Tone Mapping Video

wswartzendruber · 2nd November 2020, 16:01

I've implemented the EETF described in BT.2390. What I'm surprised to see is that I can apparently tone map R', G', and B' independently? Won't that mess up hue? The spec lists it as a viable option.

videoh · 2nd November 2020, 16:50

There is a lot of practical experience involved in choosing tone mapping the luma versus the individual RGB channels. Each has its advantages and disadvantages. Generally, tone mapping the individual values produces only very minor hue/saturation changes, mostly seen as reduction of saturation. Tone mapping just the luma tends to produce oversaturation. Search around on google and you'll find a lot of material about this. One well-known analyst said that the effects of per channel mapping is a feature, not a bug.

Try it both ways and see what you prefer.

wswartzendruber · 19th December 2020, 21:58

My current implementation tone maps the R', G', and B' channels against MaxCLL (weighted luminance), but this affords the possibility of individual channels getting clipped. Now I am modifying it so that instead of MaxCLL, the highest linear ratio encountered in all channels is used as the maximum, regardless of whether it is encountered in R, G, or B. Now there can't be any hard clipping.

Does this make sense? My brain is somewhat fatigued of this.

2nd November 2020, 16:01	#1 \| Link
wswartzendruber hlg-tools Maintainer Join Date: Feb 2008 Posts: 413	Tone Mapping Video I've implemented the EETF described in BT.2390. What I'm surprised to see is that I can apparently tone map R', G', and B' independently? Won't that mess up hue? The spec lists it as a viable option.

2nd November 2020, 16:50	#2 \| Link
videoh Useful n00b Join Date: Jul 2014 Posts: 1,667	There is a lot of practical experience involved in choosing tone mapping the luma versus the individual RGB channels. Each has its advantages and disadvantages. Generally, tone mapping the individual values produces only very minor hue/saturation changes, mostly seen as reduction of saturation. Tone mapping just the luma tends to produce oversaturation. Search around on google and you'll find a lot of material about this. One well-known analyst said that the effects of per channel mapping is a feature, not a bug. Try it both ways and see what you prefer. Last edited by videoh; 2nd November 2020 at 16:52.

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

19th December 2020, 21:58	#3 \| Link
wswartzendruber hlg-tools Maintainer Join Date: Feb 2008 Posts: 413	My current implementation tone maps the R', G', and B' channels against MaxCLL (weighted luminance), but this affords the possibility of individual channels getting clipped. Now I am modifying it so that instead of MaxCLL, the highest linear ratio encountered in all channels is used as the maximum, regardless of whether it is encountered in R, G, or B. Now there can't be any hard clipping. Does this make sense? My brain is somewhat fatigued of this.