During tuning of GK7205V300 with IMX335, the bin file generated by PQTool cannot be imported.

[deedy-wang]

Before you start

• I searched the open and closed issues for this.
• This is not about the picture, sound, ONVIF or WebUI behaviour itself - that is majestic, and there is a link for it on the previous page.
• I read the wiki page for my SoC.

SoC vendor

Goke

SoC model

gk7205v300

Camera model

imx335

Output of sysinfo

root@gk7205v300-imx335:~# sysinfo

Firmware:
  OPENIPC_VERSION=2.6.09.20

Majestic:
  internal - Lite HiSilicon (gk7205v200), master+7f2dc18, 2026-09-20 17:29
  external - Lite HiSilicon (gk7205v200), master+7f2dc18, 2026-09-20 17:29

USB devices:
  1d6b:0002 1d6b:0003

Memory:
  totalmem - 128M
  osmem    - 32M

Processor:
  detected - gk7205v300 2224180800ce0213
  recorded - gk7205v300

Sensor:
  detected - imx335
  recorded - imx335

Output of cat /etc/os-release

root@gk7205v300-imx335:~# cat /etc/os-release
NAME=Buildroot
VERSION=-g6f03cc4
ID=buildroot
VERSION_ID=2024.02.10
PRETTY_NAME="Buildroot 2024.02.10"
OPENIPC_VERSION=2.6.09.20
GITHUB_VERSION="master+6f03cc4, 2026-09-20"
BUILD_OPTION=lite
BUILD_ID=nightly-20260920-6f03cc4
BUILD_SHA=6f03cc45a1165ad45ab60cb81984b674c14837b0
BUILD_PLATFORM=gk7205v300_lite
TIME_STAMP=1789926944

Last build where this worked

No response

Where the image came from

The openipc.org download page

What happens

The following attempts have been carried out:

1. Modified parameters such as AE speed in imx335_cmos.c within the OpenIPC source code. The attempt was unsuccessful.
2. Tried loading the PQTool-generated bin file under /etc/sensor/iq. Manual loading succeeds, yet parsing fails.
3. Developed a parameter configuration tool based on OpenIPC source code to set AE speed. After writing the parameters, readback values are always 0.
4. Modified parameters in imx335_cmos.h in OpenIPC source code. This method works, but the .h file only contains a limited set of parameters and cannot meet tuning requirements.

What you expected instead

According to the debugging method described in image-quality-tuning.md on the OpenIPC wiki, how can I import the bin file exported by PQTool into the firmware image?

