Vegas to Youtube, Vimeo, Web -- A New Look

Kommentare

NickHope schrieb am 30.01.2011 um 16:44 Uhr
>> my downloads from YT and Vimeo all show the expanded levels in the files themselves <<

So what exactly is that file you have in that demo, that represents a YouTube download? If I so much as click on a YouTube downloaded file in the Vegas 8.0c explorer window, it crashes, so assuming you're seeing the same behaviour, then it must be something you've converted the YouTube mp4 to. I would have assumed it's a screen grab still, except for the fact it looks like there's audio under it. My luma comparisons were done in 10.0c, which accepts the downloaded YT files fine.

>> turning off weightp (in Handbrake anyway) fixes garbled playback in all of our players as well as Youtube and Vegas <<

OK! It's staying off!

Vimeo upload is horribly slow this time I'm afraid. 1:37 remaining, plus the rendering wait. After this upload I'll probably also throw up the file I'm sending to them for you to take a look at if you like.
musicvid10 schrieb am 30.01.2011 um 16:52 Uhr
To open YT and Vimeo downloads in Vegas 8.0c, I do a straight-through MP4 remux in AVIDemux first. That does not affect levels or flags (tested it). Here is how Vegas sees the levels from the latest Vimeo download (the 36MB one). Upload was the same 8Mbs 16-235 version done in HB. The file gamma seems to be exactly 1.125 (not very much).

To summarize, the only differences between the downloaded files and fullscreen player grabs are the gamma bumps Bob and I discussed previously, which their native players seem to negate. Level endpoints are the same. I've usually done visual comparisons to known good stills as well. Like everyone else, I don't know why the gamma game exists, but maybe there is some miniscule streaming bandwidth advantage.



If you view one of the downloaded files in Vegas 10.0c with both scope options unchecked, what do you get? Do you get consistency between the Histogram and Waveform? Are you getting exactly the opposite results in 10.0c as in 8.0c? Are we talking about a difference in scoping between 10.0c and 8.0c? If so, let's start another thread and figure out what's going on. For this tutorial I don't think that there is a question about the need to upload 16-235, because the net result always seems to be 0-255, even with Jerry's local player as well as Youtube and Vimeo.

Vimeo upload and processing times have gotten horribly sucky as of late. They want me to upgrade to $10/month or $60/year to "fix" this.

Isn't it about 1 AM where you're at?
NickHope schrieb am 30.01.2011 um 18:07 Uhr
Funnily enough I downloaded AVIDemux a couple of weeks ago to make a YouTube video useable in Powerpoint. I did a remux of your file (from http://vimeo.com/18690771) with it and got the same result as you in Vegas 8.0c.

HOWEVER, in Vegas 10.0c, all the downloaded files are giving me this:



Confused?

Before we blame the scoping, could it be that the older MPEG-4 decoder in Vegas 8.0c is expanding the luma (and changing the gamma?), but the one in in 10.0c isn't?

>> Isn't it about 1 AM where you're at? <<

2am now, and my mind is just about blown.
musicvid10 schrieb am 30.01.2011 um 18:32 Uhr
WOW!

However, if this turns out to be a decoder or scopes or player thing rather than a processing thing, I may have a lot of old posts to go back and edit.

I see yet another long thread in our future, and we can only pray for a Glenn Chan visitation . . .
NickHope schrieb am 30.01.2011 um 19:32 Uhr
Nah, it's just that crappy old mp4 decoder in Vegas 8.0. Time to upgrade Mark! ;)

As for confused uploaders, I think they've probably all long since tuned out.

Actually, in all fairness, I'm liking Vegas 10.0 more and more and it's probably worth that discounted upgrade price I paid just as an extra tool for stuff like this and for the elastique audio stretching etc., even if I end up still using 8.0c as my main editor (didn't really push 10.0 yet).

I wonder how it was in Vegas 9.0. It's at times like this that I wish someone from SCS could quickly chime in and say yes, we changed the luma of decoded mp4 files, or otherwise. Oh well...

