Skip to content

camsim

Deterministic camera sensor and ISP simulation with image-quality metrics.

camsim is a professional Python library that simulates a camera from spectral scene irradiance to a rendered image and measures image quality with standard metrics. It is built for camera and ISP engineers, and researchers who need a reproducible, physically grounded testbed. Rendered images and IQ reports can feed downstream computer-vision or ML experiments.

ISP stage contact sheet

The default ISP applied to a spectral ColorChecker scene, shown stage by stage: RAW, defect correction, black-level correction, lens shading, white balance, demosaicing, colour correction, denoise, sharpen, and tone.

What's inside

  1. Simulation pipeline — spectral scene and optics helpers, a rigorous sensor model (irradiance → RAW), and a modular ISP (RAW → rendered image). The reusable core is the sensor + ISP pair; neither depends on the scene or optics packages.
  2. Image-quality metrics — an independent harness for e-SFR/MTF, SNR, colour error, and acutance. It operates on arrays plus metadata and never imports the simulator.
  3. Reproducible tooling — camera orchestration, RAW/image IO, CLI commands, and validated presets that make every result repeatable.

Install

With uv:

uv add camsim

Or with pip:

pip install camsim

For development:

git clone https://github.com/pengzhao-ai/camsim.git
cd camsim
uv sync --all-extras

pip/venv fallback:

python -m venv .venv
source .venv/bin/activate
python -m pip install --upgrade pip
pip install -e ".[all]"

Quick check

import camsim

print(camsim.__version__)

camsim info

End-to-end example

import camsim
from camsim.config.io import PRESETS_DIR, load_isp, load_sensor
from camsim.metrics import measure_all
from camsim.scene.charts import slanted_edge_layout
from camsim.scene.targets import render

scene = render("slanted_edge", lux=200.0, shape=(256, 256), oversampling=2, seed=11)
sensor = camsim.Sensor(
    load_sensor(PRESETS_DIR / "sensors" / "flagship_phone_50mp.json"), seed=1
)
raw = sensor.expose(scene.irradiance, exposure_s=0.01, mode="LCG", seed=7)
image = camsim.ISP(load_isp(PRESETS_DIR / "isp" / "default.json")).process(raw)
report = measure_all(image, slanted_edge_layout(shape=(256, 256)))

print(image.data.shape, image.color_space)
print(report.mtf.mtf50_cp, report.acutance)

Where to go next

  • Concepts — the imaging chain, units, linear-space invariant, and determinism contract.
  • Architecture — package layout, data contracts, and dependency rules.
  • ISP stages — every stage's role, parameters, trace, and linear-space invariant.
  • Metrics — chart layouts, e-SFR, SNR, colour, acutance, and the unified report.
  • Validation — validation evidence, measured results, coverage policy, and reproduction commands.
  • Datasheet mapping — how preset fields trace to public specifications.
  • API reference — the public modules and types.
  • Release process — versioning and publishing policy.