logread | tail -200

root@gk7205v300-imx335:~# logread | tail -200
Sep 20 17:55:50 gk7205v300-imx335 authpriv.info dropbear[799]: Running in background
Sep 20 17:55:50 gk7205v300-imx335 cron.info crond[801]: crond (busybox 1.36.1) started, log level 8
Sep 20 17:55:50 gk7205v300-imx335 kern.warn kernel: open_hwrng: loading out-of-tree module taints kernel.
Sep 20 17:55:50 gk7205v300-imx335 kern.info kernel: hisi-hwrng: TRNG at 0x10080000+0x204 registered (pump)
Sep 20 17:55:50 gk7205v300-imx335 daemon.info goke: SENSOR: imx335
Sep 20 17:55:50 gk7205v300-imx335 kern.err kernel: ==== online_flag=0, cmos_yuv_flag=0, sensor=imx335, chip=gk7205v300, board=demo====
Sep 20 17:55:50 gk7205v300-imx335 kern.err kernel: ==== g_quick_start_flag=0 ====
Sep 20 17:55:50 gk7205v300-imx335 kern.err kernel: sysconfig init success!
Sep 20 17:55:50 gk7205v300-imx335 kern.err kernel: Module media: init ok
Sep 20 17:55:50 gk7205v300-imx335 kern.info kernel: Media Memory Zone Manager
Sep 20 17:55:50 gk7205v300-imx335 kern.err kernel: osal 1.0 init success!
Sep 20 17:55:50 gk7205v300-imx335 kern.err kernel: load sys.ko OK!
Sep 20 17:55:50 gk7205v300-imx335 kern.err kernel: load region.ko OK!
Sep 20 17:55:50 gk7205v300-imx335 kern.err kernel: load vgs.ko OK!
Sep 20 17:55:51 gk7205v300-imx335 kern.err kernel: load vi.ko OK !
Sep 20 17:55:51 gk7205v300-imx335 kern.info kernel: openipc_frame_ts: /dev/openipc-frame-ts minor=57, event types=2, ring_depth=256
Sep 20 17:55:51 gk7205v300-imx335 kern.err kernel: ISP Mod init!
Sep 20 17:55:51 gk7205v300-imx335 kern.err kernel: load vpss.ko OK!
Sep 20 17:55:51 gk7205v300-imx335 kern.err kernel: load chnl.ko OK!
Sep 20 17:55:51 gk7205v300-imx335 kern.err kernel: load vedu.ko OK!
Sep 20 17:55:51 gk7205v300-imx335 kern.err kernel: load rc.ko OK!
Sep 20 17:55:51 gk7205v300-imx335 kern.err kernel: load venc.ko OK!
Sep 20 17:55:51 gk7205v300-imx335 kern.err kernel: load h264e.ko OK!
Sep 20 17:55:51 gk7205v300-imx335 kern.err kernel: load h265e.ko OK!
Sep 20 17:55:51 gk7205v300-imx335 kern.err kernel: load jpege.ko OK!
Sep 20 17:55:51 gk7205v300-imx335 kern.err kernel: load ive.ko OK!
Sep 20 17:55:51 gk7205v300-imx335 kern.err kernel: Load sensor_spi.ko OK !
Sep 20 17:55:51 gk7205v300-imx335 kern.err kernel: load mipi_rx driver successful!
Sep 20 17:55:51 gk7205v300-imx335 kern.err kernel: register dev
Sep 20 17:55:51 gk7205v300-imx335 kern.err kernel: Watchdog Timer: 0.01 initialized. default_margin=60 sec (nodeamon= 0)
Sep 20 17:55:51 gk7205v300-imx335 kern.err kernel: wtdg init ok. ver=Sep 20 2026, 17:55:22.
Sep 20 17:55:51 gk7205v300-imx335 kern.err kernel: Load hi_cipher.ko success.
Sep 20 17:55:51 gk7205v300-imx335 user.notice hw: ADC raw VAL: 91
Sep 20 17:55:51 gk7205v300-imx335 user.notice hw: HW ver: v1.0 (CHN1=91)
Sep 20 17:55:53 gk7205v300-imx335 user.info majestic[978]: Majestic Lite for HiSilicon (gk7205v200), version master+7f2dc18, built on 2026-09-20 17:29
Sep 20 17:55:53 gk7205v300-imx335 user.info majestic[978]: outgoing.naluSize has moved to rtsp.naluSize; carrying 1200 over
Sep 20 17:55:53 gk7205v300-imx335 user.info majestic[978]: Loading main configuration: /etc/majestic.yaml
Sep 20 17:55:53 gk7205v300-imx335 user.debug majestic[978]: Detected Watchdog
Sep 20 17:55:53 gk7205v300-imx335 user.info majestic[978]: Watchdog armed: asked 300 s, accepted 300 s, fed every 150.000s, magic close supported
Sep 20 17:55:53 gk7205v300-imx335 user.info majestic[978]: SRTP: AES-CTR on the Cipher engine, HMAC-SHA1 in software
Sep 20 17:55:53 gk7205v300-imx335 user.debug majestic[978]: Cipher engine: key ladder slot 1 is blank (wraps as AES-256 under a zero key)
Sep 20 17:55:53 gk7205v300-imx335 user.debug majestic[978]: Cipher engine: key ladder slot 2 is blank (wraps as AES-256 under a zero key)
Sep 20 17:55:53 gk7205v300-imx335 user.debug majestic[978]: Cipher engine: key ladder slot 3 is blank (wraps as AES-256 under a zero key)
Sep 20 17:55:53 gk7205v300-imx335 user.info majestic[978]: Cipher engine: key ladder verified through slot 1 (blank)
Sep 20 17:55:53 gk7205v300-imx335 user.info majestic[978]: no identity: the identity slot (OTP key slot 1) is not programmed; majestic --otp-burn is the operator's way to do that
Sep 20 17:55:53 gk7205v300-imx335 user.debug majestic[978]: OSD face stem is 0.0312 em (8 px at a 256 px probe)
Sep 20 17:55:53 gk7205v300-imx335 user.debug majestic[978]: Set server path: /var/www
Sep 20 17:55:53 gk7205v300-imx335 user.info majestic[978]: HTTP server started on :::80
Sep 20 17:55:53 gk7205v300-imx335 user.warn majestic[978]: HTTPS port :::443 configured without cert/key; skipping TLS server
Sep 20 17:55:53 gk7205v300-imx335 user.debug majestic[978]: sensor and MIPI receiver unparked
Sep 20 17:55:53 gk7205v300-imx335 user.debug majestic[978]: SENSOR=imx335
Sep 20 17:55:53 gk7205v300-imx335 user.debug majestic[978]: matched sensor config: imx335_i2c_4M.ini
Sep 20 17:55:53 gk7205v300-imx335 user.info majestic[978]: Using /etc/sensors/imx335_i2c_4M.ini sensor configuration file
Sep 20 17:55:53 gk7205v300-imx335 user.debug majestic[978]: App was built with MPP version: Gk7205V200_MPP_V1.0.1.0.B020 Release
