Video Test Clips in h264 (x264) format

[Roco]

Hi, I'm very interested to find out if most PC systems can playback h.264 content produced and encoded at full luminance range, for maximum quality. For this reason I made some video test clips to check the black and white levels and gamma at h.264 playback.

These clips can be used in many other cases as well. I hope you'll find them usefull for your needs and I will really appreciate any feedback to help me make the correct decisions. For more information and for the download links, please visit the following page I made just for this purpose.

http://www.geocities.com/anothrosko/vtc.html



P.S If this is not the correct place to start this thread, sorry, the administrator may move it as he wishes.

[SeeMoreDigital]

Is there any particular reason why your BLACK_LEVEL_TEST_CLIP.mp4 and WHITE_LEVEL_TEST_CLIP.mp4 samples were encoded at 1920x1200 pixels and not 1920x1080 pixels?

[sheppaul]

Wow, really nice.

Thank you.

[Roco]

Is there any particular reason why your BLACK_LEVEL_TEST_CLIP.mp4 and WHITE_LEVEL_TEST_CLIP.mp4 samples were encoded at 1920x1200 pixels and not 1920x1080 pixels?

I think either way they could be used, but I choosed this resolution to fill the whole screen without leaving borders in wide-screen (16:10) monitors. Also, it might be a test for some decoders and players, to find out if they can support h.264 video in WUXGA resolution (which is outside the HD standards), because it happens to be the resolution of the content I'm referring to (but at a high frame rate). For external displays with aspect ratio 16:9 (or any other AR) they can be used with vertical scaling or cropping at playback.

[Roco]

Just changed the Gamma test clip a bit, so that it can be used on 720p displays too (because this clip must remain at actual size).

[Roco]

If you have tested your system with these free clips -or you 'd like to do so, I'd appreciate very much if you participate on a simple poll http://www.geocities.com/anothrosko/vtc.html -it's just one click.

Thanx! :)

[gngn]

after running your test-files i'm even more confused. everybody in the MPC-HC thread says that using a Nvidia card and as output VMR7/9 will result in washed out colors. this is true, so i have to use a option in the Nvidiacontrolpanel and enable Full 0-255 chroma range. then i will have "true" black and white. in the following screenshot above is with Nvidia option enabled, below without it. it's clear that below white and black is actually different grey shades.

http://i266.photobucket.com/albums/ii259/gngnmumu/dxva/pctvlevels.png
file used here (http://rapidshare.com/files/135843758/vmr9-test.avi). is it bad encoded?

but when i run your testfiles i get all levels only if i uncheck Nvidia's option of "Full-range". if i enable Nvidia's option to get the correct colors i get clipping in your testfiles, higher levels are gone in the White-test, lower levels are gone in the Black test. my Gamma-test give's me arround 2,6 (CRT LG Flatron ez)

[Roco]

after running your test-files i'm even more confused. everybody in the MPC-HC thread says that using a Nvidia card and as output VMR7/9 will result in washed out colors. this is true, so i have to use a option in the Nvidiacontrolpanel and enable Full 0-255 chroma range. then i will have "true" black and white. in the following screenshot above is with Nvidia option enabled, below without it. it's clear that below white and black is actually different grey shades.

http://i266.photobucket.com/albums/ii259/gngnmumu/dxva/pctvlevels.png
file used here (http://rapidshare.com/files/135843758/vmr9-test.avi). is it bad encoded?

but when i run your testfiles i get all levels only if i uncheck Nvidia's option of "Full-range". if i enable Nvidia's option to get the correct colors i get clipping in your testfiles, higher levels are gone in the White-test, lower levels are gone in the Black test. my Gamma-test give's me arround 2,6 (CRT LG Flatron ez)


What's happening in your case is obvious, since the clip you uploaded is encoded in standard video range 16-235.
When you enable "Full-range" it does a TV->PC levels conversion from 16-235 to 0-255 so you see the "true" black and white in your monitor. On the other side, my clips are meant to test not only the standard but the full range too, so they are encoded in full range. Leaving the option "Full-range" on, it scales up the full range clips too and that results in the clipping you see.

[gngn]

is there a way to test the color settings used in a video? problem is that all videos i have show this "washed out colors" so i could assume that everybody is encoding with TV levels?
as for the poll my answer is "Windows 0-255"

[Roco]

All videos are encoded in tv-levels indeed, this is natural -although useless while lowering the quality in most cases IMO in today's all-digital world.
There is avisynth's histogram function to check the levels of a video. Also some NLE video editing programs have a "waveform monitor" for the levels and a "vectorscope" for the chrominance. You can also grab a frame and check it in a paint program.

BTW thanks for the feedback!

[CruNcher]

Roco for that case on Digital Equipment you should change it @ decoding level not forceing it @ encoding, think about the case everyone would encode his SD stuff in PC Scale and then watch it on a old TV he would get shocked and Analog PC2TV hmm sounds not as efficient (especialy easy todo) as Digital TV2PC (setting a option) ok a Standalone or HtPC could also do it but you just move the problem from left to right :) The Standard way is TV scale and changin it now would just be confusing and not really help.

[Roco]

Roco for that case on Digital Equipment you should change it @ decoding level not forceing it @ encoding, think about the case everyone would encode his SD stuff in PC Scale and then watch it on a old TV he would get shocked and Analog PC2TV hmm sounds not as efficient (especialy easy todo) as Digital TV2PC (setting a option) ok a Standalone or HtPC could also do it but you just move the problem from left to right :) The Standard way is TV scale and changin it now would just be confusing and not really help.

I don't recommend to anyone to start encoding his videos in full range. This is just my preference for content especially produced and encoded in full range, intended to be watched from PC users on their monitors -optionally on external displays. (Even someone with an external display not capable of full range, would be able to compress the range in realtime with his graphics adapter or his software player and watch it without problems -but most today's displays, projectors etc are capable of full range).

So, in this case CruNcher, if you wanted to purchase such a product would you prefer it in video range which would have 14% less dynamic range and more banding artifacts (= less quality), or in full range?