATFORMS = sorted(PRESETS.keys())
+
+
+def get_preset(platform: str) -> PlatformPreset:
+    """Get the preset for a platform.
+
+    Args:
+        platform: Platform name (telegram, discord, signal, whatsapp).
+
+    Returns:
+        PlatformPreset with tuned DCT parameters.
+
+    Raises:
+        ValueError: If platform is not recognized.
+    """
+    key = platform.lower()
+    if key not in PRESETS:
+        available = ", ".join(PLATFORMS)
+        raise ValueError(f"Unknown platform '{platform}'. Available: {available}")
+    return PRESETS[key]
+
+
+def get_embed_positions(preset: PlatformPreset) -> list[tuple[int, int]]:
+    """Get the embed positions for a preset.
+
+    Args:
+        preset: Platform preset.
+
+    Returns:
+        List of (row, col) DCT coefficient positions.
+    """
+    from .dct_steganography import EMBED_POSITIONS
+
+    return EMBED_POSITIONS[preset.embed_start : preset.embed_end]
+
+
+def pre_verify_survival(
+    stego_image: bytes,
+    seed: bytes,
+    preset: PlatformPreset,
+) -> bool:
+    """Verify that a payload survives simulated platform recompression.
+
+    Encodes → recompresses at platform quality → attempts extraction.
+    If extraction succeeds, the payload should survive the real platform.
+
+    Args:
+        stego_image: The stego JPEG image bytes (already encoded).
+        seed: The same seed used for encoding.
+        preset: Platform preset to simulate.
+
+    Returns:
+        True if payload survived simulated recompression.
+    """
+    import io
+
+    from PIL import Image
+
+    from .dct_steganography import extract_from_dct
+
+    # Simulate platform recompression
+    img = Image.open(io.BytesIO(stego_image))
+
+    # Resize if over max dimension
+    w, h = img.size
+    if max(w, h) > preset.max_dimension:
+        scale = preset.max_dimension / max(w, h)
+        new_size = (int(w * scale), int(h * scale))
+        img = img.resize(new_size, Image.LANCZOS)
+
+    # Recompress at platform quality
+    buf = io.BytesIO()
+    if img.mode != "RGB":
+        img = img.convert("RGB")
+    img.save(buf, format="JPEG", quality=preset.recompress_quality)
+    img.close()
+    recompressed = buf.getvalue()
+
+    # Try extraction
+    try:
+        result = extract_from_dct(recompressed, seed)
+        return result is not None and len(result) > 0
+    except Exception:
+        return False