Sep 20 17:55:53 gk7205v300-imx335 user.debug majestic[978]: Current MPP version: VERSION=MPP_V1.0.1.0 B020 Release
Sep 20 17:55:53 gk7205v300-imx335 user.debug majestic[978]: SoC chip id 0x72050300
Sep 20 17:55:53 gk7205v300-imx335 user.warn majestic[978]: Set custom sensor clock mode: 37.125MHz
Sep 20 17:55:53 gk7205v300-imx335 user.info majestic[978]: VB sizing: 4 blocks for the pipeline (2 of them held by the VI path), default 5, isp.blkCnt 5 in effect
Sep 20 17:55:53 gk7205v300-imx335 user.debug majestic[978]: sensor stSnsImx335Obj
Sep 20 17:55:53 gk7205v300-imx335 user.debug majestic[978]:   input_mode MIPI, WDR NONE
Sep 20 17:55:53 gk7205v300-imx335 user.debug majestic[978]:   dev [2592x1520]@200x20 25fps, RGGB
Sep 20 17:55:53 gk7205v300-imx335 user.debug majestic[978]: trying to load /usr/lib/sensors/libsns_imx335.so
Sep 20 17:55:53 gk7205v300-imx335 user.debug majestic[978]: Free MMZ mem before allocation: 98012KB
Sep 20 17:55:53 gk7205v300-imx335 user.debug majestic[978]: VB configuration:
Sep 20 17:55:53 gk7205v300-imx335 user.debug majestic[978]:   [0]: 5909760 x 4
Sep 20 17:55:53 gk7205v300-imx335 user.debug majestic[978]:   [1]: 5909760 x 1
Sep 20 17:55:53 gk7205v300-imx335 user.debug majestic[978]: Free MMZ mem after allocation: 98012KB
Sep 20 17:55:53 gk7205v300-imx335 user.info majestic[978]: SET_DEV_ATTR: devno=0 input_mode=0 data_rate=0 img_rect={0,0,2592,1520} data_type=1 wdr=0 lanes=[0,1,2,3]
Sep 20 17:55:53 gk7205v300-imx335 user.debug majestic[978]: Setting VI-VPSS mode: pipe0=1 (vi_online=0 vpss_online=1)
Sep 20 17:55:53 gk7205v300-imx335 user.debug majestic[978]: Sensor driver loaded
Sep 20 17:55:53 gk7205v300-imx335 user.debug majestic[978]: SDK is in 'VI_OFFLINE_VPSS_ONLINE' mode
Sep 20 17:55:53 gk7205v300-imx335 user.debug majestic[978]: IntType=2 EarlyLine=128 LowDelay=0 LineCnt=128
Sep 20 17:55:53 gk7205v300-imx335 user.info majestic[978]: linear mode
Sep 20 17:55:53 gk7205v300-imx335 user.info majestic[978]: binit false IMX335 i2c init
Sep 20 17:55:53 gk7205v300-imx335 user.info majestic[978]: -------Sony  IMX335__2592_1520_10bit_52fps Initial OK!-------
Sep 20 17:55:53 gk7205v300-imx335 user.debug majestic[978]: Exposure: 125000us (8fps), aGain: 32x, dGain: 1x, ispGain: 4x, mode: slow_shutter
Sep 20 17:55:53 gk7205v300-imx335 user.debug majestic[978]: BE AF statistics enabled: 17x15 zones over 2592x1520, YUV domain
Sep 20 17:55:53 gk7205v300-imx335 user.info majestic[978]: Loading IQ profile /etc/sensors/iq/default.ini
Sep 20 17:55:54 gk7205v300-imx335 user.info majestic[978]: Loading INI profile /etc/sensors/iq/default.ini
Sep 20 17:55:54 gk7205v300-imx335 user.debug majestic[978]: SoC reports 3 VENC channels
Sep 20 17:55:54 gk7205v300-imx335 user.debug majestic[978]: h264 1920x1080@20fps 512kbit 20gop
Sep 20 17:55:54 gk7205v300-imx335 user.debug majestic[978]:   venc_chn: 0, vpss_grp: 0, vpss_chn: 1
Sep 20 17:55:54 gk7205v300-imx335 user.info majestic[978]:   vpss chn 1 compression: asked none, kept none
Sep 20 17:55:54 gk7205v300-imx335 user.debug majestic[978]:   minQp: 28, maxQp: 42
Sep 20 17:55:54 gk7205v300-imx335 daemon.info : starting pid 1008, tty '/dev/console': '/sbin/getty -L console 0 vt100'
Sep 20 17:55:54 gk7205v300-imx335 user.debug majestic[978]:   venc took 4564KB of MMZ [1012]
Sep 20 17:55:54 gk7205v300-imx335 user.debug majestic[978]: HiSilicon SDK started
Sep 20 17:55:54 gk7205v300-imx335 user.debug majestic[978]: Free MMZ mem finally: 58828KB
Sep 20 17:55:54 gk7205v300-imx335 user.warn majestic[978]: onvif discovery: multicast listener not ready, retrying
Sep 20 17:55:54 gk7205v300-imx335 user.info majestic[978]: also answering for gk7205v300-imx335.local.
Sep 20 17:55:54 gk7205v300-imx335 user.warn majestic[978]: mdns: multicast listener not ready, retrying
Sep 20 17:55:54 gk7205v300-imx335 user.info majestic[978]: RTSP server started on port 554
Sep 20 17:55:54 gk7205v300-imx335 user.debug majestic[978]: DTLS identity: P-256, 293-byte certificate
Sep 20 17:55:54 gk7205v300-imx335 user.info majestic[978]: WebRTC fingerprint sha-256 DE:40:C5:6E:E8:72:EA:70:72:42:51:BB:66:59:71:D6:62:76:BD:60:5C:4E:FC:EF:A8:29:74:0C:FA:22:5D:0A
Sep 20 17:55:54 gk7205v300-imx335 user.info majestic[978]: Live backlog budget: 28345 KiB of 124172 KiB RAM (headroom bound, memory available 70864 KiB); 1280 KiB per session, so about 22 concurrent live viewers across /ws/video, WebRTC, RTSP-over-TCP, /mjpeg and /video.mp4. Low
Sep 20 17:55:54 gk7205v300-imx335 user.info majestic[978]: H.264 SPS said nothing about picture reordering; publishing it with max_num_reorder_frames=0 (25 -> 29 bytes) so decoders that would otherwise hold a level's whole DPB show the first frame at once
Sep 20 17:55:59 gk7205v300-imx335 user.debug majestic[978]: MJPEG 1920x1080@5fps 50q
Sep 20 17:55:59 gk7205v300-imx335 user.debug majestic[978]:   venc_chn: 1, vpss_grp: 0, vpss_chn: 2
Sep 20 17:55:59 gk7205v300-imx335 user.info majestic[978]:   vpss chn 2 compression: asked none, kept none
Sep 20 17:55:59 gk7205v300-imx335 kern.warn kernel: alloc_contig_range: [44900, 44b00) PFNs busy
Sep 20 17:55:59 gk7205v300-imx335 user.debug majestic[978]:   mpeg_enc took 2048KB of MMZ [2040]
