Apple ProRes Log captures 10-bit HDR video with minimal processing, making it ideal for photogrammetry, radiance fields, and VFX. This package uses ffmpeg with Apple's AppleLogToLin-v1.0.cube LUT to decode log-encoded video to linear images. Apple log video is currently supported on iPhones 15, 16. This is the best way I know to get as close to unprocessed RAW video as possible on iPhone without making your own capture app like MotionCam on Android. The Apple ProRes Log white paper is attached here. For reference, video editing software like DaVinci Resolve provides similar log color grading functionality in their GUI.

How It Works

The script uses an ffmpeg filter graph that:

1. Converts color matrix: zscale transforms from BT.2020 limited range to full range
2. Applies the LUT: lut1d uses Apple's official LUT to decode from log to linear
3. Converts color space: zscale transforms from linear BT.2020 to sRGB (BT.709) for display-referred outputs

All three output formats (32-bit EXR, 16-bit PNG, 8-bit PNG) are generated in a single ffmpeg pass.

Requirements

• ffmpeg (better to install system level) conda install -c conda-forge ffmpeg -y
• Python >= 3.7

Installation

From the repository:

git clone https://github.com/btilmon/apple-log2linear.git
cd apple-log2linear
pip install .

Or install directly from GitHub:

pip install git+https://github.com/btilmon/apple-log2linear.git

For development:

pip install -e .

Usage

Command Line

apple-log2linear --video_path /path/to/video.MOV

With frame skipping (keep every 10th frame):

apple-log2linear --video_path /path/to/video.MOV --step 10

Python API

from appleLog2Linear import process_apple_log_video

# Process video - outputs saved directly to output_dir (or video's directory if not specified)
result = process_apple_log_video(
    video_path="/path/to/video.MOV",
    step=10,                # keep every 10th frame
    output_32bit=True,      # 32-bit EXR linear output
    output_16bit=False,     # disable 16-bit PNG
    output_8bit=True,       # 8-bit sRGB PNG for COLMAP
    output_dir="/path/to/output",  # optional: where to save outputs
)

# Access results
print(result['images_32bit_dir'])  # /path/to/output/images-32bit-linear
print(result['images_8bit_dir'])   # /path/to/output/images
print(result['exr_count'])         # Number of EXR frames
print(result['png8_count'])        # Number of 8-bit PNG frames

Returning Images Directly

Use return_imgs=True to get images loaded as numpy arrays:

from appleLog2Linear import process_apple_log_video

# Process and return images as numpy arrays
images = process_apple_log_video(
    video_path="/path/to/video.MOV",
    step=10,
    output_32bit=True,
    output_16bit=False,
    output_8bit=True,
    output_dir="/path/to/output",
    return_imgs=True,       # Return images as dict
)

# Access images by output directory name
# Keys are: 'images-32bit-linear', 'images-16bit-srgb', 'images'
linear_images = images['images-32bit-linear']  # List of 32-bit numpy arrays
srgb_images = images['images']                 # List of 8-bit numpy arrays

# Iterate over frames
for i, img in enumerate(srgb_images):
    print(f"Frame {i}: shape={img.shape}, dtype={img.dtype}")

Options

CLI Arguments

Argument	Description	Default
--video_path	Full path to the .MOV file	Required
--step	Keep every Nth frame	1 (all frames)
--lut_path	Path to LUT file	Bundled LUT
--output_dir	Directory for output folders	Video's directory
--no-32bit	Disable 32-bit EXR output	False
--no-16bit	Disable 16-bit PNG output	False
--no-8bit	Disable 8-bit PNG output	False

Python API Only

Argument	Description	Default
return_imgs	Return images as dict instead of file paths	False

Output

Frames are saved to output directory (defaults to video's directory):

Folder	Format	Description
images-32bit-linear/	32-bit EXR	Linear, unclipped (maximum fidelity)
images-16bit-srgb/	16-bit PNG	sRGB, display-referred
images/	8-bit PNG	sRGB, display-referred (for COLMAP, etc.)

Enabling ProRes Log on iPhone

1. Go to Settings → Camera → Formats
2. Enable Apple ProRes
3. Select "Log" from ProRes Encoding
4. Verify "ProRes LOG" appears in the camera app
5. I recommend enabling the setting to lock white balance on record, or using BlackMagic camera app for finer grained control of camera settings.

Note: At 4K 30FPS, expect ~6GB per minute of video.