OK, so here's the link to my "TEST 1" on Vimeo, but they say they ain't gonna start rendering it for another couple of hours:



A copy of the file I uploaded to them is available [url=https://s3.amazonaws.com/bubblevision/YT-Vimeo-Nick-Test1-MeGUI-TDeint-CQ196.mp4]here (but I make take it off in the future). Apparently you can also grab the uploaded file from Vimeo for the first week so please do that if you can and save my bandwidth.

I will post the YouTube version tomorrow. Currently uploading.

Here's the recipe...

Programs Used:

Sony Vegas 10.0c
Debugmode Frameserver 2.7
AviSynth 2.5.8.5
Tdeint 1.1
MeGUI 1911
x264 1867
Nero AAC Encoder 1.5.4

AviSynth Script:

AviSource("d:\fs.avi")
ConvertToYV12(matrix="PC.709")
TDeint(order=1)
LanczosResize(1280,720)


MeGUI settings:

x264 command line:

program --profile main --level 3.1 --crf 19.6 --deblock -2:-1 --keyint 300 --min-keyint 29 --bframes 2 --b-pyramid none --no-weightb --ref 2 --weightp 0 --qpmin 3 --subme 6 --trellis 0 --no-mixed-refs --output "output" "input" 


Nero AAC audio, AAC-LC profile, Adaptive Bitrate @ 320 kbit/s

Render time:

34 mins on my core 2 duo T7800 @ 2600 MHz with 4GB RAM

Playback of uploaded file:

Vegas 8.0c: OK (but opens slowly and plays steppily)
Vegas 10.0c: OK
VLC 1.1.5: OK
GOM: OK
WMP 11: no video, only audio, but I think this is unique to my laptop.

Jerry, thanks for the FFDShow/Haali Media Splitter link. For some bizarre reason I have a K-Lite codec pack installed on this computer (yes, I know, I'm a very naughty boy), which includes FFDShow and Haali. I should probably attempt to uninstall first to see if that fixes things. Either that or a full reinstall of Windows, which is overdue.

So, let me know how this looks against the Handbrake version (and please don't hold back regarding criticism of the AviSynth TDeint against the Handbrake Decomb). Perhaps now, with everything else even, I can brave the doom9 forums and look at some other decombers to try in my AviSynth script. But next I'm going to get a Vegas 10 / MainConcept render online. 2-pass VBR with 8 Mbps average. Any thoughts on max bitrate? Since this MeGUI render has gone all the way up to 21 Mbps max, I'm thinking at least 16 Mbps (double the average).
musicvid10 schrieb am 30.01.2011 um 20:35 Uhr
First reaction to your 8Mbs upload version:
Static detail excellent, about the same, motion detail slightly soft (many people prefer this).
Unable to see any interlace artifacts.
Unable to see noticeable transition blocking.
Levels are the same (thankfully).
Playability as good or better than HB encode.

Get some sleep!
NickHope schrieb am 30.01.2011 um 21:01 Uhr
Yes dad, but before I go, here's the YouTube version of my "Test 1".



Goodnight!
amendegw schrieb am 30.01.2011 um 21:19 Uhr
In the YouTube video (at the 49sec mark), I see the same blocking as seen in musicvid's YouTube (see third post from the top). As explained, by musicvid, this is a YouTube processing problem.

Certainly not seen in Nick's Vimeo video.

...Jerry

System Model:     Alienware M18 R1
System:           Windows 11 Pro
Processor:        13th Gen Intel(R) Core(TM) i9-13980HX, 2200 Mhz, 24 Core(s), 32 Logical Processor(s)

Installed Memory: 64.0 GB
Display Adapter:  NVIDIA GeForce RTX 4090 Laptop GPU (16GB), Nvidia Studio Driver 566.14 Nov 2024
Overclock Off

Display:          1920x1200 240 hertz
Storage (8TB Total):
    OS Drive:       NVMe KIOXIA 4096GB
        Data Drive:     NVMe Samsung SSD 990 PRO 4TB
        Data Drive:     Glyph Blackbox Pro 14TB

Vegas Pro 22 Build 239

Cameras:
Canon R5 Mark II
Canon R3
Sony A9

musicvid10 schrieb am 30.01.2011 um 21:44 Uhr
I think the slight difference in motion blur is that Nick use HEX Motion Estimation (the x264 default), and I used UMH. Again, all a matter of personal preference. Sharper is not always better. Nice of x264 to publish all the settings in the encode.
amendegw schrieb am 30.01.2011 um 21:54 Uhr
As an aside, with the proliferation of this video clip, if my wife ever discovers it, she's going to insist on scale (she's the one feeding the peacock). {grin}

