1 .. -*- coding: utf-8; mode: rst -*-
3 .. _v4l2-meta-fmt-params:
4 .. _v4l2-meta-fmt-stat-3a:
6 ******************************************************************
7 V4L2_META_FMT_IPU3_PARAMS ('ip3p'), V4L2_META_FMT_IPU3_3A ('ip3s')
8 ******************************************************************
10 .. c:type:: ipu3_uapi_stats_3a
15 For IPU3 ImgU, the 3A statistics accelerators collect different statistics over
16 an input bayer frame. Those statistics, defined in data struct :c:type:`ipu3_uapi_stats_3a`,
17 are obtained from "ipu3-imgu 3a stat" metadata capture video node, which are then
18 passed to user space for statistics analysis using :c:type:`v4l2_meta_format` interface.
20 The statistics collected are AWB (Auto-white balance) RGBS (Red, Green, Blue and
21 Saturation measure) cells, AWB filter response, AF (Auto-focus) filter response,
22 and AE (Auto-exposure) histogram.
24 struct :c:type:`ipu3_uapi_4a_config` saves configurable parameters for all above.
28 struct ipu3_uapi_stats_3a {
29 struct ipu3_uapi_awb_raw_buffer awb_raw_buffer;
30 struct ipu3_uapi_ae_raw_buffer_aligned ae_raw_buffer[IPU3_UAPI_MAX_STRIPES];
31 struct ipu3_uapi_af_raw_buffer af_raw_buffer;
32 struct ipu3_uapi_awb_fr_raw_buffer awb_fr_raw_buffer;
33 struct ipu3_uapi_4a_config stats_4a_config;
34 __u32 ae_join_buffers;
36 struct ipu3_uapi_stats_3a_bubble_info_per_stripe stats_3a_bubble_per_stripe;
37 struct ipu3_uapi_ff_status stats_3a_status;
40 .. c:type:: ipu3_uapi_params
45 IPU3 pipeline has a number of image processing stages, each of which takes a
46 set of parameters as input. The major stages of pipelines are shown here:
48 Raw pixels -> Bayer Downscaling -> Optical Black Correction ->
50 Linearization -> Lens Shading Correction -> White Balance / Exposure /
52 Focus Apply -> Bayer Noise Reduction -> ANR -> Demosaicing -> Color
54 Correction Matrix -> Gamma correction -> Color Space Conversion ->
56 Chroma Down Scaling -> Chromatic Noise Reduction -> Total Color
58 Correction -> XNR3 -> TNR -> DDR
60 The table below presents a description of the above algorithms.
62 ======================== =======================================================
64 ======================== =======================================================
65 Optical Black Correction Optical Black Correction block subtracts a pre-defined
66 value from the respective pixel values to obtain better
68 Defined in :c:type:`ipu3_uapi_obgrid_param`.
69 Linearization This algo block uses linearization parameters to
70 address non-linearity sensor effects. The Lookup table
72 :c:type:`ipu3_uapi_isp_lin_vmem_params`.
73 SHD Lens shading correction is used to correct spatial
74 non-uniformity of the pixel response due to optical
75 lens shading. This is done by applying a different gain
76 for each pixel. The gain, black level etc are
77 configured in :c:type:`ipu3_uapi_shd_config_static`.
78 BNR Bayer noise reduction block removes image noise by
79 applying a bilateral filter.
80 See :c:type:`ipu3_uapi_bnr_static_config` for details.
81 ANR Advanced Noise Reduction is a block based algorithm
82 that performs noise reduction in the Bayer domain. The
83 convolution matrix etc can be found in
84 :c:type:`ipu3_uapi_anr_config`.
85 Demosaicing Demosaicing converts raw sensor data in Bayer format
86 into RGB (Red, Green, Blue) presentation. Then add
87 outputs of estimation of Y channel for following stream
88 processing by Firmware. The struct is defined as
89 :c:type:`ipu3_uapi_dm_config`. (TODO)
90 Color Correction Color Correction algo transforms sensor specific color
91 space to the standard "sRGB" color space. This is done
92 by applying 3x3 matrix defined in
93 :c:type:`ipu3_uapi_ccm_mat_config`.