Sep 20 17:55:59 gk7205v300-imx335 user.debug majestic[978]: MJPEG channel raised
Sep 20 17:55:59 gk7205v300-imx335 user.debug majestic[978]: JPEG was sent to client
Sep 20 17:56:01 gk7205v300-imx335 auth.info login[1008]: root login on 'console'
Sep 20 17:56:04 gk7205v300-imx335 user.debug majestic[978]: JPEG was sent to client
Sep 20 17:56:09 gk7205v300-imx335 user.debug majestic[978]: JPEG was sent to client
Sep 20 17:56:14 gk7205v300-imx335 user.debug majestic[978]: JPEG was sent to client
Sep 20 17:56:19 gk7205v300-imx335 user.debug majestic[978]: JPEG was sent to client
Sep 20 17:56:29 gk7205v300-imx335 user.debug majestic[978]: Stopped 1 channel
Sep 20 17:56:29 gk7205v300-imx335 user.debug majestic[978]: MJPEG channel dropped
Sep 20 17:56:31 gk7205v300-imx335 user.debug majestic[978]: MJPEG 1920x1080@5fps 50q
Sep 20 17:56:31 gk7205v300-imx335 user.debug majestic[978]:   venc_chn: 1, vpss_grp: 0, vpss_chn: 2
Sep 20 17:56:31 gk7205v300-imx335 user.info majestic[978]:   vpss chn 2 compression: asked none, kept none
Sep 20 17:56:31 gk7205v300-imx335 user.debug majestic[978]:   mpeg_enc took 2048KB of MMZ [2040]
Sep 20 17:56:31 gk7205v300-imx335 user.debug majestic[978]: MJPEG channel raised
Sep 20 17:56:31 gk7205v300-imx335 user.debug majestic[978]: JPEG was sent to client
Sep 20 17:56:41 gk7205v300-imx335 user.debug majestic[978]: Stopped 1 channel
Sep 20 17:56:41 gk7205v300-imx335 user.debug majestic[978]: MJPEG channel dropped
Sep 20 18:25:53 gk7205v300-imx335 kern.err kernel: goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
Sep 20 18:25:53 gk7205v300-imx335 kern.err kernel: goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
Sep 20 18:25:53 gk7205v300-imx335 kern.err kernel: goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
Sep 20 18:25:53 gk7205v300-imx335 kern.err kernel: goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
Sep 20 18:25:53 gk7205v300-imx335 kern.err kernel: goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
Sep 20 18:25:53 gk7205v300-imx335 kern.err kernel: goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
Sep 20 18:25:53 gk7205v300-imx335 kern.err kernel: goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
Sep 20 18:25:53 gk7205v300-imx335 kern.err kernel: goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
Sep 20 18:25:53 gk7205v300-imx335 kern.err kernel: goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
Sep 20 18:25:53 gk7205v300-imx335 kern.err kernel: goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
Sep 20 18:25:53 gk7205v300-imx335 kern.err kernel: goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
Sep 20 18:25:53 gk7205v300-imx335 kern.err kernel: goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
Sep 20 18:25:53 gk7205v300-imx335 kern.err kernel: goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
Sep 20 18:25:53 gk7205v300-imx335 kern.err kernel: goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
Sep 20 18:25:53 gk7205v300-imx335 kern.err kernel: goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
Sep 20 18:25:53 gk7205v300-imx335 kern.err kernel: goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
Sep 20 18:25:53 gk7205v300-imx335 user.debug majestic[978]: Nameserver 95.217.179.189:53 has failed: Network unreachable
Sep 20 18:25:58 gk7205v300-imx335 user.debug majestic[978]: Nameserver 1.1.1.1:53 has failed: Network unreachable
Sep 20 18:26:03 gk7205v300-imx335 user.debug majestic[978]: Nameserver 8.8.8.8:53 has failed: Network unreachable
Sep 20 18:26:03 gk7205v300-imx335 user.debug majestic[978]: All nameservers have failed
Sep 20 18:26:08 gk7205v300-imx335 user.warn majestic[978]: dns_cb result 67, retrying
Sep 20 18:26:24 gk7205v300-imx335 user.warn majestic[978]: dns_cb result 67, retrying
Sep 20 18:26:40 gk7205v300-imx335 user.warn majestic[978]: dns_cb result 67, retrying
Sep 20 18:30:25 gk7205v300-imx335 user.info majestic[978]: Change mode to NIGHT
Sep 20 18:30:27 gk7205v300-imx335 user.info majestic[978]: Change mode to DAY
Sep 20 18:30:28 gk7205v300-imx335 user.info majestic[978]: Live image CSC: contrast=50 hue=50 saturation=50 luminance=50
Sep 20 18:30:28 gk7205v300-imx335 user.info majestic[978]: Live image orientation (vpss): mirror=0 flip=0
Sep 20 18:30:28 gk7205v300-imx335 user.info majestic[978]: Live image CSC: contrast=50 hue=50 saturation=50 luminance=50
Sep 20 18:30:28 gk7205v300-imx335 user.info majestic[978]: Live image orientation (vpss): mirror=0 flip=0
Sep 20 18:30:29 gk7205v300-imx335 user.info majestic[978]: Live image CSC: contrast=50 hue=50 saturation=50 luminance=50
Sep 20 18:30:29 gk7205v300-imx335 user.info majestic[978]: Live image orientation (vpss): mirror=0 flip=0
Sep 20 18:30:29 gk7205v300-imx335 user.info majestic[978]: Live image CSC: contrast=50 hue=50 saturation=50 luminance=50
Sep 20 18:30:29 gk7205v300-imx335 user.info majestic[978]: Live image orientation (vpss): mirror=0 flip=0
Sep 20 18:30:30 gk7205v300-imx335 user.info majestic[978]: Live image CSC: contrast=50 hue=50 saturation=50 luminance=50
Sep 20 18:30:30 gk7205v300-imx335 user.info majestic[978]: Live image orientation (vpss): mirror=0 flip=0
