1	Cropping and Scaling algorithm, used in the sh_mobile_ceu_camera driver
2	=======================================================================
3
4Terminology
5-----------
6
7sensor scales: horizontal and vertical scales, configured by the sensor driver
8host scales: -"- host driver
9combined scales: sensor_scale * host_scale
10
11
12Generic scaling / cropping scheme
13---------------------------------
14
15-1--
16|
17-2-- -\
18|      --\
19|         --\
20+-5-- .      -- -3-- -\
21|      `...            -\
22|          `... -4-- .   - -7..
23|                     `.
24|                       `. .6--
25|
26|                        . .6'-
27|                      .´
28|           ... -4'- .´
29|       ...´             - -7'.
30+-5'- .´               -/
31|            -- -3'- -/
32|         --/
33|      --/
34-2'- -/
35|
36|
37-1'-
38
39In the above chart minuses and slashes represent "real" data amounts, points and
40accents represent "useful" data, basically, CEU scaled and cropped output,
41mapped back onto the client's source plane.
42
43Such a configuration can be produced by user requests:
44
45S_CROP(left / top = (5) - (1), width / height = (5') - (5))
46S_FMT(width / height = (6') - (6))
47
48Here:
49
50(1) to (1') - whole max width or height
51(1) to (2)  - sensor cropped left or top
52(2) to (2') - sensor cropped width or height
53(3) to (3') - sensor scale
54(3) to (4)  - CEU cropped left or top
55(4) to (4') - CEU cropped width or height
56(5) to (5') - reverse sensor scale applied to CEU cropped width or height
57(2) to (5)  - reverse sensor scale applied to CEU cropped left or top
58(6) to (6') - CEU scale - user window
59
60
61S_FMT
62-----
63
64Do not touch input rectangle - it is already optimal.
65
661. Calculate current sensor scales:
67
68	scale_s = ((2') - (2)) / ((3') - (3))
69
702. Calculate "effective" input crop (sensor subwindow) - CEU crop scaled back at
71current sensor scales onto input window - this is user S_CROP:
72
73	width_u = (5') - (5) = ((4') - (4)) * scale_s
74
753. Calculate new combined scales from "effective" input window to requested user
76window:
77
78	scale_comb = width_u / ((6') - (6))
79
804. Calculate sensor output window by applying combined scales to real input
81window:
82
83	width_s_out = ((7') - (7)) = ((2') - (2)) / scale_comb
84
855. Apply iterative sensor S_FMT for sensor output window.
86
87	subdev->video_ops->s_fmt(.width = width_s_out)
88
896. Retrieve sensor output window (g_fmt)
90
917. Calculate new sensor scales:
92
93	scale_s_new = ((3')_new - (3)_new) / ((2') - (2))
94
958. Calculate new CEU crop - apply sensor scales to previously calculated
96"effective" crop:
97
98	width_ceu = (4')_new - (4)_new = width_u / scale_s_new
99	left_ceu = (4)_new - (3)_new = ((5) - (2)) / scale_s_new
100
1019. Use CEU cropping to crop to the new window:
102
103	ceu_crop(.width = width_ceu, .left = left_ceu)
104
10510. Use CEU scaling to scale to the requested user window:
106
107	scale_ceu = width_ceu / width
108
109
110S_CROP
111------
112
113The API at http://v4l2spec.bytesex.org/spec/x1904.htm says:
114
115"...specification does not define an origin or units. However by convention
116drivers should horizontally count unscaled samples relative to 0H."
117
118We choose to follow the advise and interpret cropping units as client input
119pixels.
120
121Cropping is performed in the following 6 steps:
122
1231. Request exactly user rectangle from the sensor.
124
1252. If smaller - iterate until a larger one is obtained. Result: sensor cropped
126   to 2 : 2', target crop 5 : 5', current output format 6' - 6.
127
1283. In the previous step the sensor has tried to preserve its output frame as
129   good as possible, but it could have changed. Retrieve it again.
130
1314. Sensor scaled to 3 : 3'. Sensor's scale is (2' - 2) / (3' - 3). Calculate
132   intermediate window: 4' - 4 = (5' - 5) * (3' - 3) / (2' - 2)
133
1345. Calculate and apply host scale = (6' - 6) / (4' - 4)
135
1366. Calculate and apply host crop: 6 - 7 = (5 - 2) * (6' - 6) / (5' - 5)
137
138--
139Author: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
140