saturnix-filter 0.2.0

High-performance camera film simulation filters

saturnix-filter

High-performance camera film simulation filters in Rust, designed for the SATURNIX open-source camera (Raspberry Pi Zero 2W).

Developed to accelerate on-device photo processing, saturnix-filter delivers a ~41x to ~91x speedup over traditional pure-Python image processing on the target Raspberry Pi Zero 2 W hardware by leveraging zero-copy in-memory pixel manipulation, 10-bit fixed-point integer math, and multicore scaling via Rayon.

Features

• Zero-Copy Memory Model: Manipulates Pillow image bytes directly in memory space
• Parallel Execution: Distributes row-by-row pixel computations across all available CPU cores using rayon.
• 26 Complete Film Styles:
  • S-Gold (Kodak Gold warm vintage style)
  • S-Vivid (Kodak Ektar ultra-saturated style)
  • S-Natural (Fujifilm organic greens style)
  • S-Saturnix (Signature cel-animation glow style)
  • S-MonoX (Kodak Tri-X 400 panchromatic S-curve B&W style)
  • S-Portra (Kodak Portra 400 soft, warm skin-tone style)
  • S-Cinestill (Cinestill 800T cool tungsten / teal-shadow style)
  • S-Cross (Cross-processed E-6-in-C-41 high-contrast style)
  • S-Faded (Sun-faded vintage print with milky lifted blacks)
  • S-Bleach (Bleach-bypass desaturated high-contrast silver style)
  • S-Sepia (Warm brown-toned B&W)
  • S-Cyano (Cool blue Cyanotype-toned B&W)
  • S-Noir (High-contrast neutral B&W with heavy vignette)
  • S-Teal (Cinematic teal-and-orange style)
  • S-Lomo (Lomography toy-camera oversaturated style)
  • S-Fuji (Fujifilm Velvia high-saturation landscape style)
  • S-Selenium (Cool selenium-toned B&W)
  • S-Platinum (Warm, soft platinum-print-toned B&W)
  • S-Infrared (Aerochrome-style false-colour infrared)
  • S-SplitTone (Cinematic warm-highlight / cool-shadow split-toning)
  • S-Kodachrome (Rich, warm vintage-slide style)
  • S-Polaroid (Instant-film look with milky blacks and cyan cast)
  • S-Matrix (Digital-dystopia green cast)
  • S-Cine (Filmic S-curve digital-cinema grade)
  • S-Leak (Warm light-leak flare from the corner)
  • S-Halation (Warm red-orange glow blooming from highlights)
  • VHS (Vintage VHS tape simulation)

Examples

Each column is one source photo; each row is the same image with a filter applied.

	Sample 01	Sample 02	Sample 03	Sample 04
Original
S-Gold
S-Vivid
S-Natural
S-Saturnix
S-MonoX
S-Portra
S-Cinestill
S-Cross
S-Faded
S-Bleach
S-Sepia
S-Cyano
S-Noir
S-Teal
S-Lomo
S-Fuji
S-Selenium
S-Platinum
S-Infrared
S-SplitTone
S-Kodachrome
S-Polaroid
S-Matrix
S-Cine
S-Leak
S-Halation
VHS

Installation

Install the precompiled binary wheel directly from PyPI (no compiler required on the Raspberry Pi!):

pip install saturnix-filter

The core install has no Python dependencies — ideal for the production runtime on the Raspberry Pi. To also get the saturnix-filter command-line tool for experimenting on a PC (image file conversion, gallery rendering), install the optional cli extra:

pip install "saturnix-filter[cli]"

Usage

from PIL import Image
import saturnix_filter

# 1. Load an image as RGB
img = Image.open("photo.jpg").convert("RGB")
width, height = img.size

# 2. Extract mutable bytearray (Zero-Copy pointer reference)
buf = bytearray(img.tobytes())

# 3. Apply the filter instantly inside RAM
# Supported options: "S-Gold", "S-Vivid", "S-Natural", "S-Saturnix", "S-MonoX",
#                    "S-Portra", "S-Cinestill", "S-Cross", "S-Faded", "S-Bleach",
#                    "S-Sepia", "S-Cyano", "S-Noir", "S-Teal", "S-Lomo", "S-Fuji",
#                    "S-Selenium", "S-Platinum", "S-Infrared", "S-SplitTone",
#                    "S-Kodachrome", "S-Polaroid", "S-Matrix", "S-Cine",
#                    "S-Leak", "S-Halation", "VHS"
saturnix_filter.apply_film_inplace(buf, width, height, "S-Saturnix")

# 4. Re-construct the Pillow image from the modified buffer
filtered_img = Image.frombytes("RGB", (width, height), bytes(buf))
filtered_img.save("photo_filtered.jpg", "JPEG", quality=92)

Command-line tool

With the cli extra installed, the saturnix-filter command converts image files directly:

# Apply a filter (auto-names the output photo_s-gold.jpg next to the input)
saturnix-filter convert photo.jpg -f S-Gold

# Several filters and/or images at once, downscaled to 1200 px wide
saturnix-filter convert *.jpg -f S-Gold -f S-Halation --max-width 1200

# List every available filter with a description
saturnix-filter list

# Apply every filter to one image, rendering a gallery next to it
saturnix-filter gallery photo.jpg

Performance Comparison

All numbers are pure in-memory processing times at full 16 MP camera resolution (4656 x 3496 pixels), measured with a warm-up pass and excluding file I/O. saturnix-filter uses the optimized 10-bit fixed-point integer implementation.

Hardware: Raspberry Pi Zero 2 W (ARM Cortex-A53 quad-core @ 1.0 GHz)

Filter	Original Python	saturnix-filter	Speedup
S-Gold	16.729 s	0.375 s	~44.6x
S-Vivid	16.711 s	0.405 s	~41.3x
S-Natural	16.702 s	0.374 s	~44.7x
S-Saturnix	32.083 s	0.392 s	~81.9x
S-MonoX	31.841 s	0.350 s	~91.0x
VHS	18.452 s	0.351 s	~52.5x

License

This project is licensed under the MIT License - see the LICENSE file for details.