Desparate – Two problems using FCP to Compressor to DVDSP

[Timothy Anderson]

The pulse problem occurs on both my external ASUS HD monitor as well as my 16:9 SD television with a standard SD DVD deck. The pulsing seems to be caused by the 2pass VBR encoding of the MPEG-2, when the scene is encoded by itself with a CBR encode of 7ish, the pulse disappears – my problem is obviously I cannot encode the whole film (89 mins.) with a CBR and fit it on a DVD.

[Jeremy Garchow]

Are you willing to spend money to fix this?

What version of Compressor are you using? I remember this pulsing on older versions.

Are you on anointed Mac?

[Timothy Anderson]

I already did spend money to fix this when I bought the new FCP studio. I am running Compressor 3.5.1 on a Mac yes.

[Jeremy Garchow]

Damn iPhone. Not anointed. Intel. I meant intel Mac.

I recommend BitVice for quality MPEG2 VBR encodes. It’s a one trick Pony but works really well.
Have you tried single pass VBR?

[Jeremy Garchow]

Also, how did you capture this footage? DVCPro HD doesn’t natively run @ 23.98 in 1080 resolution without removing pulldown first. DVCPro HD is always acquired @ 29.97 in camera.

Jeremy

[Michael Sacci]

could I get the ill and motion files. and then I would put the animated text into a timeline and see what I can do.

[Timothy Anderson]

Hey Michael –

Illustrator file is 57MB and I actually don’t have the Motion files – just the resultant HD QT’s for each title card which is around 100MB. Motion was only needed to create sequence matching QT from transparent ILL file – no animation, effects, etc…

Quick note: The pulsing problem DOES NOT occur in VBR One pass enconding in compressor – only in TWO PASS – What is your theory on this? Initially I thought due to the static and long take nature of the scene compressor was looking at that scene as a spot to save space and was likely using a low bit rate for it under VBR (especially since the CBR encoding at 7 solved the problem). However, my theory is shot now that the encoding at VBR one pass results in NO pulsing. Ultimately I would prefer CBR, but what kind of quality loss would a one pass VBR best result in versus a 2 PASS best? Your thoughts are appreciated.

Tim

[Jeremy Garchow]

2 pass goes through and analyzes the whole file to check for motion characteristics and takes some ‘notes’. The second pass encodes the file based in those notes.

A one pass VBR does all of that in one pass. You’ll probably end up with a slightly bigger file as it’s ultimately less efficient.

Are you encoding your audio to Dolby or keeping it PCM?

[Timothy Anderson]

Hi Jeremy –

I’m encoding the audio track to Dolby. Could you speculate why I would be getting a better result with the one pass best? Also, I don’t know if you’ve followed the thread and are briefed on the other problem I am having of the HD title sequence looking terrible on DVD – I can’t solve that one either and am really put off because the titles are originally vector based and then converted to individual QTs at the same HD codec as the film so I do not understand why there are downscale issues – isn’t that the ultimate advantage of vector titles from Illustrator?

Tim

[Timothy Anderson]

Hmmm – Yeah I re-confirmed all of my workflow settings – all derive from a 23.98 fps timebase throughout from FCP into the MPEG-2 in Compressor. To be honest the footage itself looks great on the rest of the DVD outside of some occasional black noise typical of SD – nothing concerning, it’s just this one 15 sec. patch from the scene that is part of a long, static shot with little movement except very subtle foliage moving in the breeze in the background. I think this subtle movement is ‘sneaking’ past compressor because everything else is so still for so long it drops its bitrate here, resulting in the pulsing – the fact that a CBR encoding of this scene stand-alone at 7 erases the pulses convinces me. I just have never seen VBR one pass best outperform two pass best and can’t understand why…

On the HD titles – they were created as vector graphics in Illustrator as individual cards, brought into to Motion to place and add black over transparency, then exported at DVCproHD proper codec to match sequence. Again, the problem lies within how FCP/compressor is downsizing/rescaling them because in the HD QT they are fine in the sequence….I am baffled on this front really – even a 720×480 export in ProRes422 HQ of the title sequence alone (just to test) yields the same poor results…

tim