Sep 20 18:30:30 gk7205v300-imx335 user.info majestic[978]: Live image CSC: contrast=50 hue=50 saturation=50 luminance=50
Sep 20 18:30:30 gk7205v300-imx335 user.info majestic[978]: Live image orientation (vpss): mirror=0 flip=0
Sep 20 18:30:30 gk7205v300-imx335 user.info majestic[978]: Live image CSC: contrast=50 hue=50 saturation=50 luminance=50
Sep 20 18:30:30 gk7205v300-imx335 user.info majestic[978]: Live image orientation (vpss): mirror=0 flip=0
Sep 20 18:30:30 gk7205v300-imx335 user.info majestic[978]: Live image CSC: contrast=50 hue=50 saturation=50 luminance=50
Sep 20 18:30:30 gk7205v300-imx335 user.info majestic[978]: Live image orientation (vpss): mirror=0 flip=0
Sep 20 18:30:30 gk7205v300-imx335 user.info majestic[978]: Live image CSC: contrast=50 hue=50 saturation=50 luminance=50
Sep 20 18:30:30 gk7205v300-imx335 user.info majestic[978]: Live image orientation (vpss): mirror=0 flip=0
Sep 20 18:30:30 gk7205v300-imx335 user.info majestic[978]: Live image CSC: contrast=50 hue=50 saturation=50 luminance=50
Sep 20 18:30:30 gk7205v300-imx335 user.info majestic[978]: Live image orientation (vpss): mirror=0 flip=0
Sep 20 18:30:32 gk7205v300-imx335 user.info majestic[978]: Live image CSC: contrast=50 hue=50 saturation=50 luminance=50
Sep 20 18:30:32 gk7205v300-imx335 user.info majestic[978]: Live image orientation (vpss): mirror=1 flip=0
Sep 20 18:30:33 gk7205v300-imx335 user.info majestic[978]: Live image CSC: contrast=50 hue=50 saturation=50 luminance=50
Sep 20 18:30:33 gk7205v300-imx335 user.info majestic[978]: Live image orientation (vpss): mirror=1 flip=0
Sep 20 18:30:35 gk7205v300-imx335 user.info majestic[978]: Live image CSC: contrast=50 hue=50 saturation=50 luminance=50
Sep 20 18:30:35 gk7205v300-imx335 user.info majestic[978]: Live image orientation (vpss): mirror=0 flip=0
Sep 20 21:11:13 gk7205v300-imx335 kern.info kernel: femac 10040000.ethernet eth0: Link is Down
Sep 21 00:26:08 gk7205v300-imx335 user.warn majestic[978]: dns_cb result 67, retrying
Sep 21 00:26:24 gk7205v300-imx335 user.warn majestic[978]: dns_cb result 67, retrying
Sep 21 00:26:40 gk7205v300-imx335 user.warn majestic[978]: dns_cb result 67, retrying
Sep 21 00:37:50 gk7205v300-imx335 user.warn majestic[978]: video0 is delivering far fewer frames than configured: 6.100 fps against the 20 it was set to. The sensor is most likely on a long exposure; the encoder is not the limit here
Sep 21 06:26:08 gk7205v300-imx335 user.warn majestic[978]: dns_cb result 67, retrying
Sep 21 06:26:24 gk7205v300-imx335 user.warn majestic[978]: dns_cb result 67, retrying
Sep 21 06:26:40 gk7205v300-imx335 user.warn majestic[978]: dns_cb result 67, retrying
Sep 21 08:57:04 gk7205v300-imx335 user.info majestic[978]: video0 is back within its configured rate: 386 of 512 kbit/s over the last 10 s
Sep 21 12:26:08 gk7205v300-imx335 user.warn majestic[978]: dns_cb result 67, retrying
Sep 21 12:26:24 gk7205v300-imx335 user.warn majestic[978]: dns_cb result 67, retrying
Sep 21 12:26:40 gk7205v300-imx335 user.warn majestic[978]: dns_cb result 67, retrying
Sep 21 12:30:51 gk7205v300-imx335 kern.info kernel: femac 10040000.ethernet eth0: Link is Up - 100Mbps/Full - flow control off
Sep 21 12:31:01 gk7205v300-imx335 kern.info kernel: femac 10040000.ethernet eth0: Link is Down
Sep 21 12:31:02 gk7205v300-imx335 kern.info kernel: femac 10040000.ethernet eth0: Link is Up - 100Mbps/Full - flow control rx/tx
Sep 21 14:51:16 gk7205v300-imx335 kern.err kernel: goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
Sep 21 14:51:16 gk7205v300-imx335 kern.err kernel: goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
Sep 21 14:51:16 gk7205v300-imx335 kern.err kernel: goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
Sep 21 14:51:16 gk7205v300-imx335 kern.err kernel: goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
Sep 21 14:51:16 gk7205v300-imx335 kern.err kernel: goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
Sep 21 14:51:16 gk7205v300-imx335 kern.err kernel: goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
Sep 21 14:51:16 gk7205v300-imx335 kern.err kernel: goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
Sep 21 14:51:16 gk7205v300-imx335 kern.err kernel: goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
Sep 21 14:51:16 gk7205v300-imx335 kern.err kernel: goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
Sep 21 14:51:16 gk7205v300-imx335 kern.err kernel: goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
Sep 21 14:51:16 gk7205v300-imx335 kern.err kernel: goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
Sep 21 14:51:16 gk7205v300-imx335 kern.err kernel: goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
Sep 21 14:51:16 gk7205v300-imx335 kern.err kernel: goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
Sep 21 14:51:16 gk7205v300-imx335 kern.err kernel: goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
Sep 21 14:51:16 gk7205v300-imx335 kern.err kernel: goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
Sep 21 14:51:16 gk7205v300-imx335 kern.err kernel: goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
root@gk7205v300-imx335:~#