94 Gamma correction Gamma correction :c:type:`ipu3_uapi_gamma_config` is a
95 basic non-linear tone mapping correction that is
96 applied per pixel for each pixel component.
97 CSC Color space conversion transforms each pixel from the
98 RGB primary presentation to YUV (Y: brightness,
99 UV: Luminance) presentation. This is done by applying
100 a 3x3 matrix defined in
101 :c:type:`ipu3_uapi_csc_mat_config`
102 CDS Chroma down sampling
103 After the CSC is performed, the Chroma Down Sampling
104 is applied for a UV plane down sampling by a factor
105 of 2 in each direction for YUV 4:2:0 using a 4x2
106 configurable filter :c:type:`ipu3_uapi_cds_params`.
107 CHNR Chroma noise reduction
108 This block processes only the chrominance pixels and
109 performs noise reduction by cleaning the high
111 See struct :c:type:`ipu3_uapi_yuvp1_chnr_config`.
112 TCC Total color correction as defined in struct
113 :c:type:`ipu3_uapi_yuvp2_tcc_static_config`.
114 XNR3 eXtreme Noise Reduction V3 is the third revision of
115 noise reduction algorithm used to improve image
116 quality. This removes the low frequency noise in the
117 captured image. Two related structs are being defined,
118 :c:type:`ipu3_uapi_isp_xnr3_params` for ISP data memory
119 and :c:type:`ipu3_uapi_isp_xnr3_vmem_params` for vector
121 TNR Temporal Noise Reduction block compares successive
122 frames in time to remove anomalies / noise in pixel
123 values. :c:type:`ipu3_uapi_isp_tnr3_vmem_params` and
124 :c:type:`ipu3_uapi_isp_tnr3_params` are defined for ISP
125 vector and data memory respectively.
126 ======================== =======================================================
128 A few stages of the pipeline will be executed by firmware running on the ISP
129 processor, while many others will use a set of fixed hardware blocks also
130 called accelerator cluster (ACC) to crunch pixel data and produce statistics.
132 ACC parameters of individual algorithms, as defined by
133 :c:type:`ipu3_uapi_acc_param`, can be chosen to be applied by the user
134 space through struct :c:type:`ipu3_uapi_flags` embedded in
135 :c:type:`ipu3_uapi_params` structure. For parameters that are configured as
136 not enabled by the user space, the corresponding structs are ignored by the
137 driver, in which case the existing configuration of the algorithm will be
140 Both 3A statistics and pipeline parameters described here are closely tied to
141 the underlying camera sub-system (CSS) APIs. They are usually consumed and
142 produced by dedicated user space libraries that comprise the important tuning
143 tools, thus freeing the developers from being bothered with the low level
144 hardware and algorithm details.
146 It should be noted that IPU3 DMA operations require the addresses of all data
147 structures (that includes both input and output) to be aligned on 32 byte
150 The meta data :c:type:`ipu3_uapi_params` will be sent to "ipu3-imgu parameters"
151 video node in ``V4L2_BUF_TYPE_META_CAPTURE`` format.
155 struct ipu3_uapi_params {
156 /* Flags which of the settings below are to be applied */
157 struct ipu3_uapi_flags use;
159 /* Accelerator cluster parameters */
160 struct ipu3_uapi_acc_param acc_param;
162 /* ISP vector address space parameters */
163 struct ipu3_uapi_isp_lin_vmem_params lin_vmem_params;
164 struct ipu3_uapi_isp_tnr3_vmem_params tnr3_vmem_params;
165 struct ipu3_uapi_isp_xnr3_vmem_params xnr3_vmem_params;
167 /* ISP data memory (DMEM) parameters */
168 struct ipu3_uapi_isp_tnr3_params tnr3_dmem_params;
169 struct ipu3_uapi_isp_xnr3_params xnr3_dmem_params;
171 /* Optical black level compensation */
172 struct ipu3_uapi_obgrid_param obgrid_param;
175 Intel IPU3 ImgU uAPI data types
176 ===============================
178 .. kernel-doc:: drivers/staging/media/ipu3/include/intel-ipu3.h