pixelflux 1.2.0

A performant web native pixel delivery pipeline for diverse sources, blending VNC-inspired parallel processing of pixel buffers with flexible modern encoding formats.

pixelflux

A performant web native pixel delivery pipeline for diverse sources, blending VNC-inspired parallel processing of pixel buffers with flexible modern encoding formats.

This module provides a Python interface to a high-performance C++ capture library. It captures pixel data from a source (currently X11 screen regions), detects changes, and encodes modified stripes into JPEG or H.264, delivering them via a callback mechanism. This stripe-based, change-driven approach is designed for efficient streaming or processing of visual data.

Installation

This module relies on a native C++ component.

1. Prerequisites (for the current X11 backend on Debian/Ubuntu): Ensure you have CMake, a C++ compiler, and development files for X11, XShm (Xext), libjpeg-turbo, and libx264.

sudo apt-get update && \
sudo apt-get install -y \
  cmake \
  g++ \
  gcc \
  libjpeg-turbo8-dev \
  libx11-dev \
  libxfixes-dev \
  libxext-dev \
  libx264-dev \
  make \
  python3-dev \
  python3-pip \
  python3-websockets

Note: libjpeg-turbo8-dev might be libjpeg62-turbo-dev or similar on older systems.

2. Install via pip:

pip install pixelflux

This command will:

• Build the native screen_capture_module.so (or similar) library using CMake during the installation process.
• Install the Python wrapper and the compiled shared library.

Note: The current backend is designed and tested for Linux/X11 environments. Future development aims to support a wider range of framebuffers and platforms.

3. Developer Install From Source

sudo python3 setup.py install

Usage

Basic Capture

Here's a basic example demonstrating how to use the pixelflux module to start capturing and process encoded stripes.

import ctypes
import time
# These components are provided by the pixelflux Python wrapper:
from pixelflux import CaptureSettings, ScreenCapture, StripeCallback, StripeEncodeResult

# Integer constants for OutputMode (as defined in C++ enum class OutputMode)
OUTPUT_MODE_JPEG = 0
OUTPUT_MODE_H264 = 1

# Integer constants for StripeDataType (as defined in C++ enum class StripeDataType)
STRIPE_TYPE_UNKNOWN = 0
STRIPE_TYPE_JPEG = 1
STRIPE_TYPE_H264 = 2

# Define your Python callback function.
# The StripeCallback type (a ctypes.CFUNCTYPE) expects this signature.
def my_python_callback(result_ptr: ctypes.POINTER(StripeEncodeResult), user_data_ptr: ctypes.c_void_p):
    """Callback function to process encoded stripes."""
    result = result_ptr.contents
    
    # Note: Based on the current Python wrapper's start_capture method (inferred from previous TypeError),
    # user_data_ptr might not be actively passed from Python. If user_data is needed,
    # the Python wrapper for start_capture would need to support passing it.
    # For this example, user_data_ptr is effectively ignored.

    if result.data and result.size > 0:
        type_str = "Unknown"
        if result.type == STRIPE_TYPE_JPEG:
            type_str = "JPEG"
        elif result.type == STRIPE_TYPE_H264:
            type_str = "H264"

        print(f"Received {type_str} stripe: frame_id={result.frame_id}, y_start={result.stripe_y_start}, height={result.stripe_height}, size={result.size} bytes")
    
    # Memory for result.data is managed by the C++ layer after this callback returns.
    # Do NOT free result.data here.

# Configure capture settings
capture_settings = CaptureSettings()
capture_settings.capture_width = 1280
capture_settings.capture_height = 720
capture_settings.capture_x = 0
capture_settings.capture_y = 0
capture_settings.target_fps = 30.0

# Set output mode to H.264
capture_settings.output_mode = OUTPUT_MODE_H264
capture_settings.h264_crf = 25 # H264 Constant Rate Factor (0-51, lower is better quality)

