Avisynth_challenged
26th May 2008, 20:57
Hi all,
I'm trying to clean up an analog capture. The source is a live-action movie digitally converted from laserdisc, then saved to Huffyuv 2.1.1.
The main issues with my source video are video noise, and film grain. I plan to perform the following steps to clean the source:
Step 1: Preliminary spatial filtering, using "Defreq()" and "Undot()",
Step 2: Averaging three captures which have gone through Step 1, using either "Toot" (http://forum.doom9.org/showthread.php?t=78338) or the classic Dividee formula for capture averaging (http://forum.doom9.org/showthread.php?p=232167#post232167),
Step 3: Grain removal (best method TBD)
I've performed Step 1 on one of my three captures. I ran two versions of my Avisynth script:
1. Defreq(), followed by Undot()
2. Undot(), followed by Defreq()
The first script yielded a video which was slightly smaller (i.e. better compressed) than the video output of second script. To my eye, the two videos were visually identical.
I'm inclined to run with the first version of the script, because it compressed slightly better. But I'm wondering if my decision algorithm is correct. Basically I'm running two filters, and then deciding that the best order of the two filters is whichever one produces a smaller file, all other things being equal.
So I guess my general question to the community would be: is this a reasonable approach to decide in what order to apply denoising filters in Avisynth?
Thanks in advance for any feedback: :thanks:
I'm trying to clean up an analog capture. The source is a live-action movie digitally converted from laserdisc, then saved to Huffyuv 2.1.1.
The main issues with my source video are video noise, and film grain. I plan to perform the following steps to clean the source:
Step 1: Preliminary spatial filtering, using "Defreq()" and "Undot()",
Step 2: Averaging three captures which have gone through Step 1, using either "Toot" (http://forum.doom9.org/showthread.php?t=78338) or the classic Dividee formula for capture averaging (http://forum.doom9.org/showthread.php?p=232167#post232167),
Step 3: Grain removal (best method TBD)
I've performed Step 1 on one of my three captures. I ran two versions of my Avisynth script:
1. Defreq(), followed by Undot()
2. Undot(), followed by Defreq()
The first script yielded a video which was slightly smaller (i.e. better compressed) than the video output of second script. To my eye, the two videos were visually identical.
I'm inclined to run with the first version of the script, because it compressed slightly better. But I'm wondering if my decision algorithm is correct. Basically I'm running two filters, and then deciding that the best order of the two filters is whichever one produces a smaller file, all other things being equal.
So I guess my general question to the community would be: is this a reasonable approach to decide in what order to apply denoising filters in Avisynth?
Thanks in advance for any feedback: :thanks: