Whilst I am still not sure where this is going, I have done some follow up experiments.

It seems quite clear, from reading around the subject that it is not uncommon for video encoding to lead to colour changes. The ability of the user of any encoder to influence this process may also be limited or non-existent.

To illustrate the variables :-

Use a red colour board 247,32,48 in track 1, above a green colour board 60,125,52 in track 2.
Confirm the RGB values using colorpic or similar - on my timeline they are exactly the same.

Analyse the RGB values in the preview window with the same tool.

Using VMR7
red 255,75,68
green 78,137,68

Using VMR9
red 247,31,48
green 60,126,52

Immediately (on my system but maybe not on others??) it is both visually obvious, and analytically obvious, that using VMR7 as the preview renderer gives significantly different colour to the timeline, whereas using VMR9 as the preview renderer giving virtually the same RGB values as the originals.

Therefore, from a visual perspective, the choice of renderer may be crucial to the displayed preview colour of the edited video content - easy to spot with block colours, but not so easy with normal content.

Which colour does PD export? One would expect the timeline RGB values, irrespective of the preview values. But, and here is the key issue, once exported how can we tell?

Presumably, unless we can analyse the file metadata, we have to play the produced video.

I exported the two colour board timeline in two resolutions for both .mp4 and .m2ts formats. If the produced file is played in different media players, there is a marked visual and analytical difference in RGB values, depending on the player.

The attached file "mp4 comparison" visually illustrates the difference, with the produced video @640 on the left and @1080 on the right.

The players are, from left to right, MPCHC, VLC, WMP and QT.

The screen shot should not be used for analysis as it is different to the original display but VLC is the only player that is virtually correct to the original timeline RGB values.

Similarly the file "m2ts comparison" produced @720x480 and @1080 also shows marked differences. The format and layouts are the same but this time QT is not available to be used. Again VLC is the closest player to the original.

Perhaps other players will display different results again?

So do we have a reliable and consistent methodology that allows us to analyse the colour encoding performance of PD - perhaps yes if we keep all the variables consistent on our own systems, probably no for any other system with differing variables.

If we cannot control the viewing platform, it seems unlikely that we can rely on the colour presentation being as we produce it on our own system, irrespective of PDs performance or file format or preview renderer or raw content or video card etc etc.

I suspect the best we can do is make the best shot at it for our particular purposes, however that is achieved??

Cheers
Adrian









Life is really simple, but we insist on making it complicated. (see below)
Confucius
AMD Phenom IIX6 1055T, win10, 5 internal drives, 7 usb drives, struggling power supply.

It seems this isn't a problem with PowerDirector, or at least, not a problem unique to PowerDirector.
I exported the image from Adobe After Effects as a video and got the same results, the colours were the same in SD export, but different in HD export, in the same way as PowerDirector.
What makes AE different is that it has a manual custom frame size for exported video (Which as far as I know PD doesn't have), so I could make videos of any size between 576p and 720p to see where the problem begins. It turns out it is 576, any value below that will give correct colours, any value even a pixel above will give altered colours.

AE's manual frame size also lets you break aspect ratio, and it seems that it is only the vertical frame size that causes the problem. The same video with the frame width 1500 x 560 produces the correct colours, but the frame width of 720 x 600 produces the altered colours. The horizontal frame size makes no difference to whether or not the effect is triggered.

Just to confirm that the variables between PC environments & different media players makes this nearly impossible to test.

I replicated Adrian's test (above) fully expecting to see very similar results here, using the same original colours. Not so.

1. There was no visual or analytical difference between preview displays with the video renderer set to VMR7 or VMR9.
2. There was quite consistent colour display between MPC-HC, VLC & WMP in both 480p & 1080p. Splash Pro was more "off colour"
3. Colours were closer to the original in 480p than 1080p



I won't bore you with results from AE, which were slightly different again.

Cheers - Tony

Visit PDtoots. PowerDirector Tutorials, tips, free resources & more. Subscribe!
Full linked Tutorial Catalog PDtoots happily supports fellow PowerDirector users!