dmesg | tail -200

root@gk7205v300-imx335:~# dmesg | tail -200
Virtual kernel memory layout:
    vector  : 0xffff0000 - 0xffff1000   (   4 kB)
    fixmap  : 0xffc00000 - 0xfff00000   (3072 kB)
    vmalloc : 0xc8800000 - 0xff800000   ( 880 MB)
    lowmem  : 0xc0000000 - 0xc8000000   ( 128 MB)
    modules : 0xbf000000 - 0xc0000000   (  16 MB)
      .text : 0xc0008000 - 0xc03faa20   (4043 kB)
      .init : 0xc04e7000 - 0xc0518000   ( 196 kB)
      .data : 0xc0518000 - 0xc053b520   ( 142 kB)
       .bss : 0xc053d000 - 0xc056bfe8   ( 188 kB)
SLUB: HWalign=64, Order=0-3, MinObjects=0, CPUs=1, Nodes=1
NR_IRQS:16 nr_irqs:16 16
Gic dist init...
arm_arch_timer: Architected cp15 timer(s) running at 50.00MHz (phys).
clocksource: arch_sys_counter: mask: 0xffffffffffffff max_cycles: 0xb8812736b, max_idle_ns: 440795202655 ns
sched_clock: 56 bits at 50MHz, resolution 20ns, wraps every 4398046511100ns
Switching to timer-based delay loop, resolution 20ns
clocksource: arm,sp804: mask: 0xffffffff max_cycles: 0xffffffff, max_idle_ns: 637086815595 ns
Calibrating delay loop (skipped), value calculated using timer frequency.. 100.00 BogoMIPS (lpj=500000)
pid_max: default: 32768 minimum: 301
Mount-cache hash table entries: 1024 (order: 0, 4096 bytes)
Mountpoint-cache hash table entries: 1024 (order: 0, 4096 bytes)
CPU: Testing write buffer coherency: ok
Setting up static identity map for 0x40008200 - 0x40008258
devtmpfs: initialized
VFP support v0.3: implementor 41 architecture 2 part 30 variant 7 rev 5
clocksource: jiffies: mask: 0xffffffff max_cycles: 0xffffffff, max_idle_ns: 19112604462750000 ns
futex hash table entries: 256 (order: -1, 3072 bytes)
pinctrl core: initialized pinctrl subsystem
NET: Registered protocol family 16
DMA: preallocated 256 KiB pool for atomic coherent allocations
Serial: AMBA PL011 UART driver
pstore: using zlib compression
pstore: Registered ramoops as persistent store backend
ramoops: attached 0x20000@0x41f00000, ecc: 0/0
12040000.uart: ttyAMA0 at MMIO 0x12040000 (irq = 21, base_baud = 0) is a PL011 rev2
console [ttyAMA0] enabled
12041000.uart: ttyAMA1 at MMIO 0x12041000 (irq = 22, base_baud = 0) is a PL011 rev2
12042000.uart: ttyAMA2 at MMIO 0x12042000 (irq = 23, base_baud = 0) is a PL011 rev2
ssp-pl022 12070000.spi: ARM PL022 driver, device ID: 0x00041022
ssp-pl022 12070000.spi: mapped registers from 0x12070000 to c8895000
ssp-pl022 12071000.spi: ARM PL022 driver, device ID: 0x00041022
ssp-pl022 12071000.spi: mapped registers from 0x12071000 to c8897000
usbcore: registered new interface driver usbfs
usbcore: registered new interface driver hub
usbcore: registered new device driver usb
clocksource: Switched to clocksource arch_sys_counter
NET: Registered protocol family 2
TCP established hash table entries: 1024 (order: 0, 4096 bytes)
TCP bind hash table entries: 1024 (order: 0, 4096 bytes)
TCP: Hash tables configured (established 1024 bind 1024)
UDP hash table entries: 256 (order: 0, 4096 bytes)
UDP-Lite hash table entries: 256 (order: 0, 4096 bytes)
NET: Registered protocol family 1
RPC: Registered named UNIX socket transport module.
RPC: Registered udp transport module.
RPC: Registered tcp transport module.
RPC: Registered tcp NFSv4.1 backchannel transport module.
workingset: timestamp_bits=30 max_order=15 bucket_order=0
squashfs: version 4.0 (2009/01/31) Phillip Lougher
jffs2: version 2.2 (NAND) (ZLIB) (RTIME) (c) 2001-2006 Red Hat, Inc.
yaffs: yaffs Installing.
Block layer SCSI generic (bsg) driver version 0.4 loaded (major 252)
io scheduler noop registered
io scheduler deadline registered (default)
pl061_gpio 120b0000.gpio_chip: PL061 GPIO chip @0x120b0000 registered
pl061_gpio 120b1000.gpio_chip: PL061 GPIO chip @0x120b1000 registered
pl061_gpio 120b2000.gpio_chip: PL061 GPIO chip @0x120b2000 registered
pl061_gpio 120b3000.gpio_chip: PL061 GPIO chip @0x120b3000 registered
pl061_gpio 120b4000.gpio_chip: PL061 GPIO chip @0x120b4000 registered
pl061_gpio 120b5000.gpio_chip: PL061 GPIO chip @0x120b5000 registered
pl061_gpio 120b6000.gpio_chip: PL061 GPIO chip @0x120b6000 registered
pl061_gpio 120b7000.gpio_chip: PL061 GPIO chip @0x120b7000 registered
pl061_gpio 120b8000.gpio_chip: PL061 GPIO chip @0x120b8000 registered
pl061_gpio 120b9000.gpio_chip: PL061 GPIO chip @0x120b9000 registered
brd: module loaded
loop: module loaded
bsp-sfc bsp_spi_nor.0: SPI Nor ID Table Version 1.3
bsp-sfc bsp_spi_nor.0: Force global unlock
bsp-sfc bsp_spi_nor.0: SR1:[02]->[00]
bsp-sfc bsp_spi_nor.0: all blocks are unlocked.
bsp-sfc bsp_spi_nor.0: SR1 [00], SR2 [00], SR3 [40]
bsp-sfc bsp_spi_nor.0: xt25f128b (Chipsize 16 Mbytes, Blocksize 64KiB)
5 cmdlinepart partitions found on MTD device sfc
5 cmdlinepart partitions found on MTD device sfc
Creating 5 MTD partitions on "sfc":
0x000000000000-0x000000040000 : "boot"
0x000000040000-0x000000050000 : "env"
0x000000050000-0x000000350000 : "kernel"
0x000000350000-0x000000d50000 : "rootfs"
0x000000d50000-0x000001000000 : "rootfs_data"
SPI Nand ID Table Version 2.7
Cannot found a valid SPI Nand Device
bsp_spi_nand_probe(157): Error: driver probe, result: -19
FEPHY:addr=1, la_am=0xa, ldo_am=0x4, r_tuning=0x22
libphy: femac_mii_bus: probed
libphy: Fixed MDIO Bus: probed
femac 10040000.ethernet: using random MAC address 2e:e0:9b:bf:0a:08
Generic PHY 10041100.mdio:01: attached PHY driver [Generic PHY] (mii_bus:phy_addr=10041100.mdio:01, irq=-1)
phy_id=0x20669906, phy_mode=mii
xhci-hcd xhci-hcd.0.auto: xHCI Host Controller
xhci-hcd xhci-hcd.0.auto: new USB bus registered, assigned bus number 1
xhci-hcd xhci-hcd.0.auto: hcc params 0x0220fe6c hci version 0x110 quirks 0x20010010
xhci-hcd xhci-hcd.0.auto: irq 135, io mem 0x10030000
hub 1-0:1.0: USB hub found
hub 1-0:1.0: 1 port detected
xhci-hcd xhci-hcd.0.auto: xHCI Host Controller
xhci-hcd xhci-hcd.0.auto: new USB bus registered, assigned bus number 2
usb usb2: We don't know the algorithms for LPM for this host, disabling LPM.
hub 2-0:1.0: USB hub found
hub 2-0:1.0: hub can't support USB3.0
goke_rtc 120e0000.rtc: rtc core: registered 120e0000.rtc as rtc0
goke_rtc 120e0000.rtc: RTC driver for goke enabled
i2c /dev entries driver
goke-i2c 12060000.i2c: goke-i2c0@100000hz registered
goke-i2c 12061000.i2c: goke-i2c1@100000hz registered
goke-i2c 12062000.i2c: goke-i2c2@100000hz registered
sdhci: Secure Digital Host Controller Interface driver
sdhci: Copyright(c) Pierre Ossman
sdhci-pltfm: SDHCI platform and OF driver helper
mmc0: SDHCI controller on 10010000.sdhci [10010000.sdhci] using ADMA in legacy mode
mmc1: SDHCI controller on 10020000.sdhci [10020000.sdhci] using ADMA in legacy mode
ip_tables: (C) 2000-2006 Netfilter Core Team
NET: Registered protocol family 10
NET: Registered protocol family 17
goke_rtc 120e0000.rtc: setting system clock to 1970-01-01 00:00:00 UTC (0)
VFS: Mounted root (squashfs filesystem) readonly on device 31:3.
devtmpfs: mounted
Freeing unused kernel memory: 196K (c04e7000 - c0518000)
This architecture does not have kernel memory protection.
random: crng init done
IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
femac 10040000.ethernet eth0: Link is Up - 100Mbps/Full - flow control rx/tx
IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
open_hwrng: loading out-of-tree module taints kernel.
hisi-hwrng: TRNG at 0x10080000+0x204 registered (pump)
==== online_flag=0, cmos_yuv_flag=0, sensor=imx335, chip=gk7205v300, board=demo====
==== g_quick_start_flag=0 ====
sysconfig init success!
Module media: init ok
Media Memory Zone Manager
osal 1.0 init success!
load sys.ko OK!
load region.ko OK!
load vgs.ko OK!
load vi.ko OK !
openipc_frame_ts: /dev/openipc-frame-ts minor=57, event types=2, ring_depth=256
ISP Mod init!
load vpss.ko OK!
load chnl.ko OK!
load vedu.ko OK!
load rc.ko OK!
load venc.ko OK!
load h264e.ko OK!
load h265e.ko OK!
load jpege.ko OK!
load ive.ko OK!
Load sensor_spi.ko OK !
load mipi_rx driver successful!
register dev
Watchdog Timer: 0.01 initialized. default_margin=60 sec (nodeamon= 0)
wtdg init ok. ver=Sep 20 2026, 17:55:22.
Load hi_cipher.ko success.
alloc_contig_range: [44900, 44b00) PFNs busy
goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
femac 10040000.ethernet eth0: Link is Down
femac 10040000.ethernet eth0: Link is Up - 100Mbps/Full - flow control off
femac 10040000.ethernet eth0: Link is Down
femac 10040000.ethernet eth0: Link is Up - 100Mbps/Full - flow control rx/tx
goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
goke-i2c 12060000.i2c: wait idle abort!, RIS: 0x611
root@gk7205v300-imx335:~#

Attachments

No response

[openipc-ai]

Thanks — this was four separate problems, and three of them were ours. I
reproduced all of it on a gk7205v300 + IMX335 running the same nightly.

