Probably what you're seeing is some light banding where the steps meet. I'll check it closer tonight. The way I leveled it, everything >235 should be wiped out (unless I made a mistake).

My preferred method for grabbing from VLC is full screen, scaling OFF, "Prt Sc" and paste in Irfanview. Then save as PNG at 0 compression. My older version of Photoshop prints extra png header information that is incompatible with web browsers (maps to cRGB).

btw, I understand that Handbrake's MKV flavor is fairly lossy. But worth testing at some point.

Youtube Player looks like it is putting a very light dithering pattern over its video. Maybe what you are seeing.

Note though if you are looking at a Youtube download on your system, it will show the full steps to 255, as we discovered before, because the upload was rendered in RGB space.

Here's a screengrab from YouTube in Firefox from 2:13 (the first "wrong") of the part from the 235 band to the 255 band at the top (it's near-white but you can right click and save as):



Here's it's histogram in 10.0c:



It's definitely banded. If you were expecting flat 255 then that's not what I'm seeing. Or maybe I've misunderstood.

On a different note, I noticed that the MeGUI preview window is expanding levels out to [0,255], even though it's being sent [16,235] and is rendering [16,235]. Which is misleading. I've changed the tutorial slightly to reflect this, and asked the MeGUI folk what's going on.

"If you were expecting flat 255 then that's not what I'm seeing. Or maybe I've misunderstood."
You've not misunderstood. I expected a flat 255 palette returned from >=235 RGB uploaded to Youtube.

I can see all six bands in your image. So it's not to do with my screen. They eyedropper at 234, 238, 242, 246, 250, and 254 RGB in Photoshop.

And here's my grab from the Youtube player at 2:13 at 720p fullscreen in Firefox.
The only band I can see is a remnant of 235, although there's "something" hanging around at 251 (maybe the boundary, or dithering).





Are you seeing the same blown out facial highlights and blocked blacks in the hair on Youtube as I am? That was my intention (I sent YT full 0-255 in the first "wrong" example.



Here is what was uploaded, matching very closely what you posted:
I made the stepwedge in Photoshop, and the eyedropper confirms the levels are RGB 235-255 inclusive in increments of 4.



So why are we seeing different things at 2:13? The fadeout starts at 2:14.14. Full levels are from 2:09 to 2:14.
Anyway, this version will be deleted and the faac one will be uploaded.

Are you seeing the same thing in the latest Vimeo version?
http://vimeo.com/20627058

Let's compare notes then, but maybe in the Handbrake thread so as not to clutter up this equally important MeGUI discussion.

Nick,
What version of Shockwave Flash are you running?
It's under Tools->Addons->Plugins in Firefox.
I've got 10.1.102.64

Well wouldn't ya know it, the Flash Player completely changed its behaviour and the new version 10.2 DOES NOT EXPAND [16,235]>[0,255]. i.e. No clipping.

Screen grabs from 720p/full-screen from 2:12 and 2:42 from

YouTube 2:12 Flash Player 10.1.82.76
YouTube 2:12 Flash Player 10.2.152.32
YouTube 2:42 Flash Player 10.1.82.76
YouTube 2:42 Flash Player 10.2.152.32

Adobe techs stalking this forum perhaps???

Conclusion: Assume NOTHING.

Various Flash Player versions available for testing here. Please verify what I'm seeing in case I have finally gone insane.

Didn't compare Vimeo yet. Will do that now.

Also Mark, can you please look at this reply to a question I posted on doom9 and see if it tells us anything we don't know? I'm struggling with some of the details.

Well, that pretty well mucks up both our tutorials, and all the testing I've done for the past two years. So what do we recommend now? To give Flash delivery 16-235, or 0-255? What's in the majority? How long will it stay that way? What's the safest method? Will Adobe change their minds again at some point in the future?

If any Adobe devs are lurking on this forum, I with they'd make themselves known, and tell us if this change was intentional or an oversight, and why?

Would they presume that we should upload two versions of each video to Youtube from now on? And that people would know their Flash version well enough to be able to pick the right one? Or maybe we should just encode everything at 8-245 and forget the whole thing.

Feeling a bit perplexed right now, going to bed and will d/l the latest Flash tomorrow. I've had quite enough midlife crises for one winter (in the northern hemisphere).

WRT the doom9 response, I don't quite get all that's being said, but I am certain that REC 601 and REC 709 luminance levels are essentially the same.

