(2019-08-23, 19:12)popcornmix Wrote: (2019-08-23, 18:40)krobolbus Wrote: Yep, it does work well on default settings and everything looks right with regards to black levels and with no visible banding, but the WTW and BTB are clipped (unless I'm misunderstanding something), so I just wonder whether this can be prevented.
If the issue you are describing is to do with values 0-15 and 236-255 with a limited range file then those values are illegal (although can be encoded) and behaviour is undefined.
Sorry @
popcornmix that's not the case.
In broadcast video terms the <16 and >235 can be present, and if they exist they shouldn't be clipped, as they are there to allow for a number of cases. One of the main ones is transients - particularly in imperfect legacy analogue content, where high-frequency transitions can undershoot or overshoot. If you clip these <16 undershoot and >235 overshoot transients you can introduce ringing when processing the video downstream. Most broadcast specs also allow for >100% luminance levels too. They are there for a very good reason, and if present, they should be preserved.
Bottom line - illegal levels aren't always illegal... (And in broadcast specs terms not all levels <16 and >235 are deemed out-of-gamut in the first place - 5-246 is the range that is within gamut)
Quote:Note: BTB and WTW don't make sense. It's like the volume control that goes up to 11.
They make very good sense in certain situations, which is why the specs include support for them. There are good reasons that broadcast video isn't based around 0-255 (or 1-254) level space.
The UK Broadcast technical standards delivery specs (aka DPP) state this :
Quote:1.2. Signal Parameters
In a video signal, each primary component should lie between 0 and 100% of the video range between black level and the peak level (R, G and B). Ideally, video levels should lie within the specified limits so that programmes can be distributed without adjustment.
When television signals are manipulated in YUV form, it is possible to produce "illegal" combinations that, when de-matrixed, would produce R, G or B signals outside the range 0% to 100%.
1.2.1.Video Level Tolerance
In practice, it is difficult to avoid generating signals slightly out of range, and it is considered reasonable to allow a small tolerance:
• the RGB components and the corresponding Luminance (Y) signal, should not normally exceed the “Preferred Minimum/Maximum” range of digital sample levels in the table below,
• measuring equipment should indicate an “Out-of-Gamut” occurrence only after the error exceeds 1% of an integrated area of the active image.
For further details see the EBU Recommendation, EBU R103.
Any signals outside the “Preferred Minimum/Maximum” range are described as having a gamut error (or as being out of gamut). Signals cannot exceed the “Total Video Signal Range” and will therefore be clipped.
8-bit video - Expected range = 16-235, preferred Min/Max range = 5-246, Total video signal range = 1-254
10-but video - Expected range = 64-940, preferred Min/Max range = 20-984, Total video signal range = 4-1019
Full range video levels must not be used for delivered television programmes.
Colour gamut "legalisers" should be used with caution as they may create artefacts in the picture that are more disturbing than the gamut errors they are attempting to correct. It is advisable not to “legalise” video signals before all signal processing has been carried out.
Even 'out of gamut' warnings allow 1% out-of-gamut content to pass through (transients particularly) without alarming usually (as per spec above)
Quote:If the TV could really display a blacker black and a whiter white than the maximum legal values, wouldn't it?
Yes - but video signals aren't just there to feed tellies - they pass through a number of points in the processing chain before then. If <16 and >235 levels are clipped in the chain - you can introduce artefacts. The display can clip <16 and >235 levels (or it can display them if required to cope with slightly out of spec signals) but a device upstream of the display shouldn't.
One very significant use of <16 and >235 levels is the standard PLUGE picture line-up signal used to calibrate displays, where sub-black is used to ensure black level is correctly adjusted on a display. If you clip <16 you destroy one of the useful bits of the PLUGE signal.
The <16 and >235 issue is a bit like the 702 vs 720 line width issue. Many think 720x576 is a 4:3 or 16:9 picture format, but actually (in 50Hz land) the central 702x576 area is 4:3 or 16:9. The 9 samples either side are 'wider' than 4:3 or 16:9. Just as with <16 and >235 the 9 samples each side should be preserved through the SD chain, but on a 16:9 or 4:3 display only the central 702x576 image should fill the frame (and you should not see the 9 samples either side), and if downconverting 1920x1080 or 1280x720 16:9 to SD the 1920x1080 or 1280x720 image should be scaled to 702x576 not 720x576.
https://tech.ebu.ch/docs/r/r103.pdf is very similar to the UK DPP spec in this regard, and UK DPP is based on it.