Your .bin could not have loaded, on any camera of this kind

The camera worked out how long a PQTools profile should be before part of
the video pipeline existed, and the vendor library answers that question
differently depending on whether it does. On this SoC it decided a profile
should be 99920 bytes, while the same chip's own export is 100904 — the
missing 984 bytes are the noise-reduction block. So every genuine PQTools
profile for this part was refused on length, and the only clue was a message
saying the length was wrong with a number that matched nothing you had.

Worse: when the vendor library did reject a profile, the camera reported the
error and then carried on as though the profile had loaded. That is exactly
your "manual loading succeeds, yet parsing fails" — both halves were true at
the same time.

Both fixed. The profile is imported once the pipeline can take it, the expected
length comes out right, and a refused profile is reported as refused. Verified
end to end: export, import, and a normal picture after a cold boot.

Editing the .ini could not have worked either

The text profiles in /etc/sensors/iq/ are large — yours has 27 sections — and
the camera read exactly one of them, the noise-reduction one. [static_ae],
where AE speed lives, was parsed by nothing. No warning, no message; the edit
simply had no effect.

It is applied now: AutoSpeed, AutoTolerance, MaxHistOffset,
HistRatioSlope, AutoBlackDelayFrame, AutoWhiteDelayFrame and
AERunInterval all reach the chip. Measured on the lab camera, a profile
asking for speed 16 / tolerance 9 / darkening delay 3 used to leave the chip at
64 / 2 / 8 and now leaves it at 16 / 9 / 3.

The camera also now says at startup which sections of a profile it read and
which it did not, so an edit that changes nothing is visible as such.

You should not need PQTools for AE speed at all

Four settings, no profile and no tuning tool required:

setting	what it does
isp.aeSpeed	how fast exposure chases a change in light
isp.aeTolerance	how far off target it sits before moving at all
isp.aeBlackDelay	frames of hesitation before reacting to the picture getting darker
isp.aeWhiteDelay	the same, for the picture getting brighter

They apply without a restart. Leave one empty and the camera keeps whatever its
image profile asked for.

Your own tool reading back zeros

That one is not a bug. The ISP register mappings are only valid inside the
process that owns the ISP, so a second program asking the SDK for exposure
state while the streamer is running gets zeros — you will have seen
ISP[n] Mem NOT Init alongside it. There is no way round it from outside.

Three things found on the way

• Your camera was running the IMX307 tuning, not the IMX335 one, even
  though imx335.ini is installed beside it: default.ini is a fixed link to
  one sensor's profile per SoC family. With isp.iqProfile unset the camera
  now looks for /etc/sensors/iq/<sensor>.ini first.
• Pointing isp.iqProfile at an empty file killed the camera outright
  (fatal signal, watchdog released). Fixed; it is refused with a message now.
• pqtools on a Goke board fetched its board-side server from a URL that had
  been returning 404, so the documented route could not even be started.
  That is fixed too, along with two bugs in the fix itself: a corrupt download
  could leave a half-extracted directory that the next run treated as a valid
  cache, and the download used a predictable /tmp path that a symlink could
  redirect.

How to do what you set out to do

The wiki page now covers the whole round trip — running the tuning server,
exporting, putting the file on the camera, isp.iqProfile, and baking it into
an image:

https://github.com/OpenIPC/wiki/blob/master/en/image-quality-tuning.md

Short version:

cli -s .isp.iqProfile /etc/sensors/iq/tuned.bin

No restart needed. The name matters: a file ending in .ini is read as a text
profile, anything else as a PQTools binary profile.

All of this is in the nightly from tomorrow. Please try it and say if anything
still looks wrong.

[deedy-wang]

Got it. I had downloaded the wiki documentation a month ago, before this section was updated.

I followed the instructions in the updated wiki to deploy the bin file, but the test failed.

for example:
I tried setting the AE speed parameter to 84 in PQTools. When I read the values via cat /proc/umap/isp, it still showed 64. Also, importing the bin file threw a data length error.

I had already tested this approach myself before the wiki was updated, and it didn’t work back then either.

[deedy-wang]

After re-verification: when exporting the bin file using PQTool with 3DNR unchecked, the bin file size is normal and no error is reported during export.
When importing via the cli -s command, it prompts PQ BIN import error, Unknown error code CB000006. However, verification shows that the parameters modified in PQTool have taken effect.

Does the CB000006 error have any impact?

[widgetii]

@deedy-wang — thank you, those two reports together are more diagnostic than either alone. Taking your question first.

CB000006 is the imaging library's size-mismatch code

Not an unknown error: it means the library decided the profile was not the length it expects for this chip and refused it on that basis.

And yes, it has an impact. It is not a warning after a successful import — the part it refused is the part that did not get applied. Your PQTools changes to the ISP block took effect, which is why you can see them; the noise-reduction block in the same file did not. Whatever you tuned there is not on the camera.

Your two results together say you are still on the pre-fix build

Your sysinfo shows master+7f2dc18, 2026-09-20, and the length fixes landed after that. What makes it more than a guess is that your two attempts failed in different places, and only the old build produces that pair:

export	old build (what you saw)	build with the fix
3DNR checked — the full profile	the camera's own check refuses it: data length error	imports
3DNR unchecked — 984 bytes shorter	the camera's check passes, then the library refuses it: CB000006	refused by the camera first, naming both numbers

The old build worked the expected length out too early and got a number 984 bytes short — which is, to the byte, a profile with the noise-reduction block missing. So the file it accepted is the one the library then rejected, and the file the library wants is the one it refused. On a fixed build that is the other way round.

So: update to a current nightly and export again with 3DNR checked.

Please do not uncheck 3DNR to get past this

It is the natural thing to try and it makes things worse: you get a file the library will not fully accept, and you silently lose the noise-reduction half of your tuning out of it. The full export is the one that is meant to work.

cat /proc/umap/isp will not show you AE speed

Reading 64 there while PQTools says 84 does not mean the write failed. That node does not report the live metering state, and a second process asking the SDK for it gets zeros regardless — the register mappings are only valid inside the process that owns the imaging pipeline.

Ask the camera instead. From the late-September 2026 nightly builds:

> curl -u root:PASSWORD -o profile-day.ini \
      http://<camera>/api/v1/isp/profile.ini

That writes out what the camera is actually running, as a profile file in the same format /etc/sensors/iq/*.ini uses. AutoSpeed in its [static_ae] section is the number you were trying to read.

It is also the answer to the larger thing you set out to do. There is no converter from a .bin back to text — a binary profile carries no record of which named setting any part of it came from — but this direction works: tune with PQTools, download the result as text, and from there it is an ordinary file you can edit, diff, put on another camera or build into an image.

