apps/CameraITS/tests/scene1/test_capture_result.py

# Copyright 2013 The Android Open Source Project
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#      http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

import os.path
import its.caps
import its.device
import its.image
import its.objects
import matplotlib.pyplot
import mpl_toolkits.mplot3d  # Required for 3d plot to work
import numpy


def main():
    """Test that valid data comes back in CaptureResult objects.
    """
    global NAME, auto_req, manual_req, w_map, h_map
    global manual_tonemap, manual_transform, manual_gains, manual_region
    global manual_exp_time, manual_sensitivity, manual_gains_ok

    NAME = os.path.basename(__file__).split(".")[0]

    with its.device.ItsSession() as cam:
        props = cam.get_camera_properties()
        its.caps.skip_unless(its.caps.manual_sensor(props) and
                             its.caps.manual_post_proc(props) and
                             its.caps.per_frame_control(props))

        manual_tonemap = [0,0, 1,1]  # Linear
        manual_transform = its.objects.float_to_rational(
                [-1.5,-1.0,-0.5, 0.0,0.5,1.0, 1.5,2.0,3.0])
        manual_gains = [1,1.5,2.0,3.0]
        manual_region = [{"x":8,"y":8,"width":128,"height":128,"weight":1}]
        manual_exp_time = min(props["android.sensor.info.exposureTimeRange"])
        manual_sensitivity = min(props["android.sensor.info.sensitivityRange"])

        # The camera HAL may not support different gains for two G channels.
        manual_gains_ok = [[1,1.5,2.0,3.0],[1,1.5,1.5,3.0],[1,2.0,2.0,3.0]]

        auto_req = its.objects.auto_capture_request()
        auto_req["android.statistics.lensShadingMapMode"] = 1

        manual_req = {
                "android.control.mode": 0,
                "android.control.aeMode": 0,
                "android.control.awbMode": 0,
                "android.control.afMode": 0,
                "android.sensor.sensitivity": manual_sensitivity,
                "android.sensor.exposureTime": manual_exp_time,
                "android.colorCorrection.mode": 0,
                "android.colorCorrection.transform": manual_transform,
                "android.colorCorrection.gains": manual_gains,
                "android.tonemap.mode": 0,
                "android.tonemap.curve": {"red": manual_tonemap,
                                          "green": manual_tonemap,
                                          "blue": manual_tonemap},
                "android.control.aeRegions": manual_region,
                "android.control.afRegions": manual_region,
                "android.control.awbRegions": manual_region,
                "android.statistics.lensShadingMapMode": 1
                }

        sync_latency = its.caps.sync_latency(props)
        print "Testing auto capture results"
        lsc_map_auto = test_auto(cam, props, sync_latency)
        print "Testing manual capture results"
        test_manual(cam, lsc_map_auto, props, sync_latency)
        print "Testing auto capture results again"
        test_auto(cam, props, sync_latency)


def is_close_float(n1, n2):
    """A very loose definition for two floats being close to each other.

    there may be different interpolation and rounding used to get the
    two values, and all this test is looking at is whether there is
    something obviously broken; it's not looking for a perfect match.

    Args:
        n1:     float 1
        n2:     float 2
    Returns:
        Boolean
    """
    return abs(n1 - n2) < 0.05


def is_close_rational(n1, n2):
    return is_close_float(its.objects.rational_to_float(n1),
                          its.objects.rational_to_float(n2))


def draw_lsc_plot(w_map, h_map, lsc_map, name):
    for ch in range(4):
        fig = matplotlib.pyplot.figure()
        ax = fig.gca(projection="3d")
        xs = numpy.array([range(w_map)] * h_map).reshape(h_map, w_map)
        ys = numpy.array([[i]*w_map for i in range(h_map)]).reshape(
                h_map, w_map)
        zs = numpy.array(lsc_map[ch::4]).reshape(h_map, w_map)
        ax.plot_wireframe(xs, ys, zs)
        matplotlib.pyplot.savefig("%s_plot_lsc_%s_ch%d.png"%(NAME, name, ch))


def test_auto(cam, props, sync_latency):
    # Get 3A lock first, so the auto values in the capture result are
    # populated properly.
    rect = [[0, 0, 1, 1, 1]]
    mono_camera = its.caps.mono_camera(props)
    cam.do_3a(rect, rect, rect, do_af=False, mono_camera=mono_camera)

    cap = its.device.do_capture_with_latency(cam, auto_req, sync_latency)
    cap_res = cap["metadata"]

