Camera presets

getframes ships a library of camera configurations so you can start simulating without hunting down datasheet numbers.

Listing presets

from getframes import available_presets
from getframes.presets import preset_info

available_presets()
# ['andor_ikon_m934', 'andor_ixon_ultra_888', 'generic_ccd', ...]

preset_info()
# [{'preset': 'andor_ikon_m934', 'name': 'Andor iKon-M 934',
#   'manufacturer': 'Andor', 'model': 'iKon-M 934', 'sensor_type': 'CCD'}, ...]

Loading a preset

import getframes as gf

cam = gf.Camera.from_preset("andor_ixon_ultra_888")

# Or get the raw config to tweak it:
cfg = gf.load_preset("andor_ixon_ultra_888")
warmer = cfg.replace(em_gain=100.0)
cam = gf.Camera(warmer)

Bundled cameras

Preset	Sensor	Description
andor_ikon_m934	CCD	Deep-cooled (−80 °C) back-illuminated scientific CCD
andor_ixon_ultra_888	EMCCD	Single-photon-sensitive EMCCD
leonardo_saphira	EAPD	HgCdTe avalanche IR array (AO wavefront sensing)
first_light_imaging_cred_one	EAPD	C-RED One global-reset raw reads, with 32-channel structure
zwo_asi2600mm	CMOS	Sony IMX571 cooled CMOS
hamamatsu_orca_fusion	sCMOS	Back-thinned sCMOS with per-pixel read noise
hamamatsu_orca_quest_2	sCMOS	qCMOS low-noise camera, digitized full QE curve
nuvu_hnu_240	EMCCD	CCD220 deep-depletion EMCCD, high-gain AO mode
nuvu_hnu_128_omega	EMCCD	128 x 128 high-speed midband EMCCD, Omega mode
andor_ocam2k	EMCCD	CCD220 AO EMCCD, 2000 fps high-gain mode
andor_cb1_0_5mp	sCMOS	IMX426 global-shutter 0.5 MP CB1
andor_marana_4_2b_11	sCMOS	11 µm Marana, extended-dynamic-range mode
photometrics_prime_95b	sCMOS	11 µm Prime 95B, combined-gain mode
princeton_instruments_kuro_1200b	sCMOS	11 µm KURO 1200B
qhy530_pro_ii	CMOS	Global-shutter Sony IMX530 camera
scimeasure_little_joe_ccd39	CCD	Keck/SciMeasure Little Joe with CCD39-01 QE curve
tucsen_aries_6504_pro	sCMOS	Single-photon-level sensitive mode
generic_ccd	CCD	Idealised CCD for teaching/testing
generic_cmos	CMOS	Idealised uncooled CMOS
generic_emccd	EMCCD	Idealised EMCCD
generic_eapd	EAPD	Idealised eAPD (avalanche gain, low excess noise)
generic_scmos	sCMOS	Idealised sCMOS (per-pixel read noise, nonlinearity)

Verify before quantitative use

Preset values are representative of published specifications but are not a substitute for characterising your own hardware. Treat them as realistic starting points.

leonardo_saphira describes a representative bare SAPHIRA detector. Camera-body electronics are deliberately generic. first_light_imaging_cred_one is the camera-specific full-frame profile: its raw pedestal, interleaved 32-output pattern, edge structure, per-pixel/channel read-noise distribution, and temporal common mode are fitted to 84.6 K CRED1 dark cubes. Repeated capped ramps also constrain conversion gain, physical avalanche multiplication, response nonuniformity, and the combined cap-plus-camera rate. The physical format, QE, full well, excess-noise factor, and intrinsic-background limit are constrained by FLI/Oxford specifications and published characterization.

A warm cap is not a zero-flux dark in the near infrared. Model its emission as an incident background passed to Camera.nondestructive_series; do not add it to the preset's intrinsic dark_current_e_per_s.

Adding your own preset

Presets are plain TOML files in src/getframes/presets/data/. To add a camera, drop in a <slug>.toml file whose keys mirror CameraConfig:

name = "My Camera"
manufacturer = "Acme"
model = "CAM-9000"
sensor_type = "CMOS"
resolution = [2048, 2048]
pixel_size_um = 5.0
quantum_efficiency = 0.85
full_well_e = 20000.0
bit_depth = 12
gain_e_per_adu = 1.0
bias_offset_adu = 250.0
read_noise_e = 2.0
dark_current_e_per_s = 0.3
dark_current_ref_temp_c = 20.0
dark_current_doubling_temp_c = 6.0
notes = "Where these numbers came from."

# Optional wavelength-resolved QE for spectral simulations.
[qe_curve]
wavelength_nm = [400.0, 500.0, 600.0, 700.0, 800.0]
qe = [0.45, 0.72, 0.88, 0.81, 0.55]

The loader discovers the file automatically — no code changes required. If you are working from a clone, the preset test suite will validate it loads correctly. When a manufacturer only publishes a graph, record in notes that the curve was digitized and whether the ordinate is bare QE or QE x fill factor. Keep filters, windows, atmosphere, and relay optics as separate throughput curves.

Binning is a first-class part of the config. Set supported_binnings (the integer factors the sensor supports, always including 1) and binning_method ("digital" for post-read software summation, where binned read noise grows as the factor, or "on_chip" for pre-read charge-domain/hardware binning, where one read noise applies per super-pixel). Camera.expose(binning=…, binning_mode=…) then produces the correct binned frame — you never store a per-binning read-noise value.

For facts that are useful to a particular instrument trade but do not belong in the generic detector-noise model—such as a camera body's mechanical envelope or source URLs—use an [extra] TOML table. These are exposed as CameraConfig.extra; a consumer should treat an absent value as unknown, not as a pass/fail result.

When a camera has more than one read mode (distinct read-noise operating points, e.g. a standard and an ultra-quiet mode), carry the alternates in an [[extra.read_modes]] array. Each entry gives the mode name, its read_noise_e, and any mode-specific dark_current_e_per_s, temperature_c, or narrower supported_binnings; the binning factors and method otherwise come from the config. This replaces the old per-binning detector_modes list — binned read noise now comes from the model, not from a stored table.