src/stegasoo/steganalysis.py

@@ -0,0 +1,281 @@
+"""
+Steganalysis Self-Check Module (v4.4.0)
+
+Statistical analysis to estimate detectability risk of stego images.
+Runs chi-square and RS (Regular-Singular) analysis on pixel data
+to assess how visible the embedding is to an attacker.
+
+Currently LSB-only. DCT steganalysis (calibration attack) deferred.
+
+Usage::
+
+    from stegasoo.steganalysis import check_image
+
+    result = check_image(image_data)
+    print(result["risk"])        # "low", "medium", or "high"
+    print(result["chi_square"])  # per-channel chi-square p-values
+    print(result["rs"])          # per-channel RS embedding estimates
+"""
+
+from __future__ import annotations
+
+import io
+from dataclasses import dataclass, field
+
+import numpy as np
+from PIL import Image
+
+from .constants import (
+    STEGANALYSIS_CHI_SUSPICIOUS_THRESHOLD,
+    STEGANALYSIS_RS_HIGH_THRESHOLD,
+    STEGANALYSIS_RS_MEDIUM_THRESHOLD,
+)
+
+
+@dataclass
+class SteganalysisResult:
+    """Result of steganalysis on an image."""
+
+    risk: str  # "low", "medium", or "high"
+    chi_square: dict = field(default_factory=dict)  # per-channel p-values
+    rs: dict = field(default_factory=dict)  # per-channel embedding estimates
+    width: int = 0
+    height: int = 0
+    channels: int = 0
+    mode: str = "lsb"
+
+
+def chi_square_analysis(channel_data: np.ndarray) -> float:
+    """Chi-square test on LSB distribution of a single channel.
+
+    Groups pixel values into pairs (2i, 2i+1) — so-called "pairs of values"
+    (PoVs). In a clean image, each pair has a natural frequency ratio.
+    LSB embedding with random data forces each pair toward equal frequency.
+
+    The test measures H0: "pairs are equalized" (consistent with embedding).
+
+    Args:
+        channel_data: Flattened 1-D array of pixel values (uint8).
+
+    Returns:
+        p-value from chi-square test.
+        HIGH p-value (close to 1.0) → pairs are equalized → suspicious.
+        LOW p-value (close to 0.0) → pairs are not equalized → less suspicious.
+    """
+    from scipy.stats import chi2
+
+    # Count occurrences of each value 0-255
+    histogram = np.bincount(channel_data.ravel(), minlength=256)
+
+    # Group into 128 pairs: (0,1), (2,3), ..., (254,255)
+    chi_sq = 0.0
+    degrees_of_freedom = 0
+
+    for i in range(0, 256, 2):
+        observed_even = histogram[i]
+        observed_odd = histogram[i + 1]
+        total = observed_even + observed_odd
+
+        if total == 0:
+            continue
+
+        expected = total / 2.0
+        chi_sq += (observed_even - expected) ** 2 / expected
+        chi_sq += (observed_odd - expected) ** 2 / expected
+        degrees_of_freedom += 1
+
+    if degrees_of_freedom == 0:
+        return 1.0  # No data to analyze
+
+    # p-value: probability of observing this chi-square value by chance
+    # Low p-value = LSBs are suspiciously uniform = likely embedded
+    p_value = 1.0 - chi2.cdf(chi_sq, degrees_of_freedom)
+    return float(p_value)
+
+
+def rs_analysis(channel_data: np.ndarray, block_size: int = 8) -> float:
+    """Regular-Singular groups analysis on a single channel.
+
+    Divides the image channel into groups of `block_size` pixels and measures
+    the "smoothness" (variation) of each group. Applying a flipping function
+    F1 (flip LSB) and F-1 (flip LSB of value-1) produces Regular (smoother)
+    and Singular (rougher) groups.
+
+    In a clean image: R_m ≈ R_{-m} and S_m ≈ S_{-m}.
+    LSB embedding causes R_m and S_{-m} to converge while S_m and R_{-m}
+    diverge, allowing estimation of the embedding rate.
+
+    Args:
+        channel_data: Flattened 1-D array of pixel values (uint8).
+        block_size: Number of pixels per group (default 8).
+
+    Returns:
+        Estimated embedding rate (0.0 = clean, 1.0 = fully embedded).
+        Values > 0.5 strongly indicate LSB embedding.
+    """
+    data = channel_data.ravel().astype(np.int16)
+    n = len(data)
+    # Trim to multiple of block_size
+    n_blocks = n // block_size
+    if n_blocks < 10:
+        return 0.0  # Not enough data
+
+    data = data[: n_blocks * block_size].reshape(n_blocks, block_size)
+
+    def variation(block: np.ndarray) -> float:
+        """Sum of absolute differences between adjacent pixels."""
+        return float(np.sum(np.abs(np.diff(block))))
+
+    def flip_positive(block: np.ndarray) -> np.ndarray:
+        """F1: flip LSB (0↔1, 2↔3, 4↔5, ...)."""
+        return block ^ 1
+
+    def flip_negative(block: np.ndarray) -> np.ndarray:
+        """F-1: flip LSB of (value - 1), i.e. -1↔0, 1↔2, 3↔4, ..."""
+        result = block.copy()
+        even_mask = (block % 2) == 0
