Am I doing it right?

[yourmom]

Hi everyone!

I just wanted to compare procedures with everyone else that enhances videos. Obviously, every video is different, and requires additional manual tuning as such. Apart from such unique procedures however, this is normally how I go about processing my videos:

- To deinterlace, I use a procedure known as "field extension deinterlacing" (http://en.wikipedia.org/wiki/Deinterlacing). The video is loaded up in virtualdub, a deinterlace filter is put in place to seperate each deinterlaced frame into 2 seperate frames (effectively doubling the frame rate and halving the vertical resolution), and save the results as an uncompressed avi.
- The audio is extracted from the video through virtualdub as a wav file. Primary reason is because Vreveal will often butcher the quality of the audio unnecessarily.
- The video file is then processed through Vreveal. Filters applied include Clean (high), plus Auto Contrast and Auto White balance (if needed). Results are saved as an uncompressed avi.
- The video is opened using virtualdub, and upscaled using the Super resolution VDF filter. Results are saved as an uncompressed avi.
- The video is reversed using AVS in conjunction with virtualdub, upscaled a second time using the Super resolution VDF filter (See http://forum.infognition.com/index.php/topic,277.0.html as to why I do this), have its borders cropped (where necessary. VCR footage mostly), and use a resize filter to downscale the project to 1080p (using Lancoz 3). Results are saved as an uncompressed avi.
- OPTIONALLY this video gets loaded in Vreveal again, between the cropping and resizing stages taken in the previous step. This is where I apply a stabilise filter to the video, and save as an uncompressed avi. I normally perform this step on modern videos only, while skipping this step for old footage.
- The video is re-reversed using AVS in conjunction with Virtualdub, so it runs the right way. The audio (saved of in step 2) is imported back into this footage. Results are saved as an uncompressed AVI.

It is at this stage that I make my final modifications to the video (if required. Procedures will vary from video to video). Afterwards:

- The video is loaded using MeGUI, and is converted to mkv using a "HQ-Insane" profile that I obtained from http://forum.doom9.org/showthread.php?t=101813.

Am I going about this right? What procedures does everyone else normally follow with their videos?

[Dee Mon]

Wow, that's a very sophisticated process. It sounds logical but I doubt anyone else tried to follow it.

Personally I usually just combined editing (cropping, trimming, joining) in VirtualDub and super resolution in Video Enhancer (plus deinterlacing and denoising when necessary). And now when SR is available for VDub, Video Enhancer is used when the input file is not AVI.

[yourmom]

Wow, that's a very sophisticated process. It sounds logical but I doubt anyone else tried to follow it.

I normally take video from vcr tape, and convert all of the individual scenes into their own seperate video too

It's a lot easier now, thanks to the Virtualdub plugin. Before the plugin (back when I was using the standalone video enhancer), I had to upscale the video with video enhancer, reverse the video using virtualdub, upscale the video a second time with video enhancer, then re-reverse the video with virtualdub. That one extra step back then costed me many hours and many terabytes of temporary storage

Is there anything though which can clean/denoise video better than vreveal at this time though? I dont care if it takes months of extra processing time to make a minute of footage marginally better

[Dee Mon]

Have you tried Neat Video? It's the best denoising filter we've seen for VDub.

[yourmom]

I am not familiar with this software, but I'll give it a try. Thanks Dee Mon

[yourmom]

I am playing with the demo version of neat video right now, and I must say that what I have found so far is very impressive

Would it be wise however to apply this to a video before upscaling, or afterwards?

[Dee Mon]

SR preserves noise, so unwanted noise should be removed before SR.

[yourmom]

I guess thats plausible, but if one was to perform the noise cleaning using Neat Video before upscaling with the virtualdub plugin, then wouldnt the super resolution process have less information to work with?

[Dee Mon]

Yes, denoising loses some information. Unfortunately good signal is mixed with noise and it's too hard to separate them. So you either lose noise and some good information too, or you upsize good signal together with the noise.

[yourmom]

Well, what I might do in the immediate future then is experiement with both procedures to see what I like

I purchased Neat Video last night btw

[yourmom]

Well, while we're comparing notes, may I ask what plugins you utilise to enhance contrast, white balance, sharpness, and anything else that I cant think of at the moment Dee Mon?

[Dee Mon]

After comparing different VD filters for these tasks we came to the following choices:
color correction - Color Mill
automatic brightness correction - MSU Smart Brightness
deinterlacing - "deinterlace area based"
denoising - Neat Video
sharpening - don't know yet.

Also, Deshaker is a nice filter too, helped me a few times for shaky videos shot from transport.