...Jerry

System Model:     Alienware M18 R1
System:           Windows 11 Pro
Processor:        13th Gen Intel(R) Core(TM) i9-13980HX, 2200 Mhz, 24 Core(s), 32 Logical Processor(s)

Installed Memory: 64.0 GB
Display Adapter:  NVIDIA GeForce RTX 4090 Laptop GPU (16GB), Nvidia Studio Driver 566.14 Nov 2024
Overclock Off

Display:          1920x1200 240 hertz
Storage (8TB Total):
    OS Drive:       NVMe KIOXIA 4096GB
        Data Drive:     NVMe Samsung SSD 990 PRO 4TB
        Data Drive:     Glyph Blackbox Pro 14TB

Vegas Pro 22 Build 239

Cameras:
Canon R5 Mark II
Canon R3
Sony A9

musicvid10 schrieb am 30.01.2011 um 21:57 Uhr
Heck, I want a cut from the poppy harvest!
NickHope schrieb am 30.01.2011 um 22:25 Uhr
Can't sleep. Head full of Mbps...

>> I think the slight difference in motion blur is that Nick use HEX Motion Estimation (the x264 default), and I used UMH. <<

I'll render a Handbrake file with as identical settings as possible to my MeGUI file so we can compare.

>> As an aside, with the proliferation of this video clip, if my wife ever discovers it, she's going to insist on scale (she's the one feeding the peacock). {grin} <<

Facebook comparison version on the way (uploading now). She's going to be famous!

Can you guys suggest a frame or two and possibly a specific portion of the frame to take some standardised screengrabs from the files we are uploading to illustrate the differences in resizing/deinterlacing/decombing etc.? I'm thinking a frame from stringer's highway clip for starters, with a fast moving car and a moving sloping white line.
amendegw schrieb am 30.01.2011 um 22:38 Uhr
My favorite place to look for blocking is frame 1453:



This was from a 1Mbps Handbrake render.

...Jerry

System Model:     Alienware M18 R1
System:           Windows 11 Pro
Processor:        13th Gen Intel(R) Core(TM) i9-13980HX, 2200 Mhz, 24 Core(s), 32 Logical Processor(s)

Installed Memory: 64.0 GB
Display Adapter:  NVIDIA GeForce RTX 4090 Laptop GPU (16GB), Nvidia Studio Driver 566.14 Nov 2024
Overclock Off

Display:          1920x1200 240 hertz
Storage (8TB Total):
    OS Drive:       NVMe KIOXIA 4096GB
        Data Drive:     NVMe Samsung SSD 990 PRO 4TB
        Data Drive:     Glyph Blackbox Pro 14TB

Vegas Pro 22 Build 239

Cameras:
Canon R5 Mark II
Canon R3
Sony A9

amendegw schrieb am 30.01.2011 um 22:59 Uhr
"Heck, I want a cut from the poppy harvest!"In case anyone cares, the poppies were from Longwood Gardens in Chadds Ford, Pa. The peacock calls the Fountain of Youth in St. Augustine, Fl home.

...Jerry

System Model:     Alienware M18 R1
System:           Windows 11 Pro
Processor:        13th Gen Intel(R) Core(TM) i9-13980HX, 2200 Mhz, 24 Core(s), 32 Logical Processor(s)

Installed Memory: 64.0 GB
Display Adapter:  NVIDIA GeForce RTX 4090 Laptop GPU (16GB), Nvidia Studio Driver 566.14 Nov 2024
Overclock Off

Display:          1920x1200 240 hertz
Storage (8TB Total):
    OS Drive:       NVMe KIOXIA 4096GB
        Data Drive:     NVMe Samsung SSD 990 PRO 4TB
        Data Drive:     Glyph Blackbox Pro 14TB

Vegas Pro 22 Build 239

Cameras:
Canon R5 Mark II
Canon R3
Sony A9

musicvid10 schrieb am 30.01.2011 um 23:07 Uhr
Nick, since you're up, try a quick test for me.
Take the remuxed download and open in Vegas 10. Do the scopes match your un-remuxed download (is that a word)?

