I am recompressing some video files and the results are good, but just a shade jaggy. I'd like to get the advice of people here to see if I could do it any better.
Currently I'm taking mpeg2 files at 704x480 and recompressing them to mpeg4 at 512x336. This reduces the file size from 3gb to 500mb. The resulting movie plays extremely well on the computer screen, but if I burn back to dvd I do get a little big of jaggies. Not a lot, but for a person looking they are there. Any advice?
I am using the deinterlace and deblock filter, two pass and high quality options, mpeg4 mencoder settings.
Dave
Results 1 to 10 of 10
-
-
You get jaggies because the 336 line tall image has to be stretched back to 480 lines for display on the TV. This is probably accomplished by duplicating some of the scan lines. Recommend you keep the original frame size.
Avoid deinterlacing the source. One of the most common deinterlacing techniques is to simply throw out one field and duplicate, or interpolate from, the first. This results in jaggies.
If your source is a telecined movie try inverse telecining it to restore the original film frames.
-
Hmm, i've been reading the manual:
"Selects the algorithm used to resize the image. Choose the default "fast-bilinear" when speed is important. For better image quality but slower encoding, you can choose either Bicubic (best for upscaling, ie scaling to a larger size), Bilinear (best for downscaling, ie scaling to a smaller size than source), Lanczos, or Bicubic spline (sharp picture)."
It looks like using Bilinear, Lanczos, or Bicubic would give me a better picture but it is greek to me. If someone can tell me what the difference is that would be great.
Dave
-
When you shrink an image down you loose detail. when you expand it again you can't restore the detail that was lost. At best you can "guess" what might have been there. Avoid resizing if at all possible.
I think your main problem is the deinterlacing. See this thread for some sample images I uploaded a whiel back:
https://www.videohelp.com/forum/viewtopic.php?t=230828
Note especially the jaggies on the woman's arm in the deinterlaced samples.
-
I thought I needed to deinterlace. Is this incorrect?
It can take hours to explain all the possibilities. If you tell me a little about your source and what you plan to do with it it will be much easier for me:
Is it a movie that was originally shot at 24 fps, then telecined for television?
Is it a true 60 field per second video? Say, a sporting event that was broadcast live?
Where do you plan to watch it? On television via a standard DVD player? On television via a special MPEG 4 DVD player? Only on the computer?
-
I am recording tv shows to my panasonic dmr-e30 dvd recorder from dish network. i'm pulling the video off the dvd-ram disk as a .vro which apparently is the same thing as a .vob. I use ffmpegx to recompress as mpeg4, and finally run it through divxdoctor so I can edit out the commercials using quicktime. I have two goals:
1) to have all the shows available on my laptop to watch
2) maybe burn back to dvd minus the commercials
If I had known I would go down this path originally I would have bought the dvd record with the hard drive, edited out the commercials there, and then burned to dvd. It would have been much easier if a little less precise.
Dave
-
For saving on DVD (to watch on TV) I recommend you use a program like Womble MPEG Video Wizard (EDIT: Just realized you're on a Mac so you won't be able to use Womble). This will allow you to cut out commercials and such without reencoding. No deinterlacing is necessary. This will definitely give the best quality. And since you're not reencoding it will go pretty quickly.
For converting to MPEG 4 and watching on the computer you have two main possibilities.
If the source material is from 24 fps film (most movies, many TV shows) it's definitely worth it to inverse telecine (IVTC). This is a special form deinterlacing which restores the 24 fps film frames from interlaced video that has been telecined (converted to 30 fps for TV). I normally reduce the frame size from 704x480 to 640x480 to maintain the proper overall image aspect ratio on the computer. If you need to reduce the frame size IVTC is still your best choice.
(I inverse telecine recordings from digital cable all the time. Occasionally it doesn't work well because the broadcast looses the 3:2 telecine pattern. It usually happens at scene changes so I assume the video was edited on video tape after being telecined. If it's only a few times during a movie I'll encode in pieces. But sometimes there are just too many chunks to deal with.)
If you can't IVTC then you'll want to deinterlace, especially if you're scaling the frame size. I'm not familiar with ffmpegx so I can't give you specific suggestions for it. But you should still consider a 640x480 frame size.
Similar Threads
-
Subpicture highlights look jagged on TV
By Henkie in forum Authoring (DVD)Replies: 9Last Post: 10th Feb 2014, 13:08 -
Jagged lines upon playback
By seven_deuce offsuit in forum Video ConversionReplies: 10Last Post: 17th Jul 2010, 08:31 -
No jagged DV to HD possible?
By chamo in forum EditingReplies: 8Last Post: 22nd Dec 2008, 17:49 -
Removing Jagged Effect
By Vid-Kid in forum RestorationReplies: 5Last Post: 20th Jan 2008, 21:03 -
Procoder 2 - Jagged Edges
By pbar in forum Video ConversionReplies: 4Last Post: 24th May 2007, 09:32