+        result[even_mask] -= 1
+        result[~even_mask] += 1
+        return result
+
+    r_m = s_m = r_neg = s_neg = 0
+
+    for i in range(n_blocks):
+        block = data[i]
+        v_orig = variation(block)
+
+        v_f1 = variation(flip_positive(block))
+        if v_f1 > v_orig:
+            r_m += 1
+        elif v_f1 < v_orig:
+            s_m += 1
+
+        v_fn1 = variation(flip_negative(block))
+        if v_fn1 > v_orig:
+            r_neg += 1
+        elif v_fn1 < v_orig:
+            s_neg += 1
+
+    # Estimate embedding rate using the RS quadratic formula
+    # d0 = R_m - S_m, d1 = R_{-m} - S_{-m}
+    # The embedding rate p satisfies: d(p/2) = d0, d(1 - p/2) = d1
+    # Simplified estimator: p ≈ (R_m - S_m) / (R_{-m} - S_{-m}) divergence
+    d0 = r_m - s_m
+    d1 = r_neg - s_neg
+
+    if n_blocks == 0:
+        return 0.0
+
+    # Use the simplified dual-statistic estimator
+    # In clean images: d0 ≈ d1 (both positive)
+    # In embedded images: d0 → 0 while d1 stays positive
+    if d1 == 0:
+        # Can't estimate — likely very embedded or degenerate
+        return 0.5 if d0 == 0 else 0.0
+
+    # Ratio-based estimate: how much has d0 dropped relative to d1
+    ratio = d0 / d1
+    if ratio >= 1.0:
+        return 0.0  # d0 ≥ d1 means no evidence of embedding
+    if ratio <= 0.0:
+        return 1.0  # d0 collapsed or inverted
+
+    # Linear interpolation: ratio=1 → 0% embedded, ratio=0 → 100% embedded
+    estimate = 1.0 - ratio
+    return float(np.clip(estimate, 0.0, 1.0))
+
+
+def assess_risk(chi_p_values: dict[str, float], rs_estimates: dict[str, float]) -> str:
+    """Map analysis results to a risk level.
+
+    RS analysis is the primary metric (reliable for both sequential and
+    random-order embedding). Chi-square is supplementary — high p-values
+    indicate equalized PoV pairs, which is suspicious for random LSB embedding.
+
+    Args:
+        chi_p_values: Per-channel chi-square p-values (high = suspicious).
+        rs_estimates: Per-channel RS embedding rate estimates (high = suspicious).
+
+    Returns:
+        "low", "medium", or "high" detectability risk.
+    """
+    if not chi_p_values and not rs_estimates:
+        return "low"
+
+    # RS is the primary indicator: any channel with high embedding estimate
+    max_rs = max(rs_estimates.values()) if rs_estimates else 0.0
+
+    # Chi-square: high p-value means pairs are equalized (suspicious)
+    max_chi_p = max(chi_p_values.values()) if chi_p_values else 0.0
+    chi_suspicious = max_chi_p > STEGANALYSIS_CHI_SUSPICIOUS_THRESHOLD
+
+    # High risk: RS strongly indicates embedding
+    if max_rs > STEGANALYSIS_RS_HIGH_THRESHOLD:
+        return "high"
+
+    # Medium risk: moderate RS signal, or RS + chi-square both flagging
+    if max_rs > STEGANALYSIS_RS_MEDIUM_THRESHOLD:
+        return "medium"
+    if chi_suspicious and max_rs > 0.05:
+        return "medium"
+
+    return "low"
+
+
+def check_image(image_data: bytes, mode: str = "lsb") -> dict:
+    """Run steganalysis on an image and return detectability assessment.
+
+    Args:
+        image_data: Raw image bytes (PNG, BMP, etc.).
+        mode: Analysis mode — currently only "lsb" is supported.
+
+    Returns:
+        Dict with keys: risk, chi_square, rs, width, height, channels, mode.
+    """
+    if mode not in ("lsb", "auto"):
+        raise ValueError(f"Unsupported steganalysis mode: {mode}. Use 'lsb' or 'auto'.")
+
+    img = Image.open(io.BytesIO(image_data))
+    if img.mode not in ("RGB", "RGBA", "L"):
+        img = img.convert("RGB")
+
+    width, height = img.size
+    pixels = np.array(img)
+    img.close()
+
+    channel_names = ["R", "G", "B"] if pixels.ndim == 3 else ["L"]
+    if pixels.ndim == 2:
+        pixels = pixels[:, :, np.newaxis]
+
+    num_channels = min(pixels.shape[2], 3)  # Skip alpha
+
+    chi_p_values = {}
+    rs_estimates = {}
+
+    for i in range(num_channels):
+        name = channel_names[i]
+        channel = pixels[:, :, i].ravel()
+        chi_p_values[name] = chi_square_analysis(channel)
+        rs_estimates[name] = rs_analysis(channel)
+
+    risk = assess_risk(chi_p_values, rs_estimates)
+
+    result = SteganalysisResult(
+        risk=risk,
+        chi_square=chi_p_values,
+        rs=rs_estimates,
+        width=width,
+        height=height,
+        channels=num_channels,
+        mode=mode,
+    )
+
+    return {
+        "risk": result.risk,
+        "chi_square": result.chi_square,
+        "rs": result.rs,
+        "width": result.width,
+        "height": result.height,
+        "channels": result.channels,
+        "mode": result.mode,
+    }