I noticed the AVIDemux version, although the levels appear and play exactly the same, did not keep the 709 flag. If that affects both versions of Vegas, it explains our paradox.
NickHope schrieb am 30.01.2011 um 23:25 Uhr
Jerry, that frame is perfect for the blocking. We need a similar standard frame for the deinterlacing/decombing, but not a whole frame so that we can see the image at 1:1 here on the forum. Max 750px wide. Mark, where are you looking to judge this stuff?

>> Take the remuxed download and open in Vegas 10. Do the scopes match your un-remuxed download (is that a word)? <<

Sure it's a word, and yes, the scopes match precisely. Um.. struggling with the last bit.
amendegw schrieb am 30.01.2011 um 23:58 Uhr
"Jerry, that frame is perfect for the blocking. We need a similar standard frame for the deinterlacing/decombing, but not a whole frame so that we can see the image at 1:1 here on the forum. Max 750px wide. Mark, where are you looking to judge this stuff?"Wow, that's a tough one. I see no deinterlacing/decombing issues anywhere. However, if I were to look for them I would look for mouseteeth/jaggies in a diagonal line with motion. That said, frame 1735 catches a diagonal blade of grass in the midst of a pan. I cannot see any deinterlacing artifacts - even in my 1Mbps test. Maybe I'll test at a very low bitrate and see what happens - a test for tomorrow (I sleep!).


...Jerry

System Model:     Alienware M18 R1
System:           Windows 11 Pro
Processor:        13th Gen Intel(R) Core(TM) i9-13980HX, 2200 Mhz, 24 Core(s), 32 Logical Processor(s)

Installed Memory: 64.0 GB
Display Adapter:  NVIDIA GeForce RTX 4090 Laptop GPU (16GB), Nvidia Studio Driver 566.14 Nov 2024
Overclock Off

Display:          1920x1200 240 hertz
Storage (8TB Total):
    OS Drive:       NVMe KIOXIA 4096GB
        Data Drive:     NVMe Samsung SSD 990 PRO 4TB
        Data Drive:     Glyph Blackbox Pro 14TB

Vegas Pro 22 Build 239

Cameras:
Canon R5 Mark II
Canon R3
Sony A9

NickHope schrieb am 31.01.2011 um 00:14 Uhr
Jerry, at the moment I'm looking for a frame specifically to study Handbrake decomb vs Tdeint vs Vegas interpolate in the versions already created.

For those interested, Facebook upload of the same MeGUI "Test 1" render is here. You need a facebook account to view it. I'll embed a version for non facebook members on my site later. Vimeo version also now live.
NickHope schrieb am 31.01.2011 um 00:43 Uhr
Wow, the Facebook render is really nice. Possibly even better than the Vimeo, which still gives hints of that infamous Vimeo stutter.
amendegw schrieb am 31.01.2011 um 00:49 Uhr
'Jerry, at the moment I'm looking for a frame specifically to study Handbrake decomb vs Tdeint vs Vegas interpolate in the versions already created."As I mentioned earlier, I can see no deinterlace issues anywhere. So, I decided to do a test and see if my theory about diagonal images with motion was a good predictor of deinterlace issues. I intentionally tried to create a problem clip by turning deinterlace off, and render at 4Mbps 640x360 using the Sony AVC encoder. I think my theory is correct - note the deinterlace artifacts!


So... the good news is: frame 1735 (or 1732 in my second test - was my mouse shakey when I placed the clip in the timeline?) is a good predictor of deinterlacing problems.

The bad news is... I can see no deinterlacing artifacts in any of our test clips.