Two caveats that apply directly to you. isp.aeSpeed and friends are applied on top of the profile, so if any of them are set, theirs are the numbers you get back rather than PQTools' — the file says so in a comment. And for AE speed specifically you do not need PQTools at all: isp.aeSpeed sets it directly and applies without a restart.

The wiki page

Updated again since the comment above, and it now covers the text half as well — what the camera reads at boot, why the two formats are not interchangeable, and getting your tuning back out as text:

https://wiki.openipc.org/en/image-quality-tuning.md

Verified end to end on a gk7205v300 + IMX335 running the shipped IMX335 profile.

[deedy-wang]

Thanks,@widgetii
Got it. After upgrading to the latest nightly build and enabling 3DNR, Majestic no longer throws errors when importing the BIN file.

However, when the sensor engineer uses PQTools on this latest nightly to modify the four parameters under TSWDR->DRC->Tone mapping: after writing the values, switching to another tab and returning to this page, those four parameters revert back to their default values.

I then tried another workflow: export the BIN file first, load it onto the board with cli -s .isp.iqProfile, and then modify those four parameters. In this scenario, the modified values can be retained.

Is this expected behavior? Or is it mandatory for the sensor engineer to load the BIN file via cli -s .isp.iqProfile before starting tuning?

[widgetii]

@deedy-wang — good, that closes the import half. Thank you for coming back with the result, and for the second observation: it was a real gap and there is now a key for it.

What was putting those four back

Those four are not static tuning on this chip. In the text profile format they exist only as an ISO ladder, and your camera is running one. /etc/sensors/iq/imx335.ini, which majestic loads when isp.iqProfile is unset, carries this:

[dynamic_linear_drc]
Enable                    = "1"
IsoLevel                  = "100,  200,   400,   800,  1600,   3200,   6400,  12800,  25600,  51200"
...
Asymmetry                 = "  4,    4,     4,     4,     6,      6,      6,      6,      6,      6"
SecondPole                = "160,  160,   160,   160,   160,    160,    160,    150,    150,    150"
Compress                  = "150,  150,   150,   150,   150,    150,    150,    150,    150,    150"
Stretch                   = " 45,   45,    45,    45,    45,     45,     45,     50,     50,     50"

majestic reads the live ISO every 10 seconds and writes that ladder's value for the current ISO into the DRC attribute — on every tick, not only when the ISO moves. So a value PQTools writes into those four lives at most ten seconds. Switching tabs and coming back takes longer than that.

What you saw is therefore not a reset to defaults. It is the profile's rung for the light the camera is in. Under office light at ISO 100–200 that ladder reads 4 / 160 / 150 / 45.

There is now a switch for this

> cli -s .isp.externalTuner true

No restart. While it is set, majestic stops writing the ISP so your engineer's tuner owns it: the ISO ladders stand down, automatic tuning stops moving contrast and dehaze, and a day/night change does not re-read the profile. The ISO monitor keeps running and keeps logging the ISO and exposure, which is what tells you which rung you are tuning.

Turn it off and the camera takes the picture back on the next tick:

> cli -s .isp.externalTuner false

Nothing is saved. Read the result out before you clear it:

> curl -u root:PASSWORD -o profile.ini \
      http://<camera-address>/api/v1/isp/profile.ini

Is the .bin route mandatory? No — and it was quietly costing you

Your workaround worked because a .bin is not a .ini: majestic parses no ladders out of it, so nothing rewrites DRC afterwards.

But it was doing something you would not have wanted, and this is worth knowing because you have been importing .bin files throughout. majestic was overwriting part of every profile you imported. After the import it wrote its own compiled-in tables over sharpening, noise reduction and local contrast. Measured on a lab hi3516ev300 by exporting a profile from the camera, importing it, and reading the chip back:

key	what the profile held	what the camera ran
AutoLumaWgt_0	31 ×16	31,31,31,31,31,28,26,24,24,31…
LDCIGaussLPFSigma	28	58
FineStr	64,72,75,90,90,…	64,64,64,64,64,64,78,78,…

That is fixed. An accepted binary profile now owns every ISP module and majestic keeps its hands off.

What a .bin still costs, by design: it carries no ISO ladders, so DRC, dehaze and the VPSS 3DNR table do not follow gain. Fine on a tuning bench; not what you want on a shipped camera. So no — do not make it a prerequisite for tuning. Use isp.externalTuner for the session and keep the .ini.

The four parameters cannot be exported, and that will not change

GET /api/v1/isp/profile.ini will not contain Asymmetry, SecondPole, Compress or Stretch. There is no [static_drc] key for them in the vendor's own format — a chip readback is one rung, and writing it out as a flat value would claim one light level's answer for every light level.

So for these four the deliverable is the ladder. Tune live to find the values, then write them into the ladder rows for the ISO band you tuned under, in a copy of the profile:

> cp /etc/sensors/iq/imx335.ini /etc/sensors/iq/imx335-tuned.ini
> vi /etc/sensors/iq/imx335-tuned.ini          # the four rows above
> cli -s .isp.iqProfile /etc/sensors/iq/imx335-tuned.ini
> killall -1 majestic

Edit the [ir_dynamic_linear_drc] copy too — that is the same ladder for when the IR-cut filter is out of the light path.

One change to expect in the picture

Also landing: [dynamic_gamma] now loads on the IMX335 profile. It never used to. That file declares its climb thresholds (3200, 6400, 25600) on an ISO scale against fall thresholds (400000, 800000, 1600000) in exposure units — two different quantities in one section — so the whole section was refused and gamma was left at whatever the sensor library set. majestic now ignores the ISO-scaled array and switches on the exposure one.

This moves the picture, so you will see it between builds. Measured against a colour checker: highlights up about 24 luma, midtones up 13, the darkest patch down 14 — the vendor's own tuned curve, where previously none was applied. If your engineer is comparing tuning results across builds, that is why.

Two other things that will fight a live session

Both are covered by isp.externalTuner, but worth knowing if you tune without it:

• The dehaze and CSC pages. image.tuning is on by default and walks contrast, saturation and dehaze strength every couple of seconds.
• A day/night transition re-reads the whole profile.

The static sections — sharpen, NR, LDCI, AE and the rest — are applied once at bring-up and never rewritten, which is why your earlier PQTools work in those stayed put.

Availability

All of the above is on master now and should be in tonight's nightly. Please do come back if the four parameters still do not hold with isp.externalTuner set — that path is verified on a hi3516ev300 here, and your gk7205v300 is a different member of the same family.