logo Sign In

Post #488414

Author
Moth3r
Parent topic
Puggo Strikes Back! (Released)
Link to post in topic
https://originaltrilogy.com/post/id/488414/action/topic#488414
Date created
4-Apr-2011, 8:57 AM

Puggo - Jar Jar's Yoda said:

Yes, it's barrel distortion. I tried correcting it in the Puggo Grande with a VirtualDub filter designed for that purpose, but the results were horrible.  On some scenes it was ok, but on others it was completely unacceptable, such as on the crawl.

There isn't really a fix for it during capture.  The WorkPrinter wasn't initially designed to capture the entire frame of a scope print, and with the enlarged gate you see it.  Frankly, after being upscaled, I think it's pretty minor.

If you want something to play with, I think I still have a raw clip here:

http://rapidshare.com/files/454420238/Tdark.avi

I had a little play with the raw clip last night.

Firstly I looked at the luma levels. It's interesting that there is a substantial amount of "super-white" information in the clip - pixels with a luma value greater than 235, which would normally get clamped when converting to RGB for display. This detail can be recovered by lowering the white level.

I also lowered the black level to roughly where it should be.

My levels adjustment did give a nice improvement IMO, but there was still some shadow detail lost. If you can work out a method to apply HDR techniques to merge a light and dark capture together that would be much better.

Although I didn't come up with anything, I think that some colour adjustments are needed, particularly to reduce the blue levels. The catwalk that Luke and Vader fight on should be more of a steely grey than a blue-grey.

I also had a play with the DeSpot filter for AviSynth. After a bit of parameter tweaking, I was getting some pretty impressive dirt spot removal in the Hoth scenes; the natural film grain was untouched.

Unfortunately I then checked the transporter fly-by shot and found that the filter was eating half the stars from the panning starfield. Would be worth amending the parameters so that it only looks for dark spots, or even attempt a mo-comped script (see the "combined usage" notes at the above link) to reduce the number of false-positive spot detections.