Tweet
Excuse my bad English.
I am an electronic engineer and I participated in the development of the ITU-R BT standards. Edius has always satisfy the full specs buts there is a clipping issue in this filter. I ask to correct this damageable situation.
Because many reader dont know works the math of YUV (and it's digital YCbCr sub part), I do first a very little introduction.
If you think that YUV has the same gammut and color depth than sRGB because they have the same bit depth, you are not right. YUV encodes colors in the same way as the LAB, XYZ and CIE 1931 color spaces with specific coeficients and complicated color transform math. I go not in the details but you must know that the YUV color space is wider than the sRGB color space and can even code for negative RGB value (see https://en.wikipedia.org/wiki/XvYCC, a good but incomplete introduction). Furthermore, even if you respec the Zebra at 100% to shoot in 16-235 range for Y, there are peeks in the bright saturated colors wich have RGB color values far above de max. 255 range. For Ex. my Sony has shoot Rec.709 YCbCr=244, 63, 207 (bright red part in a lava flow) = RGB 365.63!!, 219.75, 126.08. I have very often G or B=290. This values are above 1023 in the Videoscopes.
Edius has a full YUV and filter core so that there is never clipping of the wider YUV color space. Furthermore, Edius is fully comptible with the xvYCC color space. There is thus never clipping of the datas (even if you shoot at 109 IRE) to the lower gammut sRGB space limited to 255/1023 or even to 235/960 like by other NLE's. This is OK.
But the Primary Color Correction filter does no work in the good way. This filter that seems to work in RGB instead of YUV and clips the colors at 255 even if you don't apply a LUT and that all the cursors are to the default. I ask to update this filter to work in full range YUV with no clipping and to accept Shaper LUT's. Another way is to add an negative gain input cursor and a output cursor to gain the signal back up to his original level. This workaround will only produce little banding artifacts in a 8 bit project and can compromise real time but it's not as bad as clipping. At the moment, this filter don't satisfy the ITU-R BT wich says: "the signal under 0 and above 1 may never be clipped during production, mastering nor in the intermediate files". The dramatic consequences of clipping are for ex.: hue shift, greenish sky, not possible to decently import REC.709 in the wider Rec.2020 gammut or in HDR because the bright saturated Rec.709 colors where sadly clipped before the conversion.
I ask also to extend the upper scale of the Scopes to 1535 so that you can immediately see what is above 255/1023 like in the previous Edius Scopes.
TIP: the full unclipped YUV range is send to the external monitor trough the Intensity Pro 4K. The Panasonic, Philips and probably the Sony monitors and TV's display the above 255 RGB colors without clipping. You see thus exactly what you have shoot. This is not the case with almost all the other NLE's, even Resolve. Switch never to another NLE!
Computer: Intel i7 8700K with Intel Graphic 630 + Blackmagic Intensity Pro 4K
I am an electronic engineer and I participated in the development of the ITU-R BT standards. Edius has always satisfy the full specs buts there is a clipping issue in this filter. I ask to correct this damageable situation.
Because many reader dont know works the math of YUV (and it's digital YCbCr sub part), I do first a very little introduction.
If you think that YUV has the same gammut and color depth than sRGB because they have the same bit depth, you are not right. YUV encodes colors in the same way as the LAB, XYZ and CIE 1931 color spaces with specific coeficients and complicated color transform math. I go not in the details but you must know that the YUV color space is wider than the sRGB color space and can even code for negative RGB value (see https://en.wikipedia.org/wiki/XvYCC, a good but incomplete introduction). Furthermore, even if you respec the Zebra at 100% to shoot in 16-235 range for Y, there are peeks in the bright saturated colors wich have RGB color values far above de max. 255 range. For Ex. my Sony has shoot Rec.709 YCbCr=244, 63, 207 (bright red part in a lava flow) = RGB 365.63!!, 219.75, 126.08. I have very often G or B=290. This values are above 1023 in the Videoscopes.
Edius has a full YUV and filter core so that there is never clipping of the wider YUV color space. Furthermore, Edius is fully comptible with the xvYCC color space. There is thus never clipping of the datas (even if you shoot at 109 IRE) to the lower gammut sRGB space limited to 255/1023 or even to 235/960 like by other NLE's. This is OK.
But the Primary Color Correction filter does no work in the good way. This filter that seems to work in RGB instead of YUV and clips the colors at 255 even if you don't apply a LUT and that all the cursors are to the default. I ask to update this filter to work in full range YUV with no clipping and to accept Shaper LUT's. Another way is to add an negative gain input cursor and a output cursor to gain the signal back up to his original level. This workaround will only produce little banding artifacts in a 8 bit project and can compromise real time but it's not as bad as clipping. At the moment, this filter don't satisfy the ITU-R BT wich says: "the signal under 0 and above 1 may never be clipped during production, mastering nor in the intermediate files". The dramatic consequences of clipping are for ex.: hue shift, greenish sky, not possible to decently import REC.709 in the wider Rec.2020 gammut or in HDR because the bright saturated Rec.709 colors where sadly clipped before the conversion.
I ask also to extend the upper scale of the Scopes to 1535 so that you can immediately see what is above 255/1023 like in the previous Edius Scopes.
TIP: the full unclipped YUV range is send to the external monitor trough the Intensity Pro 4K. The Panasonic, Philips and probably the Sony monitors and TV's display the above 255 RGB colors without clipping. You see thus exactly what you have shoot. This is not the case with almost all the other NLE's, even Resolve. Switch never to another NLE!
Computer: Intel i7 8700K with Intel Graphic 630 + Blackmagic Intensity Pro 4K
Comment