Compressor -- What settings do you use?

[Ed Mohar]

Hey guys,

I'm working on putting out a soccer highlight video, mixing 1080i footage and SD-widescreen footage, down to DVD. The video is only 4 min. I'm having a lot of trouble getting compressor to put out something that isn't pixelated.

I don't really have time to set the resize and deinterlace frame controls to best settings, it says it takes 300 hours and I can't believe that 4 min will take 300 hours. At my TV station, Avid Liquid takes a 1 hr documentary 6 hrs to press down to DVD and it looks spectacular (again using HD and SD footage)... Knowing that, I really do not want to test a 300 hour export just to find that there's a good chance its bad.

I've been reading through previous posters posts, a lot of theories discussed but not a lot of numbers, so my question today is:

What settings do you use for your shorter videos, settings that have been successful, and how long did it take to process down?

Many thanks,

Ed Mohar

[William Hohauser]

PIxelated where? On a DVD or on your HD screen?

Try downconverting the project to SD in Compressor first and then make the DVD. Works for great for me.

[Josh Bass]

For issues like this, just in general, I do ten second tests. So you make a clip from the project, then just throw that ten seconds in compressor and try different settings. You could even do one second, if it'll show you enough about how it's working. You can also use these tests to make a guess about the final file size for the whole thing (if ten seconds is x MB, then 10*6*4 will give you the total file size). Of course this isn't exact, since the amount of movement from frame to frame and color/luminance changes affect the file size, but you usually get close, in my experience.

I try to test the most difficult parts of a movie/file/clip to see file size (this can make your size estimate high, but better to have something come out much smaller than you expected than larger), artifacts, etc. So take something with lots of movement, etc. vs a static monochrome portion.

As for the 300 hours, it judges how long something will take by the difficulty of the frame it's on multiplied by the total number of frames in the file/project. There may be something really difficult to process at the very beginning of your project.

[Chris Korrow]

500 minutes for a 4 min clip, something isn't right.

Are you working in an HD sequence or SD?

I never send to compressor straight from the timeline, have never been satisfied with the results of interlaced to compressor. Now I always make a self contained prores 30p movie and just drag that into compressor. That also gives me an archival backup if I don't want to save the entire project.

This also makes the de-interlace setting a moot issue.
2 pass, Frame controls on, Detail levels set to 20.

For 4 min, would take no more than maybe a couple of hours + - for both of these renders. - 2.93 quad MP

[Simon Denny]

Hi Chris,

I'm interested in your frame controls settings and the details setting set to 20. What does setting this to 20 achieve?

[Chris Korrow]

Several months ago I poured over everything I could find on HDV to DVD in these threads & elsewhere, and after trying many different settings, these seem to work best for me.

I believe that particular setting came from a Robert Lane post (not sure) and someone else here could probably give you a better explanation of what this does. I can only guess that it takes a bit longer to analyze each frame for a better quality.

Since there were so many post/settings, I just copied & pasted bits & pieces into a document, so I don't have a link to the thread. But here's the post I've been mostly working from.

*sorry I can't give credit to who wrote it!

Here are the settings I use in Compressor. The results are very good. Since you have that software, I strongly recommend you make good use of it as it has a lot more flexibility than just using DVDSP.

You can also "batch process" in Compressor, having it do several different tasks, unattended, overnight.

I'm shooting 1080p30 so that might make a difference. Export Apple ProRes HQ, self-containted. 1920 x 1080 p30.

Drag that file into Compressor --

File Format: MPEG-2
Stream Usage: SD DVD
Video Format: NTSC, frame rate 29.97, aspect ratio 16:9
Quality: Mode - Two-pass VBR.

NOTE: Two-pass Variable Bitrate Compression provides the best quality output and is the most efficient. One-pass VBR isn't quite as good. Two-pass VBR is also slower since it has to go through the entire file to determine compression rates throughout, then go back to apply those rates accordingly. Average bit rate 6.2 Mbps. Maximum bit rate 7.7 Mbps. Motion estimation, best.

Frame Controls: On. Resize filter, better. Output fields, same as source (I'm using progressive so this doesn't really matter). Deinterlace, fast (no deinterlacing taking place in progressive so this doesn't matter). Adaptive details: on.

This is another reason why I shoot progressive frame HD. Eliminates the problems caused by interlaced video. Interlaced video was invented to solve problems caused by the slower scanning rates of television picture tubes in the 1940s. We don't have that problem in the 21st Century so why make things more complicated?

Anti-alias, off. Details level: 20.

Rate conversion: fast (no rate conversion so it doesn't matter). My own personal opinion again: I produce for TV and 30 fps is the native frame rate. So no 24 fps for me since, again, I don't want to make things more complicated.

You can try setting the "resize filter" to "best".

Be aware that setting the "details level" to anywhere above zero will increase the compression time considerably. You might want to do your compression in segments instead of the entire program at once. Our show consists of eight segments and each one is done on its own. In case Compressor should crash overnight at least not all is lost.

IMHO, if you're shooting HD, you should edit in HD. Conversion to SD should be last. That way your program can be re-purposed to HD anytime in the future. Some make the mistake of editing in SD then changing their minds later, and having to re-scale other elements such as graphics and titles. It's almost like doing the whole job twice, and that doesn't make sense. It also makes things more complicated.

Because you're working with interlaced, you might want to do some systematic tests to see what works best. That's how I set my details level at 20. I batch-processed several settings with short test files in Compressor and looked at each one. You can use these settings as a starting point, then change one variable and see what happens. Because Compressor can process several things at a time, you can set up a variety of tests, let them all run, and compare the results afterward.

Just be sure to work methodically and take careful notes.

[William Hohauser]

These are excellent suggestions. I have to semi-correct the statement about interlaced video. Interlacing is an analog compression technique to cram enough scan lines into a frame to get acceptable detail while remaining within a set broadcasting bandwidth. Progressive SD transmission would require nearly double the bandwidth which would reduce the number of transmitting stations within a set area. The switch over to HD was delayed for years because of bandwidth issues. Once MPEG compression was agreed upon the switch over could go forward.