logo Sign In

GOUT image stabilization - Released — Page 43

Author
Time

What happened? Everyone has real pics. What did I miss?

-G

Author
Time

g-force said:

What happened? Everyone has real pics. What did I miss?

-G

LOL, looks like it is your turn now.

Author
Time

dark_jedi said:

g-force said:

What happened? Everyone has real pics. What did I miss?

-G

LOL, looks like it is your turn now.

Sorry to go Off Topic, but seriously DJ, I thought you had changed your avatar to Henry Rollins!

Save London’s Curzon Soho Cinema

Author
Time
 (Edited)

I hate to propose changes to the script at this late of a date, but I've been playing around with something lately that might really be useful.

Currently two NNEDI2 calls are used to antialias, and assumingly remove any remaining combing artifacts (or line shimmer.)

qtgmc() is a deinterlacing filter that is a more updated riff on TempGaussMC, and has a progressive processing component.

It is not nearly as slow as when processing field based video.  I calculated about 12x faster than using it as a field deinterlacer.  (And MUCH faster than a double NNEDI2 call.)

http://forum.doom9.org/showthread.php?t=156028

This filter identifies edges/lines, cleans them and is really robust without destroying detail or creating new artifacts.

The basic usage for this is: qtgmc(Preset="Slower", InputType=1)

I don't have GOUT discs ATM to test this on, but since I've heard chatter that there are still some artifacts remaining after the current version of script, I was wondering if someone wanted to give it a try as a faster more effective answer.

Dr. M

Author
Time

Ever notice that a new filter always pops up AFTER you've finished all of your video encoding for a project? ;)

I'll check it out and see how it looks compared to what I had been using on my Hyperspace Collection (which was a combination of NNEDI2 and AAA).

Thanks, Doctor M ;)

Author
Time

I think the reason there are still jaggies and shimmering remaining in the image is that the instability combined with the low resolution results in some details ending up on completely separate lines, which could well be beyond fixing.  The script cleans up most of the lesser instances, which gives a huge improvement in the overall watchability, but the worst aliasing (the interrogation droid and the grid in the Yavin control room come to mind) will probably still be there no matter what.  Tis unfortunate, but that's what you get with a source that is mostly crap.

Still, I'd be quite interested in seeing what improvements could come from new anti-aliasing filters.

Author
Time

So far, from the test I just ran...this option isn't going to cut it, unfortunately :(

The image was a lot sharper, but it adds a weird "bob" to the picture and seems to do nothing for aliasing.  It tried it both with and without all of the "temporal repair" functions activated, but with similar results.

Someone else may want to have a go at this filter to double check.  I'd hate to have done something wrong and have a Holy Grail of GOUT restoration overlooked.

Author
Time
 (Edited)

I think you are right hairy, I am trying it out now just for shits and giggles, I replaced the double NNEDI2 lines with DrM's line, and LOL it is running 12 times SLOWER! and I have a fast QUAD core PC with 8 GB's RAM running a 64 bit system, I will see the results in a little while, but for now, way to slow for me.

I am pretty sure I am going to stick with what I already have, colors look great, and it just looks a lot better than what I am seeing at the moment with this tweak.

Author
Time

I think the reason there are still jaggies and shimmering remaining in the image is that the instability combined with the low resolution results in some details ending up on completely separate lines, which could well be beyond fixing.

Plus the vertical blur Lucasfilm added to the DVD to hide the problem only exacerbates it.  Even though the LD Definitive Collection suffers from the same issues overall, it seems the aliasing cleans up a tiny bit more.

Author
Time

^But an LD image isn't quite as clear, so you wouldn't see the aliasing as much thanks to it being a bit blurry. Also, if there really is a vertical blur, does that mean that the release was modified and not sent out just as-is straight rip? Interesting... shame they couldn't have been assed to do any more.

A Goon in a Gaggle of 'em

Author
Time

Okay, I've had some really good success with qtgmc to clean poorly deinteralced material.  I guess just not in this case.

@Dark_Jedi - I believe qtgmc can utilize multiple cores/threads... but not having them myself, I can't be more specific.

Dr. M

Author
Time

bkev said:

^But an LD image isn't quite as clear, so you wouldn't see the aliasing as much thanks to it being a bit blurry. Also, if there really is a vertical blur, does that mean that the release was modified and not sent out just as-is straight rip? Interesting... shame they couldn't have been assed to do any more.

If you check the end of the X0 Project thread, there are some screenshot samples addressing this.  You can see the detail loss from the vertical blur on some of Artoo's finer details.

Author
Time

I know; in fact, I posted them. I just wonder if it was an error in the encoding in some fashion or post-processing.

A Goon in a Gaggle of 'em

Author
Time

bkev said:

I know; in fact, I posted them. I just wonder if it was an error in the encoding in some fashion or post-processing.

 Whoops!  That'll teach me to reference past threads while I'm not looking at them.  This week really needs to end ;)

Author
Time

bkev said:

^But an LD image isn't quite as clear, so you wouldn't see the aliasing as much thanks to it being a bit blurry. Also, if there really is a vertical blur, does that mean that the release was modified and not sent out just as-is straight rip? Interesting... shame they couldn't have been assed to do any more.

The more I think about it, I seem to remember reading that the Lucasfilm marketing team showed a "before and after" when they were promoting the GOUT DVDs at one of the home video conventions.  Maybe they did do some minor cleaning (instead of leaving well enough alone).

Author
Time

Boy, it's been a while, so I can't answer your question off the top of my head, but I have detailed instructions on how I did it, I can check tonight when I get home. It's basically the same way Moth3r did it so that I would end up with the same frame numbers as he did. I know OSJ's instructions are correct also.

Let me check tonight and get back to you.

-G

Author
Time
 (Edited)

What is wrong with ripping file from DVD in IFO mode with DVD Decrypter and then running it through DGIndex using force Film? that's what I did, and all audio is in sync.

Author
Time

Here's what I did:

insert SW bonus disc into drive
open DVD decrypter
Mode > IFO
Edit > Select Main Movie PGC
Tools > Settings... under "IFO Mode" Tab check "Chapter Information - IfoEdit", click OK
File > Decrypt

open dgindex.exe
Video > Field Operation > Force Film
File > Open > (path where you saved the Decrypted files) "VTS_03_1.VOB" "VTS_03_2.VOB" "VTS_03_3.VOB" "VTS_03_4.VOB" "VTS_03_5.VOB" "VTS_03_6.VOB" "VTS_03_7.VOB"
File > Save Project (save to some path other than where the decrypted files are or you may get a file protection error)

Author
Time

I think some people are just making it harder than it really is, you can't get any easier than that.

Author
Time
 (Edited)

Nobody's making things harder.  Demuxing in different ways led to different frame counts, and nobody knew for sure how g-force had done it, so it needed to be established. That's all.

Author
Time
 (Edited)

dark_jedi said:

...you can't get any easier than that.

Said the dude with the porn star avatar ;-)