    gains = cap_res["android.colorCorrection.gains"]
    transform = cap_res["android.colorCorrection.transform"]
    exp_time = cap_res["android.sensor.exposureTime"]
    lsc_obj = cap_res["android.statistics.lensShadingCorrectionMap"]
    lsc_map = lsc_obj["map"]
    w_map = lsc_obj["width"]
    h_map = lsc_obj["height"]
    ctrl_mode = cap_res["android.control.mode"]

    print "Control mode:", ctrl_mode
    print "Gains:", gains
    print "Transform:", [its.objects.rational_to_float(t)
                         for t in transform]
    if props["android.control.maxRegionsAe"] > 0:
        print "AE region:", cap_res["android.control.aeRegions"]
    if props["android.control.maxRegionsAf"] > 0:
        print "AF region:", cap_res["android.control.afRegions"]
    if props["android.control.maxRegionsAwb"] > 0:
        print "AWB region:", cap_res["android.control.awbRegions"]
    print "LSC map:", w_map, h_map, lsc_map[:8]

    assert(ctrl_mode == 1)

    # Color correction gain and transform must be valid.
    assert(len(gains) == 4)
    assert(len(transform) == 9)
    assert(all([g > 0 for g in gains]))
    assert(all([t["denominator"] != 0 for t in transform]))

    # Color correction should not match the manual settings.
    assert(any([not is_close_float(gains[i], manual_gains[i])
                for i in xrange(4)]))
    assert(any([not is_close_rational(transform[i], manual_transform[i])
                for i in xrange(9)]))

    # Exposure time must be valid.
    assert(exp_time > 0)

    # Lens shading map must be valid.
    assert(w_map > 0 and h_map > 0 and w_map * h_map * 4 == len(lsc_map))
    assert(all([m >= 1 for m in lsc_map]))

    draw_lsc_plot(w_map, h_map, lsc_map, "auto")

    return lsc_map


def test_manual(cam, lsc_map_auto, props, sync_latency):
    cap = its.device.do_capture_with_latency(cam, manual_req, sync_latency)
    cap_res = cap["metadata"]

    gains = cap_res["android.colorCorrection.gains"]
    transform = cap_res["android.colorCorrection.transform"]
    curves = [cap_res["android.tonemap.curve"]["red"],
              cap_res["android.tonemap.curve"]["green"],
              cap_res["android.tonemap.curve"]["blue"]]
    exp_time = cap_res["android.sensor.exposureTime"]
    lsc_obj = cap_res["android.statistics.lensShadingCorrectionMap"]
    lsc_map = lsc_obj["map"]
    w_map = lsc_obj["width"]
    h_map = lsc_obj["height"]
    ctrl_mode = cap_res["android.control.mode"]

    print "Control mode:", ctrl_mode
    print "Gains:", gains
    print "Transform:", [its.objects.rational_to_float(t)
                         for t in transform]
    print "Tonemap:", curves[0][1::16]
    if props["android.control.maxRegionsAe"] > 0:
        print "AE region:", cap_res["android.control.aeRegions"]
    if props["android.control.maxRegionsAf"] > 0:
        print "AF region:", cap_res["android.control.afRegions"]
    if props["android.control.maxRegionsAwb"] > 0:
        print "AWB region:", cap_res["android.control.awbRegions"]
    print "LSC map:", w_map, h_map, lsc_map[:8]

    assert(ctrl_mode == 0)

    # Color correction gain and transform must be valid.
    # Color correction gains and transform should be the same size and
    # values as the manually set values.
    assert(len(gains) == 4)
    assert(len(transform) == 9)
    assert( all([is_close_float(gains[i], manual_gains_ok[0][i])
                 for i in xrange(4)]) or
            all([is_close_float(gains[i], manual_gains_ok[1][i])
                 for i in xrange(4)]) or
            all([is_close_float(gains[i], manual_gains_ok[2][i])
                 for i in xrange(4)]))
    assert(all([is_close_rational(transform[i], manual_transform[i])
                for i in xrange(9)]))

    # Tonemap must be valid.
    # The returned tonemap must be linear.
    for c in curves:
        assert(len(c) > 0)
        assert(all([is_close_float(c[i], c[i+1])
                    for i in xrange(0,len(c),2)]))

    # Exposure time must be close to the requested exposure time.
    assert(is_close_float(exp_time/1000000.0, manual_exp_time/1000000.0))

    # Lens shading map must be valid.
    assert(w_map > 0 and h_map > 0 and w_map * h_map * 4 == len(lsc_map))
    assert(all([m >= 1 for m in lsc_map]))

    draw_lsc_plot(w_map, h_map, lsc_map, "manual")


if __name__ == "__main__":
    main()