|
|||||||||
|
Thread Tools | Search this Thread |
September 8th, 2010, 08:16 AM | #1 |
Major Player
Join Date: Mar 2010
Location: Red Lodge, Montana
Posts: 889
|
Suddenly very slow DVD encoding w/ CS5
I'm trying to burn a 23 minute timeline/sequence to disk for client playback at a function. (This is an excerpt from what will be a much longer project and I am specifically making it for the client's use at this gathering.) Encoding to DVD used to take real time or less. (By "used to" I mean as recently as last Saturday). This morning it took 2½ hours for this 23 minute timeline.. I'm trying to figure out what may have gone wrong.
I'm using CS5 under Win 7 with a "DIY7" recipe machine: I7/920, 12 g RAM, GTX260 card with the MPE "hack" enabling hardware MPE. The video and project files are on a G-Speed Raid 10 with a separate Raid 0 as the target for the encode files. Neither array is more than 50% occupied by files. I have all the most CS5 current updates as of a couple of days ago. The PProCS5 timeline is a 23 minute multi-cam plus two additional sets of clips cut into the multi-cam and 5 audio tracks (cut so that only two or three are active at any time). PPRo sequence presets are Matrox 1920x1080/60i. The video is a mix from three AVCHD cams and 3 HDV cams. A minor amount of PPro color correction effect used on one of the HDV tracks. Sixteen dissolves and three wipes plus fade-ups and downs at the ends of the sequence. Everything rendered before Dynamic Link invoked. I've been doing similar mixes of AVCHD and HDV ever since CS5 came out in May, and encoding has been fast. I used Dynamic Link to move the sequence to Encore CS5. Transcode setting was High Quality NTSC 7 mb. (Same as I have been using for months.) The box for "maximum render quality" did not have an activation check-mark in either PPro or Encore. When I checked the system monitors during transcoding, I found that all eight cores were maxed out at 100% and 8.2 g of the 12g of RAM was being used. I kind of remember a posting (which I can't find but maybe it was by Harm Millaard) which indicated that the media encoder in CS5.02 will automatically set everything to max render quality if you have hardware MPE enabled. Anybody know if that is true and, if so, what can be done about it? Anybody else run into this with CS5.02 or have any ideas of things I can try this evening when I get back to the editing bay? Last edited by Jay West; September 8th, 2010 at 08:19 AM. Reason: typos |
September 8th, 2010, 10:00 AM | #2 |
Major Player
Join Date: Mar 2010
Location: Raleigh, NC, USA
Posts: 710
|
May not have any bearing, but... I've had something similar happen that was due to using the "wrong" media type in a drive that didn't like it. Specifically, I used what I had available which was a DV+R blank. The drive was unhappy, but did the job anyway -- just very slowly. Turns out that it works much faster with a DV-R blank. Don't know why, don't know if this has any bearing on your particular problem. Just something else to check maybe.
|
September 8th, 2010, 02:13 PM | #3 |
Major Player
Join Date: Jun 2007
Location: The Netherlands
Posts: 710
|
Even if you have MPE enabled you should also check Max Render Quality for those instances that are not covered by MPE rendering.
I do not use DL to Encore. I export in Premiere to so i can set my settings just the way i want to. Import into Encore, make menu, make iso, check iso (virtial clone drive and power dvd) and burn with Nero. Nero tells you if the burn was a succes, Encore does not. |
September 9th, 2010, 01:30 AM | #4 |
Major Player
Join Date: Mar 2010
Location: Red Lodge, Montana
Posts: 889
|
My workflow is similar to Ann's. I always burn to *.iso and use Nero to burn the disks.
However, the problem here is not disk burning or image file generation. The problem is the encoding time before I even get to image generation and disk burning.Once the files are encoded, Encore generated the *.iso file in a couple of minutes and Nero burned the disks almost that quickly. I suppose it could be possible that Dynamic Link is creating problems, but that seems counter-intuitive to me because I thought it was just a way to feed files to a DVD project in Encore. I sometimes use Dynamic link and sometimes do not. Sometimes, I "export" (clicking File --> export) from PPro which puts me into AME where I chose DVDMpeg2 settings. I've also built several projects in Encore by doing file imports of PPro timelines/sequences, too. I previously did this with CS4 and now do it with CS5. Thing is, in the past, I had not noticed that these differing methods resulted in any noticeable difference in encoding times. Also, I'm not sure what settings are available when you go from PPro to AME versus DL into Encore and then going into the AME encoding settings from there. I think I've been able to access any setting I wanted with whatever method I used; I just had to use different paths and menus depending on the application I started from. I asked about "maximum render quality" (MRQ) because (a) using it seems to slow things down to an extraordinary degree on my system and (b) I usually do not see that it makes much, if any, difference on short projects like this one. I've mostly found that a 2 pass VBR at a high bit rate is usually sufficient for my projects. I've used MRQ in the past mainly with longer projects with lots of motion and high contrast images where there may be disk space and image quality issues (say, with a dance recital DVD that runs more than 90 minutes). What puzzled me here was that I had not enabled it for this project and I still got the very long encoding time I would have expected if I had enabled MRQ. So, I will try some tests to see if differing export paths make any difference here. When I get some time tomorrow or Saturday, Ill try exporting this particular timeline/sequence from PPro via AME and see if that speeds anything up. In case there might be an issue with the mix of AVCHD and HDV, I'll also try exporting to Cineform and see if the Cineform files encode any faster. That will have to wait a couple of days until I get the time to try this, but I'll report back when I've run the tests. In the meantime, any other feedback would be appreciated. |
September 9th, 2010, 12:32 PM | #5 |
Trustee
Join Date: Aug 2009
Location: Chicago, IL
Posts: 1,554
|
Hey Jay,
The MRQ debate was Harm and me. Technically, hardware MPE + No-MRQ is better than software + MRQ. With hardware MPE + MRQ checked, AME will only use MRQ if certain requirements are met. For example, when using 720p source footage and encoding to 720p or 1080, MRQ is not used. However, when using 720 or 1080 and encoding to 480p, MRQ is engaged if its box is checked. I have found significant increases in quality when going from 720/1080 to 480p or smaller. The main advantage is with edges and lines - they are sharp whereas with non-MRQ, they are soft. I now use MRQ on 100% of 720x406 exports where the bitrate is 1000kb/s and above (anything less, there is no difference) and 100% of 720x480. For you, could it be possible that the 5.02 update is causing a problem with your Matrox drivers? If you updated to 5.02, then you must re-do the hack as well. |
September 9th, 2010, 07:28 PM | #6 |
Major Player
Join Date: Mar 2010
Location: Red Lodge, Montana
Posts: 889
|
Thanks Steve.
I redid the hack as soon as the 5.02 upgrade was installed, so that is not the problem. With the Matrox sequence presets which worked fine prior to 5.02, I suppose something got corrupted or just went goofy when I applied the 5.02 patch. I had not thought of this before, but the Matrox Mini instructions say to install the Matrox Mini software only after you've installed the Adobe applications. So, maybe I need to uninstall the Matrox software and then reinstall it. We'll see if that fixes the encoding time problems. One more thing to check out on Saturday morning. I'm thinking that I also should test this by copying the footage from the Matrox timeline (rather than nesting) into a new sequence with an Adobe AVCHD preset and then see if using non-Matrox sequence presets makes any difference. If the encode then runs faster, I'll know if there's a problem with CS 5.02 and the Matrox drivers. |
| ||||||
|
|