test_codec.py 7.18 KB
Newer Older
André Anjos's avatar
André Anjos committed
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
#!/usr/bin/env python
# vim: set fileencoding=utf-8 :
# Andre Anjos <andre.anjos@idiap.ch>
# Tue 12 Nov 09:25:56 2013

"""Runs quality tests on video codecs supported
"""

import os
import numpy
import nose.tools
from . import test_utils
from .utils import color_distortion, frameskip_detection, quality_degradation

# These are some global parameters for the test.
INPUT_VIDEO = test_utils.datafile('test.mov', __name__)

@test_utils.ffmpeg_found()
def check_format_codec(function, shape, framerate, format, codec, maxdist):

  length, height, width = shape
  fname = test_utils.temporary_filename(suffix='.%s' % format)

  try:
    orig, framerate, encoded = function(shape, framerate, format, codec, fname)
    reloaded = encoded.load()

    # test number of frames is correct
    assert len(orig) == len(encoded), "original length %d != %d encoded" % (len(orig), len(encoded))
    assert len(orig) == len(reloaded), "original length %d != %d reloaded" % (len(orig), len(reloaded))

    # test distortion patterns (quick sequential check)
    dist = []
    for k, of in enumerate(orig):
      dist.append(abs(of.astype('float64')-reloaded[k].astype('float64')).mean())
    assert max(dist) <= maxdist, "max(distortion) %g > %g allowed" % (max(dist), maxdist)

    # assert we can randomly access any frame (choose 3 at random)
    for k in numpy.random.randint(length, size=(3,)):
      rdist = abs(orig[k].astype('float64')-encoded[k].astype('float64')).mean()
      assert rdist <= maxdist, "distortion(frame[%d]) %g > %g allowed" % (k, rdist, maxdist)

    # make sure that the encoded frame rate is not off by a big amount
    assert abs(framerate - encoded.frame_rate) <= (1.0/length), "reloaded framerate %g differs from original %g by more than %g" % (encoded.frame_rate, framerate, 1.0/length)

  finally:

    if os.path.exists(fname): os.unlink(fname)

def test_format_codecs():

  length = 30
  width = 128
  height = 128
  framerate = 30.
  shape = (length, height, width)
  methods = dict(
      frameskip = frameskip_detection,
      color     = color_distortion,
      noise     = quality_degradation,
      )

  # distortion patterns for specific codecs
  distortions = dict(
      # we require high standards by default
      default    = dict(frameskip=0.1,  color=9.0,  noise=45.),

      # high-quality encoders
      zlib       = dict(frameskip=0.0,  color=0.0, noise=0.0),
      ffv1       = dict(frameskip=0.05, color=9.,  noise=46.),
      vp8        = dict(frameskip=0.3,  color=9.0, noise=65.),
      libvpx     = dict(frameskip=0.3,  color=9.0, noise=65.),
      h264       = dict(frameskip=0.4,  color=9.0, noise=50.),
      libx264    = dict(frameskip=0.4,  color=9.0, noise=50.),
      theora     = dict(frameskip=0.5,  color=9.0, noise=70.),
      libtheora  = dict(frameskip=0.5,  color=9.0, noise=70.),
      mpeg4      = dict(frameskip=1.0,  color=9.0, noise=55.),

      # older, but still good quality encoders
      mjpeg      = dict(frameskip=1.2,  color=9.0, noise=50.),
      mpegvideo  = dict(frameskip=1.3,  color=9.0, noise=55.),
      mpeg2video = dict(frameskip=1.3,  color=9.0, noise=55.),
      mpeg1video = dict(frameskip=1.4,  color=9.0, noise=50.),

      # low quality encoders - avoid using - available for compatibility
      wmv2       = dict(frameskip=3.0,  color=10., noise=50.),
      wmv1       = dict(frameskip=2.5,  color=10., noise=50.),
      msmpeg4    = dict(frameskip=5.,   color=10., noise=50.),
      msmpeg4v2  = dict(frameskip=5.,   color=10., noise=50.),
      )

