Page 1 of 1

Proper procedure to expand profile quality?

Posted: Wed Dec 26, 2018 11:42 pm
by stevevido
I am editing several videos that are digitized versions of 16mm film. The group that did the videos telecined and added duplicate frames plus did a harsh compression of the video. I have broken up each full video into smaller scene videos (clips) to get more consistent levels and noise for processing. I detelecined the videos and removed most duplicate frames with AviSynth. I am now using VirtuaDub to remove noise and make other adjustments to each clip.

I just installed NV as a VirtualDub plugin. I am building my first profile. Each video is only 720 x 480 pixels so finding large enough areas with no features is difficult. I am manually selecting small areas to build a custom profile. There are only one or two small areas in a frame to use for tuning. To select enough areas to build a profile, I need to use frames in different parts of the clip. I need to pick frames that are more than the +5 or -5 frames from the first chosen frame.

To access frames that are far apart in a clip, is the following process proper?

- Select first frame then click Build Noise Profile button
- Select small area in frame and click Auto Profile button
- Click Apply button to go back to "Please select a frame..." window
- Select frame in another part of video. Click Build Noise Profile button
- Select area and click Manual Fine-Tune button
- Continue picking areas, manually adding them, going back to "Please select a frame.." and back to Build Noise Profile as often as necessary to get a good profile

If this works ok, is it ever appropriate to use a frame or frames from a different clip to select areas to manually add to a profile? This is using multiple clips to enhance a single profile. I ask because there are groups of clips with similar noise but have additional lighting conditions that would seem to help enhance the quality of a given profile.

Opinions?

Thanks,
Steve

Re: Proper procedure to expand profile quality?

Posted: Tue May 07, 2019 12:22 pm
by NVTeam
Hi Steve,

Sorry about delayed reply but I though it may still be useful.

Yes, your VirtualDub-based workflow is correct. You generally use Auto Profile only and then manually fine-tune the profile using area from the same frame, from the same clip and/or from clips showing the same noise.

If you use the new version 5 of Neat Video, you also have one more option to build a profile in a difficult situation: you can create a generic profile and tune it to the noise in your clip.

Vlad