src/stegasoo/steganography.py

@@ -107,13 +107,14 @@ EXT_TO_FORMAT = {
 # - v3.1.0: 76 bytes (had date field - 10+1 bytes)
 # - v3.2.0: 65 bytes (removed date, simpler)
 # - v4.0.0: 66 bytes (added flags byte for channel key)
+# - v4.4.0: 82 bytes (added 16-byte message nonce for HKDF)
 
-HEADER_OVERHEAD = 66  # What the crypto layer adds to any message
+HEADER_OVERHEAD = 82  # What the crypto layer adds to any message (v6 format)
 LENGTH_PREFIX = 4  # We prepend the payload length for LSB extraction
-ENCRYPTION_OVERHEAD = HEADER_OVERHEAD + LENGTH_PREFIX  # Total: 70 bytes
+ENCRYPTION_OVERHEAD = HEADER_OVERHEAD + LENGTH_PREFIX  # Total: 86 bytes
 
-# That 70 bytes is your minimum image capacity requirement.
-# A tiny 100x100 image gives you ~3750 bytes capacity, minus 70 = ~3680 usable.
+# That 86 bytes is your minimum image capacity requirement.
+# A tiny 100x100 image gives you ~3750 bytes capacity, minus 86 = ~3664 usable.
 
 # DCT output format options (v3.0.1)
 DCT_OUTPUT_PNG = "png"
@@ -609,6 +610,9 @@ def embed_in_image(
     dct_output_format: str = DCT_OUTPUT_PNG,
     dct_color_mode: str = "color",
     progress_file: str | None = None,
+    quant_step: int | None = None,
+    jpeg_quality: int | None = None,
+    max_dimension: int | None = None,
 ) -> tuple[bytes, Union[EmbedStats, "DCTEmbedStats"], str]:
     """
     Embed data into an image using specified mode.
@@ -636,49 +640,54 @@ def embed_in_image(
         embed_mode in VALID_EMBED_MODES, f"Invalid embed_mode: {embed_mode}. Use 'lsb' or 'dct'"
     )
 
-    # DCT MODE
-    if embed_mode == EMBED_MODE_DCT:
-        if not has_dct_support():
-            raise ImportError(
-                "scipy is required for DCT embedding mode. " "Install with: pip install scipy"
-            )
+    # Dispatch via backend registry
+    from .backends import registry
 
-        # Validate DCT output format
+    backend = registry.get(embed_mode)
+    if not backend.is_available():
+        raise ImportError(
+            f"Dependencies for '{embed_mode}' mode are not installed. "
+            f"Install with: pip install stegasoo[dct]"
+        )
+
+    if embed_mode == EMBED_MODE_DCT:
+        # Validate DCT-specific options
         if dct_output_format not in (DCT_OUTPUT_PNG, DCT_OUTPUT_JPEG):
             debug.print(f"Invalid dct_output_format '{dct_output_format}', defaulting to PNG")
             dct_output_format = DCT_OUTPUT_PNG
-
-        # Validate DCT color mode (v3.0.1)
         if dct_color_mode not in ("grayscale", "color"):
             debug.print(f"Invalid dct_color_mode '{dct_color_mode}', defaulting to color")
             dct_color_mode = "color"
 