  # some exceptions
  if test_utils.ffmpeg_version_lessthan('0.6'):
    distortions['ffv1']['frameskip'] = 0.55
    distortions['mpeg1video']['frameskip'] = 1.5
    distortions['mpegvideo']['color'] = 9.0
    distortions['mpegvideo']['frameskip'] = 1.4
    distortions['mpeg2video']['color'] = 9.0
    distortions['mpeg2video']['frameskip'] = 1.4

  from .version import externals
  if externals['FFmpeg']['ffmpeg'] != 'unavailable':
    from .version import supported_videowriter_formats
    SUPPORTED = supported_videowriter_formats()
    for format in SUPPORTED:
      for codec in SUPPORTED[format]['supported_codecs']:
        for method in methods:
          check_format_codec.description = "%s.test_%s_format_%s_codec_%s" % (__name__, method, format, codec)
          distortion = distortions.get(codec, distortions['default'])[method]
          yield check_format_codec, methods[method], shape, framerate, format, codec, distortion

@test_utils.ffmpeg_found()
def check_user_video(format, codec, maxdist):

  from . import VideoReader, VideoWriter
  fname = test_utils.temporary_filename(suffix='.%s' % format)
  MAXLENTH = 10 #use only the first 10 frames

  try:

    orig_vreader = VideoReader(INPUT_VIDEO)
    orig = orig_vreader[:MAXLENTH]
    (olength, _, oheight, owidth) = orig.shape
    assert len(orig) == MAXLENTH, "original length %d != %d MAXLENTH" % (len(orig), MAXLENTH)

    # encode the input video using the format and codec provided by the user
    outv = VideoWriter(fname, oheight, owidth, orig_vreader.frame_rate,
        codec=codec, format=format)
    for k in orig: outv.append(k)
    outv.close()

    # reload from saved file
    encoded = VideoReader(fname)
    reloaded = encoded.load()

    # test number of frames is correct
    assert len(orig) == len(encoded), "original length %d != %d encoded" % (len(orig), len(encoded))
    assert len(orig) == len(reloaded), "original length %d != %d reloaded" % (len(orig), len(reloaded))

    # test distortion patterns (quick sequential check)
    dist = []
    for k, of in enumerate(orig):
      dist.append(abs(of.astype('float64')-reloaded[k].astype('float64')).mean())
    assert max(dist) <= maxdist, "max(distortion) %g > %g allowed" % (max(dist), maxdist)

    # make sure that the encoded frame rate is not off by a big amount
    assert abs(orig_vreader.frame_rate - encoded.frame_rate) <= (1.0/MAXLENTH), "original video framerate %g differs from encoded %g by more than %g" % (encoded.frame_rate, framerate, 1.0/MAXLENTH)

  finally:

    if os.path.exists(fname): os.unlink(fname)

def test_user_video():

  # distortion patterns for specific codecs
  distortions = dict(
      # we require high standards by default
      default    = 1.5,

      # high-quality encoders
      zlib       = 0.0,
      ffv1       = 1.7,
      vp8        = 2.7,
      libvpx     = 2.7,
      h264       = 2.5,
      libx264    = 2.5,
      theora     = 2.0,
      libtheora  = 2.0,
      mpeg4      = 2.3,

      # older, but still good quality encoders
      mjpeg      = 1.8,
      mpegvideo  = 2.3,
      mpeg2video = 2.3,
      mpeg1video = 2.3,

      # low quality encoders - avoid using - available for compatibility
      wmv2       = 2.3,
      wmv1       = 2.3,
      msmpeg4    = 2.3,
      msmpeg4v2  = 2.3,
      )

  from .version import externals
  if externals['FFmpeg']['ffmpeg'] != 'unavailable':
    from .version import supported_videowriter_formats
    SUPPORTED = supported_videowriter_formats()
    for format in SUPPORTED:
      for codec in SUPPORTED[format]['supported_codecs']:
        check_user_video.description = "%s.test_user_video_format_%s_codec_%s" % (__name__, format, codec)
        distortion = distortions.get(codec, distortions['default'])
        yield check_user_video, format, codec, distortion