# Instantiate the ScreenCapture module
module = ScreenCapture()

# Create a C-callable function pointer from your Python callback.
# StripeCallback is the CFUNCTYPE provided by the pixelflux module.
c_callback_func_ptr = StripeCallback(my_python_callback)

try:
    # Call start_capture with settings and the callback pointer.
    module.start_capture(capture_settings, c_callback_func_ptr) 
    
    mode_str = "JPEG" if capture_settings.output_mode == OUTPUT_MODE_JPEG else "H264"
    print(f"Capture started (Mode: {mode_str}). Press Enter to stop...")
    input() # Keep capture running
finally:
    module.stop_capture()
    print("Capture stopped.")

Capture Settings

The CaptureSettings class, exposed as a ctypes.Structure by the Python wrapper, allows configuration of various parameters:

# Python representation of the C++ CaptureSettings struct,
# provided by the pixelflux Python wrapper.
class CaptureSettings(ctypes.Structure):
    _fields_ = [
        ("capture_width", ctypes.c_int),
        ("capture_height", ctypes.c_int),
        ("capture_x", ctypes.c_int),
        ("capture_y", ctypes.c_int),
        ("target_fps", ctypes.c_double),
        ("jpeg_quality", ctypes.c_int),             # (JPEG mode) Quality for changed stripes (0-100)
        ("paint_over_jpeg_quality", ctypes.c_int),  # (JPEG mode) Quality for static "paint-over" stripes (0-100)
        ("use_paint_over_quality", ctypes.c_bool),  # (JPEG/H264 mode) Enable paint-over with different quality
        ("paint_over_trigger_frames", ctypes.c_int),# Frames of no motion to trigger paint-over/IDR request
        ("damage_block_threshold", ctypes.c_int),   # Consecutive changes to trigger "damaged" state for a stripe
        ("damage_block_duration", ctypes.c_int),    # Frames a stripe stays "damaged" (affects paint-over logic)
        ("output_mode", ctypes.c_int),              # 0 for JPEG, 1 for H264
        ("h264_crf", ctypes.c_int),                 # (H264 mode) CRF value (0-51, lower is better quality/higher bitrate)
        ("h264_fullcolor", ctypes.c_bool),          # Enable H.264 full color (I444)
        ("h264_fullframe", ctypes.c_bool),          # Enable H.264 full frame encoding (activates NVENC if available)
        ("capture_cursor", ctypes.c_bool),          # Enable cursor capture
        ("watermark_path", ctypes.c_char_p),        # Absolute path to watermark PNG file
        ("watermark_location_enum", ctypes.c_int),  # 0-6 for values table below 
    ]
WATERMARK_LOCATION_NONE = 0
WATERMARK_LOCATION_TL = 1 # Top Left
WATERMARK_LOCATION_TR = 2 # Top Right
WATERMARK_LOCATION_BL = 3 # Bottom Left
WATERMARK_LOCATION_BR = 4 # Bottom Right
WATERMARK_LOCATION_MI = 5 # Middle
WATERMARK_LOCATION_AN = 6 # Animated bounces around

Adjust these settings to fine-tune capture performance and quality.

Stripe Callback and Data Structure

The start_capture function requires a callback function, invoked by the native C++ module when an encoded stripe is ready. The Python wrapper provides the necessary StripeCallback type (a ctypes.CFUNCTYPE) and the StripeEncodeResult structure (a ctypes.Structure).

