Reply to topic

Idea: YUV histogram

TheDiveO
Registered Member
Posts
595
Karma
3
OS

Idea: YUV histogram

Fri Jan 22, 2016 6:28 pm
I would like to ask/discuss the following idea here before filing a bug/feature report. I'm unsure as to whether this would need some new MLT support too.

We currently have the RGB parade/scope and the YUV scope. We also have the RGB histogram. But we don't have the YUV histogram, if I'm not mistaken. So, with the luma range discussion in the back of my head I would like to ask for a YUV scope, if possible? Would this make sense after all?
Inapickle
Registered Member
Posts
157
Karma
3

Re: Idea: YUV histogram

Sun Feb 07, 2016 8:07 pm
I expressed my views on this subject and the reasoning behind them in that thread TheDiveO was referring to, and and specifically post #48:

viewtopic.php?f=272&t=130641&p=351154#p351154

The first paragraph in that particular post is of course irrelevant.
Inapickle
Registered Member
Posts
157
Karma
3

Re: Idea: YUV histogram

Wed Feb 10, 2016 6:55 pm
In the course of that discussion, attention was drawn to the fact that both the Histogram and Luma WFM monitors appear to be displaying "limited" 16-235 ("TV levels") scale luma expanded out to full scale 0-255.

viewtopic.php?f=272&t=130641&start=45#p351206

As a case in in point, here are screen shots (posted in that thread) of the Y Luma Histogram and Luma produced from a 1080/29.97 YV12 source (in lossless UTVideo format) with the luma range intentionally pre-compressed to 32-220. No effects were applied. This is just the clip loaded on the timeline.

http://i.imgur.com/tPYQSj3.png
http://i.imgur.com/S3frr1X.png

Looking to the source codes to try and glean more insight:

https://quickgit.kde.org/?p=kdenlive.gi ... olorscopes

Although the code reveals that that the luma is being derived by Rec709 conversion from RGB (and not taken the source luma directly) it is not clear, at least to a layman like myself, how the programming results in (what should surely be) limited 16-235 range luma being presented as full scale. In other words, are full scale Y values actually being derived directly from RGB, or is it simply that derived 16-235 scaled Y values are being presented as though 0-255?

I would be grateful if the developers could clarify what's happening there.

Thanks.

 
Reply to topic

Bookmarks



Who is online

Registered users: Baidu [Spider], bassoongb, bcooksley, Bing [Bot], claydoh, Google [Bot], ipwizard, musheghkhachatryan, robertotamburrino, Sogou [Bot], teleseryereplays