PR: Fix and improve "Sony S-Log" curves implementation errors and consistency.

[KelSolaar]

References #336.

[coveralls]

Coverage increased (+0.02%) to 98.197% when pulling 34cdc5b on feature/sony_slog into 124461c on develop.

[nick-shaw]

For consistency, I think log_encoding_SLog3 and log_decoding_SLog3 should be updated to include the bit_depth, in/out_legal, and in/out_reflection kwargs that the other two have.

It won't work quite the same as the other two. For example, for log_encoding_SLog3, out_legal=True could just return the current y value. out_legal=False would need to return:
CV_to_IRE(y * (2**bit_depth - 1), bit_depth, True)/100

This is needed to make all three S-Logs able to return the values in the Table in the S-Log3 white paper, because the IRE column is based on out_legal=False, and the CV column is based on out_legal=True.

Make sense?

I would also argue for the default to be out_legal=True rather than the current False. This will produce float values compatible with the ACES IDTs.

[coveralls]

Coverage decreased (-0.03%) to 98.155% when pulling 4e5f345 on feature/sony_slog into 124461c on develop.

[coveralls]

Coverage increased (+0.02%) to 98.198% when pulling 4e5f345 on feature/sony_slog into 124461c on develop.

[nick-shaw]

The function RGB_range in ycbcr.py has been replaced with the new CV_range function. However there are still references to RGB_range in the docstrings.

[coveralls]

Coverage increased (+0.02%) to 98.198% when pulling 609166e on feature/sony_slog into 124461c on develop.

[coveralls]

Coverage increased (+0.02%) to 98.198% when pulling 850df560d8d4b25eeedd77ee6e7ea266ab84d4ed on feature/sony_slog into 124461c on develop.

[coveralls]

Coverage increased (+0.02%) to 98.198% when pulling bccdbd84b7be8c5b537999e41288160561cf208c on feature/sony_slog into 124461c on develop.

[coveralls]

Coverage increased (+0.02%) to 98.198% when pulling aba669a on feature/sony_slog into 124461c on develop.

[KelSolaar]

@nick-shaw : I have implemented the requested changes. The following snippet matches the table in [1]:

from colour import log_encoding_SLog2, log_encoding_SLog, log_encoding_SLog3

for x in [0.0, 0.18, 0.9]:
    print( x, "S-Log3 IRE", 100 * log_encoding_SLog3(x, bit_depth=10, out_legal=False))
    print( x, "S-Log2 IRE", 100 * log_encoding_SLog2(x, bit_depth=10, out_legal=False))
    print( x, "S-Log IRE", 100 * log_encoding_SLog(x, bit_depth=10, out_legal=False))
    print( x, "S-Log3 CV", 1023 * log_encoding_SLog3(x, bit_depth=10, out_legal=True))
    print( x, "S-Log2 CV", 1023 * log_encoding_SLog2(x, bit_depth=10, out_legal=True))
    print( x, "S-Log CV", 1023 * log_encoding_SLog(x, bit_depth=10, out_legal=True) )

0.0 S-Log3 IRE 3.53881278539
0.0 S-Log2 IRE 3.00012228519
0.0 S-Log IRE 3.00012228519
0.0 S-Log3 CV 95.0
0.0 S-Log2 CV 90.2810712183
0.0 S-Log CV 90.2810712183
0.18 S-Log3 IRE 40.6392694064
0.18 S-Log2 IRE 32.3449512215
0.18 S-Log IRE 37.6512722255
0.18 S-Log3 CV 420.0
0.18 S-Log2 CV 347.3417727
0.18 S-Log CV 393.825144695
0.9 S-Log3 IRE 60.9469471967
0.9 S-Log2 IRE 59.1365542288
0.9 S-Log IRE 65.3529251225
0.9 S-Log3 CV 597.895257443
0.9 S-Log2 CV 582.036215044
0.9 S-Log CV 636.491624073

References

1. Sony Corporation. (n.d.). Technical Summary for S-Gamut3.Cine/S-Log3 and S-Gamut3/S-Log3. Retrieved from http://community.sony.com/sony/attachments/sony/large-sensor-camera-F5-F55/12359/2/TechnicalSummary_for_S-Gamut3Cine_S-Gamut3_S-Log3_V1_00.pdf

[coveralls]

Coverage decreased (-0.03%) to 98.156% when pulling 634348b055321f0d00f62ead9fcf1a5c0d4b36cc on feature/sony_slog into 124461c on develop.

[coveralls]

Coverage increased (+0.02%) to 98.2% when pulling 2af7b71 on feature/sony_slog into 124461c on develop.

[nick-shaw]

That all looks good to me now.

[KelSolaar]

Thanks @nick-shaw! :)