# The pixelflux Python wrapper provides these definitions:

# StripeCallback is a ctypes.CFUNCTYPE defining the callback signature:
# StripeCallback = ctypes.CFUNCTYPE(
#    None, ctypes.POINTER(StripeEncodeResult), ctypes.c_void_p # Result pointer, User data pointer
# )

# StripeEncodeResult is a ctypes.Structure for the callback data:
class StripeEncodeResult(ctypes.Structure):
    _fields_ = [
        ("type", ctypes.c_int),             # StripeDataType: 0 UNKNOWN, 1 JPEG, 2 H264
        ("stripe_y_start", ctypes.c_int),
        ("stripe_height", ctypes.c_int),
        ("size", ctypes.c_int),
        ("data", ctypes.POINTER(ctypes.c_ubyte)), # Pointer to the encoded data (includes custom header)
        ("frame_id", ctypes.c_int)              # Frame counter for this stripe
    ]

Your Python callback function must match the StripeCallback signature: def my_callback(result_ptr: ctypes.POINTER(StripeEncodeResult), user_data_ptr: ctypes.c_void_p):

• result_ptr: A ctypes pointer to a StripeEncodeResult structure. Access its fields via result_ptr.contents.
• user_data_ptr: A ctypes.c_void_p representing the user data you passed to start_capture. If you passed None, this will be None or 0. You are responsible for casting and interpreting this pointer if you use it.

Inside the callback, process result_ptr.contents. The Python wrapper handles freeing the memory of result.data by calling the native free_stripe_encode_result_data function after your callback returns. Do not attempt to free it manually.

Features

• Efficient Pixel Capture: Leverages a native C++ module using XShm for optimized X11 screen capture performance.
• Stripe-Based Encoding (JPEG & H.264): Encodes captured frames into horizontal stripes, processed in parallel using a number of threads based on system core count. Each stripe is an independent data unit.
• Change Detection: Encodes only stripes that have changed (based on XXH3 hash comparison) since the last frame, significantly reducing processing load and bandwidth. This approach is inspired by VNC.
• Hardware-Accelerated H.264 Encoding (NVENC):
  • Automatically utilizes NVIDIA's NVENC for H.264 encoding when h264_fullframe is enabled, significantly reducing CPU usage.
  • This feature has no build-time dependencies. It uses dynamic loading at runtime to detect and use installed NVIDIA proprietary drivers. If drivers are not found, it seamlessly falls back to CPU-based x264 encoding.
• Configurable Capture Region: Specify the exact X, Y, width, and height of the screen region to capture.
• Adjustable FPS, Quality, and Encoding Parameters: Control frame rate, JPEG quality (0-100), and H.264 CRF (0-51).
• Dynamic Quality Optimizations:
  • Paint-Over for Static Regions: After a stripe remains static for paint_over_trigger_frames, it is resent. For JPEG, this uses paint_over_jpeg_quality if use_paint_over_quality is true. For H.264, this triggers a request for an IDR frame for that stripe, ensuring a full refresh.
  • Adaptive Behavior for Highly Active Stripes (Damage Throttling):
    • Identifies stripes that change very frequently (exceeding damage_block_threshold updates).
    • For these "damaged" stripes, damage checks are done less frequently, saving resources on high motion.
    • For JPEG output, the quality of these frequently changing stripes dynamically adjusts (reducing slightly on change) and resets to higher base/paint-over quality after a cooldown period of damage_block_duration frames. This manages resources effectively for volatile content.
• Direct Callback Mechanism: Provides encoded stripe data, including a custom header, directly to your Python code for real-time processing or streaming.

Example: Real-time H.264 Streaming with WebSockets

A comprehensive example, screen_to_browser.py, is located in the examples directory of this repository. This script demonstrates real-time screen capture, H.264 encoding, and streaming via WebSockets. It sets up:

• A WebSocket server to stream encoded H.264 stripes.
• An HTTP server to serve a client-side HTML page for viewing the stream.
• The pixelflux module to perform the screen capture and encoding.

To run this example:

Note: This example assumes you are on a Linux host with a running X11 session and will only work from localhost unless https is added.

1. Navigate to the examples directory within the repository:

   cd examples
   

2. Execute the Python script:

   python3 screen_to_browser.py
   

3. Open your web browser and go to the URL indicated by the script's output http://localhost:9001 to view the live stream.

License

This project is licensed under the Mozilla Public License Version 2.0. A copy of the MPL 2.0 can be found at https://mozilla.org/MPL/2.0/.