Dynamic range split, above and below middle gray
at ISO 800 / 1600 / 3200

Figures from the Blackmagic PYXIS manual dynamic range chart. The camera switches to the high base (ISO 3200) at ISO 1250, which is why ISO 1600 shows less headroom than ISO 800: it sits low on the second gain circuit.

Single native ISO with a gain structure that trades shadow for highlight as ISO rises. Total stays at 16 stops across 125–3200 in the 12K/9K readout modes.

Figures from the official URSA Mini Pro manual dynamic range chart, the first-generation 12K RGBW sensor, and the direct ancestor of the PYXIS 12K's. The pattern is identical (one stop from shadows to highlights per ISO doubling, constant total), just with roughly two stops less overall range than its large-format successor.

Blackmagic's long-running Super 35 workhorse, and the camera the 12K replaced at the top of the URSA line. Like the PYXIS 6K it runs a dual-base sensor with circuits at 400 and 3200, which is why Blackmagic quotes its 15 stops specifically at ISO 3200, so ISO 800 sits on the low base while 3200 lands on the clean high base. The G2 added redesigned electronics and up to 300 fps, but measured range is essentially unchanged from the G1.

RED does not publish an above/below-middle-gray table. In REDCODE RAW, ISO is metadata (FLUT): the sensor capture is identical at any ISO, and each doubling of the ISO dial re-maps middle gray one stop lower into the capture, gaining one stop of highlight protection and giving up one stop of shadow separation. The table shows that shift relative to ISO 800.

The second-generation KOMODO: updated sensor, image processing and audio hardware, a big jump in frame rates (4K 120p now possible) and, per RED, improved dynamic range. The ISO behavior is unchanged from the original: RED publishes no native ISO, and in REDCODE RAW the ISO dial only moves where code values sit on the waveform without affecting what clips, so the split below is the redistribution relative to ISO 800.

Canon specifies C-Log2 dynamic range as 1600% for ISO 800 and above, which places the clip point ≈ +6.5 stops over middle gray. Unlike the RED, ISO on the C70 is real gain: headroom above middle gray stays fixed from base ISO upward, while the rising noise floor eats into the shadow side. The DGO (dual gain output) advantage is strongest at ISO 800 and diminishes as ISO climbs.

The R5C has a dual base ISO sensor (800 and 3200, for both Cinema RAW Light and C-Log3) but no C-Log2 and no DGO. Canon's C-Log3 spec of 1600% at ISO 800 and above pins the clip point at ≈ +6.5 stops over middle gray, the same headroom as the C70. The difference between the two cameras is almost entirely below middle gray, where the C70's dual gain output holds cleaner shadows. ISO 1600 is the R5C's awkward spot: it's a gain-up from the first base, so shadows suffer until the second base engages at 3200.

The C80 uses the C400's back-illuminated stacked full frame sensor: no DGO, but a triple base ISO of 800 / 3200 / 12800 in C-Log2, C-Log3 and RAW. Canon's C-Log2 spec of 1600% pins the clip point at ≈ +6.5 stops over middle gray at every base, so all the ISO behavior plays out below the mid point: in auto mode the camera rides base 800 up to ISO 2500, so ISO 1600 is a gain-up from the first base, and ISO 3200 lands exactly on the clean second base.

Same sensor and base ISO structure as the C80. The C400 adds the pro body, 6K 60p RAW, internal ND and 12G-SDI, but the exposure math is identical: C-Log2's 1600% spec fixes the clip point at ≈ +6.5 stops over middle gray at every base, ISO 1600 is a gain-up from base 800, and ISO 3200 lands on the clean second circuit.

Canon's compact hybrid cinema body keeps the family C-Log2 ceiling, 1600%, so ≈ +6.5 stops over middle gray at both bases, but places its second base at ISO 6400 rather than 3200. That makes it behave more like the Sonys at these three ISOs: 1600 and 3200 both run on gain from base 800, with the clean second circuit sitting one click above 3200. Unusually, Canon rates the Super 35 crop higher than full frame (16 vs 15+ stops), and lab tests confirm the crop mode really does measure better.

Essentially the C50's sensor in a mirrorless body: same 32.5MP full frame chip, same dual base ISO of 800 and 6400, same C-Log2 1600% ceiling at ≈ +6.5 stops over middle gray, and 7K open gate internal RAW. Reviewers describe it as a cross between the C50 and the R5 Mark II: you gain IBIS, stills performance and a smaller body, and give up some thermal headroom and about half a stop of dynamic range. Like the C50, both 1600 and 3200 run on gain from base 800.

Canon's first V-series full frame body, announced May 2026, the same 32.5MP front-side-illuminated sensor as the R6 Mark III and the C50, in a video-first package: no EVF, no mechanical shutter, but an internal cooling fan, vertical tripod mount and UI, tally lamp, front record button, power zoom lever and full-size HDMI. Because the sensor and log curve are shared, the exposure math is the C50's: ≈ +6.5 stops of C-Log2 headroom, dual base 800 / 6400, and both 1600 and 3200 running on gain from the first base.

