0002534: Histogram and RGB Parade

Notes
(0007918) j-b-m (administrator) 2012-03-05 11:29	Oh, yes there seems to be a problem... will look into it.

(0007919) j-b-m (administrator) 2012-03-06 14:59	Should be fixed in current git, probably available tomorrow in sunab's experimental PPA

(0007920) ansztyk (reporter) 2012-03-09 14:52	Thank you. So, I'm waiting for new update PPA.

(0007921) yellow (reporter) 2012-03-10 11:20 edited on: 2012-03-10 12:41	I can confirm that histogram looks great and compare near identically to controlled output from Avisynth, within a few values of each other on each channel, for the same files. Although this is not part of this bug report and perhaps i should open one, but for sometime I think the Waveform scope is raising levels by 10 8bit levels, so something at luma 16 is showing at 26, the top end is correct. I've tested this with a simple 0 - 255 test patten with 16 & 235 areas. http://www.yellowspace.webspace.virginmedia.com/fullrangetest.zip [^] EDIT On further investigation the waveform is giving identical results to Avisynths ColorYUV(analyze=true) for 'Loose' min and max. That appears to be filtering out any bright or dark 'noise', or what it thinks is noise, rather than the 'real' min and max. But even my lossless h264 full range test file with 16 & 235 areas, the 16 is getting reported by the waveform as 26. If you find time to look at the waveform would it be possible to add a Min and Max value to the Waveform scope like the RGB Parade has, especially if the waveform is going to do some filtering of 'real' levels. EDIT Above I was testing sunabs svn PPA build, but having just built from source using the build script, something isn't right with the latest code done with the build script vs sunabs earlier build. The Canon h264AVC in a MOV container, shows the histogram and RGB Parade combing / stretching which they didn't show in sunabs build. This appears specific to the Canon MOV's which are signaled as full range luma and ffmpeg reads them as yuvj420 instead of just yuv420. Disabling the full range flag in the source file by doing a quick remux with MP4Box which doesn't alter levels just changes file header meta data, so ffmpeg reads it as yuv420 and then forcing full range in the clip properties is the 'correct' result and the histogram in both builds displays correctly no combing so something is not quite right with latest histogram fix. I think it's something to do with the latest commit about swapping blue and red over?

(0007940) ansztyk (reporter) 2012-03-21 15:29 edited on: 2012-03-22 09:43	I checked a new app and scopes seem to be good.

Issue History
Date Modified	Username	Field	Change
2012-02-29 21:31	ansztyk	New Issue
2012-03-05 11:29	j-b-m	Note Added: 0007918
2012-03-05 11:29	j-b-m	Assigned To	=> j-b-m
2012-03-05 11:29	j-b-m	Status	new => feedback
2012-03-05 11:29	j-b-m	Status	feedback => acknowledged
2012-03-06 14:59	j-b-m	Note Added: 0007919
2012-03-09 14:52	ansztyk	Note Added: 0007920
2012-03-10 11:20	yellow	Note Added: 0007921
2012-03-10 12:05	yellow	Note Edited: 0007921	View Revisions
2012-03-10 12:06	yellow	Note Edited: 0007921	View Revisions
2012-03-10 12:36	yellow	Note Edited: 0007921	View Revisions
2012-03-10 12:37	yellow	Note Edited: 0007921	View Revisions
2012-03-10 12:39	yellow	Note Edited: 0007921	View Revisions
2012-03-10 12:39	yellow	Note Edited: 0007921	View Revisions
2012-03-10 12:40	yellow	Note Edited: 0007921	View Revisions
2012-03-10 12:41	yellow	Note Edited: 0007921	View Revisions
2012-03-21 15:29	ansztyk	Note Added: 0007940
2012-03-21 20:17	ansztyk	Note Added: 0007941
2012-03-22 09:43	ansztyk	Note Deleted: 0007941
2012-03-22 09:43	ansztyk	Note Edited: 0007940	View Revisions

Relationships