...Jerry

System Model:     Alienware M18 R1
System:           Windows 11 Pro
Processor:        13th Gen Intel(R) Core(TM) i9-13980HX, 2200 Mhz, 24 Core(s), 32 Logical Processor(s)

Installed Memory: 64.0 GB
Display Adapter:  NVIDIA GeForce RTX 4090 Laptop GPU (16GB), Nvidia Studio Driver 566.14 Nov 2024
Overclock Off

Display:          1920x1200 240 hertz
Storage (8TB Total):
    OS Drive:       NVMe KIOXIA 4096GB
        Data Drive:     NVMe Samsung SSD 990 PRO 4TB
        Data Drive:     Glyph Blackbox Pro 14TB

Vegas Pro 22 Build 239

Cameras:
Canon R5 Mark II
Canon R3
Sony A9

amendegw schrieb am 31.01.2011 um 00:57 Uhr
"Wow, the Facebook render is really nice. Possibly even better than the Vimeo, which still gives hints of that infamous Vimeo stutter."I concur - watch the white car at the 36 sec mark. Not as smooth as playing locally via WMP, but better than YouTube, Vimeo and JW Player.

...Jerry

System Model:     Alienware M18 R1
System:           Windows 11 Pro
Processor:        13th Gen Intel(R) Core(TM) i9-13980HX, 2200 Mhz, 24 Core(s), 32 Logical Processor(s)

Installed Memory: 64.0 GB
Display Adapter:  NVIDIA GeForce RTX 4090 Laptop GPU (16GB), Nvidia Studio Driver 566.14 Nov 2024
Overclock Off

Display:          1920x1200 240 hertz
Storage (8TB Total):
    OS Drive:       NVMe KIOXIA 4096GB
        Data Drive:     NVMe Samsung SSD 990 PRO 4TB
        Data Drive:     Glyph Blackbox Pro 14TB

Vegas Pro 22 Build 239

Cameras:
Canon R5 Mark II
Canon R3
Sony A9

musicvid10 schrieb am 31.01.2011 um 01:18 Uhr
"The bad news is... I can see no deinterlacing artifacts in any of our test clips."

That's bad news? A year ago I couldn't get away from it.

For the deinterlace shootout, I suggest a separate clip -- 1080i of an acoustic guitar player. Can one of you shoot it?
amendegw schrieb am 31.01.2011 um 01:29 Uhr
I guess that is kind of an oxymoron isn't it? It's only bad news if you're looking for a frame that contains artifacts and can't find one.

A deinterlace shootout? You are a sadist, arn't you?

...Jerry

System Model:     Alienware M18 R1
System:           Windows 11 Pro
Processor:        13th Gen Intel(R) Core(TM) i9-13980HX, 2200 Mhz, 24 Core(s), 32 Logical Processor(s)

Installed Memory: 64.0 GB
Display Adapter:  NVIDIA GeForce RTX 4090 Laptop GPU (16GB), Nvidia Studio Driver 566.14 Nov 2024
Overclock Off

Display:          1920x1200 240 hertz
Storage (8TB Total):
    OS Drive:       NVMe KIOXIA 4096GB
        Data Drive:     NVMe Samsung SSD 990 PRO 4TB
        Data Drive:     Glyph Blackbox Pro 14TB

Vegas Pro 22 Build 239

Cameras:
Canon R5 Mark II
Canon R3
Sony A9

NickHope schrieb am 31.01.2011 um 07:43 Uhr
I did a render in Sony AVC in Vegas 10.0c to the "Internet 1280x720-30p" template. The template specifies 8 Mbps bit rate but the render is actually only 5.8 Mbps (16.0 Mbps max). I'll try one at 11 Mbps and see what it delivers. The other interesting thing is that Avinaptic reports "8x8dct: Yes" for the Sony AVC render.

Now here is the YouTube version of my MainConcept AVC render from Sony Vegas 10.0c. I'm calling this "Test 2".

Video: 29.970 fps, 1280x720 Progressive, 2-pass VBR, Ave. 8,000 Kbps, Max. 40,000 Kbps. Best.
Deinterlace Method: Interpolate Fields
Audio: 320 Kbps, 44,100 Hz, 16 Bit, Stereo, AAC
Render time: 52 mins on my core 2 duo T7800 @ 2600 MHz with 4GB RAM

(I decided to raise the max really high and let the MC encoder do whatever it likes around the 8 Mbps average)

A copy of the file I uploaded is here. I can't upload anything to Vimeo until Saturday so if anyone has a Plus account or a clean upload slate, please feel free to upload it there.

If you're using 10.0 then please render your own and save my bandwidth. Thanks!