The closest Nikon gets to the C80 on a spec sheet, and the comparison is genuinely interesting because RED engineers built it. Nikon acquired RED in 2024, and RED reworked the Z6 III's semi-stacked sensor for dual base ISOs of 800 and 6400 and a claimed 15+ stops, recordable in RED's own R3D NE with the identical color space and log curve as RED's cinema bodies. On the spec line it reads like a C80 rival at a fraction of the price. In the lab it doesn't get there, but its rolling shutter does.

In S-Log3 the FX3's clip point sits ≈ +6 stops over middle gray (the curve places middle gray at 41 IRE and runs out of code values about six stops up), leaving ≈ 9 stops below at base to reach the claimed 15+. The catch for this comparison: the FX3's second base is way up at ISO 12800, so both 1600 and 3200 are gain-up settings from base 800, each doubling costs roughly a stop of shadow floor with no clean base to rescue them.

The FX6 shares the FX3's sensor generation, log curve and base ISO layout, so the distribution table is the same: ≈ +6 stops of S-Log3 headroom, ≈ 9 below at base 800, and both 1600 and 3200 stranded on gain because the second base doesn't arrive until 12800. What the FX6 adds is workflow, not stops, electronic variable ND and a proper Cine EI mode, which changes how you reach these numbers in practice.

Sony's newest Cinema Line body (July 2026) keeps the same ≈ +6-stop S-Log3 highlight ceiling as the FX3 but fixes its mid-ISO gap: the triple base of 800 / 4000 / 12800 puts a clean circuit right where the FX3 has none, and a separate Dual Gain mode combines readouts for a claimed 16+ stops at the cost of readout speed. At the three ISOs compared here the FX5 still rides base 800 (the second base starts at 4000), so on paper it mirrors the FX3, its advantages appear just outside this table.

A gimbal and camera in one body, with 9 stops of built-in ND. D-Log places its clip point around +6 stops over middle gray, and the dual native ISO sits at 800 / 5000, which puts both 1600 and 3200 on gain from the first base, like the FX3. The lab results are the good news: at ISO 800 in ProRes RAW HQ, 12.4 stops at SNR = 2 and 13.7 at SNR = 1, with the second base only about 0.4 stops behind. Unusually, RAW measures better than the compressed ProRes here, DJI deliberately leaves noise in the ProRes files rather than scrubbing it, to preserve texture for post.

The strongest full-frame Lumix for BRAW work. Blackmagic enabled external Blackmagic RAW from the S1II and S1IIE in Video Assist 3.20 (June 2025). V-Log puts its clip point around +6 stops over middle gray, and the native ISOs are 640 and 5000, so ISO 800 is effectively base performance while 1600 and 3200 run on gain. The headline feature is Dual Gain "DR Boost", which merges two ISO readouts into one frame, worth two extra stops in APS-C crop, at a heavy rolling-shutter cost.

The S5IIX ships with BRAW compatibility out of the box; the S5II gains it via firmware v2.0 plus the DMW-SFU2 upgrade key. Output is 12-bit over HDMI at 5.9K full frame, 4.1K APS-C and 3.5K anamorphic. Same V-Log ceiling of ≈ +6 stops, with dual native ISOs at 640 and 4000, so again ISO 800 is essentially base, and 3200 sits just under the second base.

The original BRAW-capable Lumix group. The S1H got Blackmagic RAW output in firmware v2.3 (March 2021), with the S1 (v2.1, requires the DMW-SFU2 key) and S5 (v2.3) following, and the BS1H box camera shipping with it. All four output 12-bit BRAW to a Video Assist 12G at 5.9K full frame, 4K Super 35 and 3.5K anamorphic, and all share V-Log's ≈ +6-stop ceiling with native ISOs of 640 and 4000.

The odd one out, and worth including precisely because it breaks the pattern. The fp L has no log picture profile at all: you shoot 12-bit CinemaDNG and develop it in Resolve using BMD Film as the color space, which is why its numbers don't map cleanly onto the log-curve math used elsewhere here. Its base ISOs are 100 and 1250, so unlike every other camera on this page, ISO 800 is not near a base: it's a gain-up from 100, while 1600 and 3200 sit above the second base. Treat this table as indicative only.

The reference point the rest of this page gets measured against. ARRI's EI system works like RED's metadata ISO: the sensor's 17 stops are constant at every EI, and the dial only chooses where middle gray sits. Each doubling hands one stop from shadows to highlights. What's unmatched is the ceiling: at EI 3200 the ALEXA 35 holds more stops above middle gray (11.4) than several cameras on this page have in total.

Readout time is a property of the sensor mode, not the frame rate: a camera scans just as fast at 24 fps as at 60, so these are the values in effect for standard 24 fps cinema work in each listed mode. The shorter the scan relative to the 41.7 ms frame window, the less skew on pans and the safer the camera is around strobes and LED walls. Green = fast (<10 ms), amber = moderate, red = slow (>18 ms).