-        dct_mod = _get_dct_module()
-
-        # Pass output_format and color_mode to DCT module (v3.0.1)
-        stego_bytes, dct_stats = dct_mod.embed_in_dct(
+        stego_bytes, dct_stats = backend.embed(
             data,
             image_data,
             pixel_key,
-            output_format=dct_output_format,
-            color_mode=dct_color_mode,
             progress_file=progress_file,
+            dct_output_format=dct_output_format,
+            dct_color_mode=dct_color_mode,
+            quant_step=quant_step,
+            jpeg_quality=jpeg_quality,
+            max_dimension=max_dimension,
         )
-
-        # Determine extension based on output format
-        if dct_output_format == DCT_OUTPUT_JPEG:
-            ext = "jpg"
-        else:
-            ext = "png"
-
+        ext = "jpg" if dct_output_format == DCT_OUTPUT_JPEG else "png"
         debug.print(
             f"DCT embedding complete: {dct_output_format.upper()} output, "
             f"color_mode={dct_color_mode}, ext={ext}"
         )
         return stego_bytes, dct_stats, ext
 
-    # LSB MODE
-    return _embed_lsb(data, image_data, pixel_key, bits_per_channel, output_format, progress_file)
+    # LSB and other image backends
+    stego_bytes, stats = backend.embed(
+        data,
+        image_data,
+        pixel_key,
+        progress_file=progress_file,
+        bits_per_channel=bits_per_channel,
+        output_format=output_format,
+    )
+    ext = getattr(stats, "output_extension", "png")
+    return stego_bytes, stats, ext
 
 
 def _embed_lsb(
@@ -844,6 +853,7 @@ def extract_from_image(
     bits_per_channel: int = 1,
     embed_mode: str = EMBED_MODE_AUTO,
     progress_file: str | None = None,
+    quant_step: int | None = None,
 ) -> bytes | None:
     """
     Extract hidden data from a stego image.
@@ -860,32 +870,40 @@ def extract_from_image(
     """
     debug.print(f"extract_from_image: mode={embed_mode}")
 
-    # AUTO MODE: Try LSB first, then DCT
+    from .backends import registry
+
+    # AUTO MODE: Try LSB first (cheaper), then other backends
     if embed_mode == EMBED_MODE_AUTO:
-        result = _extract_lsb(image_data, pixel_key, bits_per_channel)
-        if result is not None:
-            debug.print("Auto-detect: LSB extraction succeeded")
-            return result
-
-        if has_dct_support():
-            debug.print("Auto-detect: LSB failed, trying DCT")
-            result = _extract_dct(image_data, pixel_key, progress_file)
+        auto_order = [EMBED_MODE_LSB] + [
+            m for m in registry.available_modes(carrier_type="image") if m != EMBED_MODE_LSB
+        ]
+        for mode in auto_order:
+            backend = registry.get(mode)
+            debug.print(f"Auto-detect: trying {mode}")
+            result = backend.extract(
+                image_data,
+                pixel_key,
+                progress_file=progress_file,
+                bits_per_channel=bits_per_channel,
+                quant_step=quant_step,
+            )
             if result is not None:
-                debug.print("Auto-detect: DCT extraction succeeded")
+                debug.print(f"Auto-detect: {mode} extraction succeeded")
                 return result
-
         debug.print("Auto-detect: All modes failed")
         return None
 
-    # EXPLICIT DCT MODE
-    elif embed_mode == EMBED_MODE_DCT:
-        if not has_dct_support():
-            raise ImportError("scipy required for DCT mode")
-        return _extract_dct(image_data, pixel_key, progress_file)
-
-    # EXPLICIT LSB MODE
-    else:
-        return _extract_lsb(image_data, pixel_key, bits_per_channel)
+    # EXPLICIT MODE
+    backend = registry.get(embed_mode)
+    if not backend.is_available():
+        raise ImportError(f"Dependencies for '{embed_mode}' mode are not installed.")
+    return backend.extract(
+        image_data,
+        pixel_key,
+        progress_file=progress_file,
+        bits_per_channel=bits_per_channel,
+        quant_step=quant_step,
+    )
 
 
 def _extract_dct(
@@ -1099,9 +1117,9 @@ def peek_image(image_data: bytes) -> dict:
 
     # Try DCT extraction (requires scipy/jpeglib)
     try:
-        from .dct_steganography import HAS_JPEGIO, HAS_SCIPY
+        from .dct_steganography import HAS_JPEGLIB, HAS_SCIPY
 
-        if HAS_SCIPY or HAS_JPEGIO:
+        if HAS_SCIPY or HAS_JPEGLIB:
             from .dct_steganography import extract_from_dct
 
             # Extract first few bytes to check header