Osmo Action 6

This reference preserves an Osmo Action 6 loaner survey from 21 September 2026, using camera firmware V01.02.0521 and DJI Mimo 2.12.0 (1000061) on an iPhone 16 Pro Max. It documents physical Mimo behavior, command/reply exchanges, camera readback, and inspected recordings. The survey alone does not establish Action 6 support in the OpenPocketCine apps.

OpenPocketCine app status

Both apps now implement the following from this survey. None of it has been checked on an Action 6 in OpenPocketCine yet; the loaner was returned before the implementation.

Area	App behavior	Evidence
Live view	One 09/A8 enable to receiver 41; no Nano 02/09 gate and no Pocket 02/68; AVC with the Nano private SEI; square or other feed geometry from the decoded picture	Connection
Colour	Normal 10-bit 3F and D-Log M 3D; D-Log M binds the official Action 6 cube	Colour
Aperture	APERTURE tile in the FOCUS position: live iris from cam_expo_param @13, strategy SET 02/8E pid 0044 from camcap_aperture_ctrl_strategy	Aperture
Capture	Photo 02/E1 05; Timelapse start/stop 02/01; other video modes 02/02	Commands
Album favorite	Action-specific 02/BF layout, On/Off in byte 1	Coverage
Hidden	Gimbal, tap focus and focus modes	No AF control in the base-lens survey

Not implemented: zoom (02/B8 uses a different encoding), the Auto aperture range (pid 004D), Portrait Mode 4B, lapse and Hyperlapse parameters, Multiview, and first-time pairing approval.

The loaner’s firmware predates several features in current DJI documentation. Use the firmware reference to distinguish advertised features from the behavior captured on this unit. No firmware update, factory reset, storage format, or deletion of pre-existing media was part of the survey.

Find a command or capability

Reference	Contents
Command comparison	Shared/Pocket differences, qualified command families, parameter index and untested families
Capture coverage and media commands	Every observed command family, media-list framing, reply/error patterns and decoder limits
Bluetooth and pairing	Advertising model ID, GATT, MTU, packed pairing request, credential reads, cached reconnect, fresh-client limitation
Connection and live view	UDP startup, command cursors, live-enable receiver, status codes, subscriptions, AVC/SEI validation
Video, aperture and exposure	Video format writes, aperture strategies and ranges, iris telemetry, manual exposure, WB, image adjustments
Shooting modes	SuperNight, Photo, Slow Motion, Timelapse, Hyperlapse, Portrait Mode, loop recording and custom presets
Orientation and device settings	Remaining 4:3/square/portrait Video formats, explicit reconnect, timecode-menu blocking state, storage, Wi-Fi choices and general settings
Stabilization and advanced controls	Horizon modes, rate restrictions, zoom, film tones, exposure limits and additional control evidence
Original media	Complete-file hashes, dimensions, rational rates, codec/bit depth, audio and metadata; RAW preservation status
Firmware and published specifications	DJI source links, release differences, advertised matrices, lens accessories and features outside the tested scope

Shared framing and reusable commands remain in the command catalog. Each page above gives the Action-specific addresses, values, restrictions and evidence needed to apply it.

Evidence levels

Level	What it establishes
UI	Mimo displayed the option or selected value
Request	The capture contains the corresponding outgoing command
Accepted reply	A matched camera response accepted that request
Status	An independent camera push/readback reports the resulting value
Physical effect	A visible or recorded effect was observed
Inspected original	A complete file was preserved and its actual contents checked

A pending UI selection, a capability entry, and an accepted SET are different evidence. Tables identify missing readbacks and distinguish advertised choices from selections actually exercised. Different resolutions, frame rates, colors, exposure states, stabilization modes and zoom levels change the capability lists. Honor each list’s logical count; stale bytes can remain beyond it.

Implementation priorities

1. Qualify Action 6 explicitly in model discovery and connection routing. Its captured live-enable receiver and AVC framing are described in the connection reference; shared behavior alone does not prove the existing Nano or Pocket startup sequence works unchanged.
2. Keep configured aperture strategy/range separate from instantaneous iris, configured exposure compensation separate from metered exposure, and logical capability counts separate from backing payload length.
3. Preserve square and portrait geometry, fractional shutter representations, and mode-dependent color/ISO/EIS/zoom restrictions. Inspect the sample files rather than inferring recording properties from the live feed or generic metadata labels.
4. Gate unverified controls. No AF-S/AF-C/MF selector or remote focus-distance control was established in this base-lens Mimo survey. Focus peaking is a display aid, and the optional Macro Lens’s physical ring is separate evidence.
5. Prove future app support on the physical camera for both shells. Captured Mimo commands are reference evidence, not a replacement for that validation.

Scope and reproducibility

Continuous iPhone network and Bluetooth traces, timestamped action labels, screenshots, offline decoders, complete media and hash manifests are retained privately under the session’s ignored capture directory. An independently verified copy is kept outside the task worktree. Raw captures, credentials, device identifiers, private interiors and original media are not published.

The reference uses UTC action times and identifies capture takes and packet numbers where needed. Network timestamps are batched by iPhone RVI; they must not be treated as latency measurements. The Bluetooth page describes its separate clock correction. File-size rotation can end a take before the usual five-minute interval; use packet timestamps rather than assuming fixed boundaries.

Fresh-client camera approval, station-mode provisioning, cold wake, optional lenses, camera-body-only settings and direct USB output require evidence beyond the successful Mimo reconnects and base-lens survey. Each reference records its remaining limits explicitly.

Archive and restored state

The continuous network capture ran from 08:48:40 to 12:12:22 UTC and closed as 42 readable PCAPNG files containing 3,015,110 packets. The capture tool reported zero host drops. This does not rule out upstream RVI loss: incomplete HTTP reconstruction was observed, and complete originals were validated separately. The closed Bluetooth log contains 1,591,008 complete records with no trailing partial record. These are capture-integrity counts, not counts of distinct commands or proof that every camera workflow was exercised.

A later timecode-menu follow-up adds one closed network file with 40,389 packets, preserving Mimo’s blocked state and subsequent return to normal preview. Its counts are separate from the continuous survey above; reset/sync operations remain unqualified.

The retained media includes 16 complete originals, 402,022,877 bytes: ten MP4s and three JPEG+RAW pairs. The media reference gives exact hashes, actual dimensions, rational frame rates and full RAW validation.

Before disconnecting, camera readback confirmed restoration to Video, Custom 4K1:1 at 25 fps, Normal 10bit, Auto aperture f/2–f/4, Auto white balance and exposure, EV 0, Wide, RockSteady, Daily, 1× zoom, neutral Texture/Noise Reduction, loop Off and Voice Off. Pro On was confirmed in Mimo. The tested photo’s favorite flag was restored to Off and independently re-listed. This restores the observed starting video setup; it does not promise restoration of every inactive per-mode memory modified during the survey.