As far as the rest of the response, I'm going to have to digest a bit further.

I actually admire your and johnmeyer's tenacity in posting on that forum. I'm a bit of a coward when it comes to sadomasochism.

I don't care if I appear ignorant or if they're short with me at doom9. I just want the info and between them they've usually got it.

Yes, 601 and 709 levels are essentially the same regarding luminance. The colour balance comes out a bit different, that's all. I was playing with them yesterday.

Anyway, to mix metaphors, I've dug deeper and the plot thickens...

Vimeo and Facebook are both behaving the same in Flash Player 10.2 as they were in 10.1.

So actually it's not a blanket change in the Flash Player at all but rather either:

a) A recent change in the colour space instructions that YouTube is sending in their Flash/HTML,

or

b) The new Flash Player 10.2 has started respecting YouTube's existing colour space instructions correctly,

or

c) Adobe have done introduced specific new behaviour for YouTube.

I suspect (b) but can't be sure.

Any recent change that YouTube has made is apparently in their Flash/HTML, not in their H.264/mp4 encoding, because my uploaded REC709 gradient from 18th Feb ( and yours from yesterday are showing the same levels.

Incidentally I found no differences between FF and IE regarding Flash Player version behaviour. And Google Chrome's even later version of 10.2 is behaving the same way as 10.2 in FF and IE. So it's not a browser-specific issue.

So apparently levels jiggery pokery is no longer required for WYSIWYG from YouTube. One up for them over Vimeo and Facebook, I say. One up for YouTube uploaders too, as long as they don't change it back again. One down for us, as our tutorials now have to get more complicated.

As I said earlier, a verification of all this on someone else's computer would be reassuring.

As auto-update is the default state of the Flash Player and updating is really easy and fast (I did it myself the other day almost without thinking... hence this new discovery), penetration of new versions is rapid. You can see here how fast version 10.1 was adopted over 10.0. So I'm going to change my tutorial to say that squeezing the levels is for Vimeo/Facebook but now not for YouTube.

I hunted for documented evidence of a change in a Flash Player changelog and all I can find is this and this. It makes me slightly concerned that levels behaviour will vary according to the viewer's individual hardware/software setup.

It looks like we've succeeded in alienating just about all visitors to this thread (again, lol), so when we've fully got our heads around what's happening here, we might want to start a new thread to try and stop people merrily adding Computer-RGB-to-Studio-RGB filters to their YouTube projects when they should no longer be.

Digital video really is the new wild, wild west...

Nick, did you find anything interesting with your 1080i to 1080p tests?

Andy

So with 10.2 Flash,

At Vimeo the Studio RGB still looks right and Computer RGB is blown out?
http://vimeo.com/14151131

And at Youtube the Studio RGB looks flat and the Computer RGB now looks correct?


And with 10.1 we know Studio RGB is correct for both services.

Wonder about local players -- JWPlayer, Flash Player, VLC, SMPlayer, GOM Player, VLC Player, etc. etc. ?

Wonder is "someone" might have time to run another series of tests? (Cue Jerry stage right).

@ Andy - Yes, sorry for the delay. I did a shoot-out of a number of methods and will start a new thread on it soon. Just downloaded the BCC trial to complete the set. In brief, QTGMC is amazing, even at "ultra-fast". Yadif Vegas plugin is really good for the speed and convenience. I prefer it to the Mike Crash Smart Deinterlacer. Both can now be used as a track fx in Vegas 10.0 with pre-toggle (in 8.0 Smart Deint had to be Media fx). I'm keen to know if the Yadif plugin supports 64-bit Vegas.

@ Musicvid - Yes, that's the correct summary as far as I see it. Is that what you're seeing too? If so I'll start an "Important information..." thread for those that aren't looking here. The REC709 gradient in Jerry's JWPlayer test is still the same in Flash Player 10.2 as in 10.1. It's really looking like a YouTube-specific change.

musicvid said: "Wonder about local players -- JWPlayer, Flash Player, VLC, SMPlayer, GOM Player, VLC Player, etc. etc. ?I'm ready and willing! What I'd plan to do is take my existing page: http://www.jazzythedog.com/testing/DNxHD/mediaplayers.aspx and do a screen capture of your REC709 chart, plop that into Vegas & scope it. I have the latest Flash installed. Were you looking for something different? Also, the only Flash players included on that page are JW Player & FlowPlayer (as well as embedded YouTube & Vimeo). I'm going to have to do a little research into the other players you suggest - could take a couple of days.Nick Hope said: "The REC709 gradient in Jerry's JWPlayer test is still the same in Flash Player 10.2 as in 10.1"Looks like Nick has already done some of this testing, but I'll confirm on my computer.

...Jerry

"And at Youtube the Studio RGB looks flat and the Computer RGB now looks correct?Either I'm doing something terribly wrong, or I'm terribly confused or I'm not seeing this with Flashplayer 10.2.152.32 and IE9. Is the following screen capture too subtle to show that sRGB looks correct & cRGB looks blown out?



...Jerry

Do you guys think these different results may be the result of the new hardware acceleration in Flash 2.10?

In other words, does Flash work differently with different cards and source? Or are the results always the same whether HW acceleration is off or on in the Flash control panel?

This is what I see in Firefox with Flash Player 10,2,152,32 and Google Chrome with Flash Player 10,2,154,12.



However, in IE8 I'm seeing the same as you Jerry. i.e. blown out

So there are browser differences after all.

What a complete and utter head****. Deep breaths...

How does look to you in IE9? Is the left-most edge grey (16) or white (0)?

Do you have other browsers to compare Musicvid's test in?

It would be useful for someone to upload Musicvid's test again in case something has changed in YouTube's encoding since July 2010. But it will need modifying or they will reject it as a duplicate.

Right, I'm off to have color space nightmares.

Oh by the way, Jøran, if you're reading, the Yadif plugin developer has told me that the 64-bit version is for 64-bit Vegas and that he actually only tested it inside Vegas 64-bit. I'm hoping to post a new thread of deinterlacer comparisons tomorrow.

>> are the results always the same whether HW acceleration is off or on in the Flash control panel? <<

"Settings..." is greyed out for me in all 3 browsers. It was there in IE, with acceleration checked, but when I unchecked it, "Settings" became greyed out.

I will put a new grayscale test on both YT and Vimeo sometime tonight or in the morning,

That old one can be a tiny bit misleading because, although the 0, 16, 235, 255 levels are spot on, it used an old printer gamma test I made a decade ago, and is gamma 2.2 instead of 1.0.

The new one will be truly linear, size 1920x1080, and I will make the PSD available for those who want to play with it or incorporate into their own work.

I think Adobe got a few "unintended consequences" in the latest Flash version, perhaps a result of HW acceleration, and it may be altogether different in their next release, perhaps even stable again. There are certainly enough complaints about it in the forums, and many people are recommending to just turn HW acceleration off for now.

I'm on my laptop Display adapter is an ATI Mobility Radeon HD 545v. 512MB. Flash HW acceleration is "on".

...Jerry

btw: confirming what my eyes see. Here's the sRGB flipped on it's side:



Whoops! Detail is lost. Click here: http://dl.dropbox.com/u/20447760/IE9-sRGB-cRGB-Waveform.png

"How does White.



"Do you have other browsers to compare Musicvid's test in?"I see no difference in Firefox 3.6.13.

First w/ Flashplayer 10.1



Next w/ Flashplayer 10.2



Finally, I turned hardware acceleration off & see no difference.

Strange!! Are you having a bad Cosmic Ray storm across the Pacific? Or maybe it IS the Display Adapter??

...Jerry

This may be a red herring, but I'll post anyway.

When I posted earlier in this thread, I mentioned in passing that I wasn't sure the levels on my YouTube video were correct. What I didn't post is that I was getting all sorts of different results, depending on what I used to playback the MP4 file (i.e., when playing the file locally). I realize that we are talking about playback from an online source (YouTube, Vimeo, etc.) but what I found when using Vegas vs. using VLC and Media Player is that the overlay settings in your graphic card are used when playing from VLC and Media Player, but NOT when playing back from Vegas. Thus, you will get completely different results -- levels, colors, you name it -- depending on whether the video is being played through the Windows "overlay" function.

So -- and here is the reason for posting -- is it possible, depending on what browser plugins and settings are enabled, that the graphic card overlay setting (or some other setting in the browser) might be altering the levels and colors?

"So -- and here is the reason for posting -- is it possible, depending on what browser plugins and settings are enabled, that the graphic card overlay setting (or some other setting in the browser) might be altering the levels and colors?"Pardon me for being obtuse, but what is the "graphic card overlay setting" and where is it set? I have done a Spyder calibration on my laptop - is this calibration adjustment the "overlay"?

...Jerry