g-force

Chewtobacca said:

This looks great so far.  Stupid question: you guys do force film the video in DGIndex before using the script, don't you? 

I did this as a matter of course without thinking about it.  I cannot see why it would make much of a difference to functioning of the script, but just checking.

edit - G-Force, are you planning to work with the PAL GOUT for your ROTJ script?  Unlike the other two movies, the PAL is supposed to be better quality.  I would prefer to have an NTSC output, but one could easily work PAL to NTSC conversion into the script.  This would mean one could work with the best source material and it could produce PAL and NTSC outputs by just adjusting or deleting a few lines.

I also hope you will give me permission to release an MKV of this. 

Yes, force film is a must. I don't own the PAL versions, is there a thread that describes what makes the PAL ROTJ better?

Permission not needed, but if it makes you feel better, you've got it.

-G

JediTray said:

doubleofive said:

Yeah, I feel the same way as I don't understand any of the scripting they do, but we'll see the results in other preservation attempts.

 I certainly hope so.  The ones I currently have jog and bounce all over the place.  While I did notice it a bit on the old 4:3 TV, having an HDTV now really shows everything, warts and all.

Also, I wonder if this script, or a variant of it, would work on things like Super 8 transfers?

some form of the stabilization and degrain stages should work quite well actually.

-G

Chewtobacca said:

Even though you are cropping what looks like part of the picture, you are really not, as that part of the picture could just as easily be black pixels: it would make no difference. 

good point! Just do the top and bottom crop before the stabilization, the sides after the stabilization and you should be fine.

-G

Chewtobacca said:

Thanks, G-Force!  It was just this part from my earlier post.

Crop(8,96,-8,-96,align=true)

Crops about 4 more pixels off from the right side than I found you need to for ANH and 2 more than you need to for ESB, but maybe that's your preference. If you do this before image stabilization, then your left and right borders are going to move. If you do it after the stabilization, you just wasted a lot of processing on the borders. Maybe you better break up where you do the vertical and horizontal crops.

Chewtobacca said:

obviously the values have to change if one no longer resizes horizontally.

That would be true, if I had been resizing by A LOT or cropping A LOT, but I was zooming by a factor of 1.011. I'd have to be cropping later in the script by more than 100 to make the values change even 1 pixel. I think my max crop is 14, so the values of the crops shouldn't change.

-G

Chewtobacca said:

I would be grateful if someone could check the vertical cropping values that I changed.  I think they are right.

Sure, I would be more than happy to check the values, but what are the final final changes you are proposing?

-G

Guys,

Maybe it would help to state what the design goals were:

1. stabilize the image. 2. Anti-alias 3. Remove over-sharpening artifacts 4. crop off the border over-sharpening that couldn't be removed from the convolution.

One thing to keep in mind is that the Anti-Aliasing filter is a line doubler. You MUST resize afterwards unless you want an image that is stretched vertically. The entire script really is not as compartmentalized as the comments in the script would lead you to believe. If you really want to do away with the resize, you also have to remove the anti-aliasing.

The other thing to keep in mind is that the borders of the original were put in after telecine, so they don't move with the shakiness of the image. SO, when you do the image stabilization, NOW the borders move. Bummer right? So SOME cropping is justified. Yeah, one could argue that your TV won't display the edges anyway due to overscan... But not everyone is watching these on a TV. No harm done just blacking out what you don't want and what the TV doesn't show anyway, right?

One other thing. I know everybody thinks Spline32 is better than Spline16, but I could never see the difference, and the script ran just a bit faster with Spline16, so I went with it. I really don't feel too strongly about that, so if anyone can show some even slight improvement with Spline32, I would definitely change the script.

I kinda like discussing the finer details of the script, but please keep in mind that it's been a while, so I may be a little fuzzy on the exact details. I expect a small re-learning curve once I jump back into it to get the convolution for ROTJ done.

-G

 

 

dark_jedi said:

g-force, I am not rushing you in any way, shape or form, but do you have any idea at all when we might see the ROTJ script being final, whether it be a few days, a month, whatever?

Thanks man just want to keep these together, but if it is going to be a pretty long time yet then I can slow down a bit on finishing these 2.

 Most likely a few months, Sorry.

-G

dark_jedi said:

Yes whatever dude, I never claim ANYONE'S work as my own, EVER!, so you stay perched on your high and mighty horse, sorry for saying anything, maybe someday you will finish and share with everyone your hard works your self then, and NO ONE is perfect, well except you ofcourse.

NO need to even read this shit anymore.

Asta La Vista Big Guy, Internet sure makes some people really big, bad, and tough LOL

 Hey DJ, I may have over-reacted a bit. What do you say we just put it all behind us, as I'm sure people are looking forward to having some different audio options on this thing. That was something I never really got to, and it's nice that you care enough to do that.

Sorry, I never actually saw the credits on your release before, so I made some assumptions and that just makes an a** out of me.

Please don't get mad at anyone else here because of some sophmoric comments I made. I apologize.

-G

dark_jedi said:

g-force said:

(hey, I don't even have time to do ROTJ and it would be about the same amount of work)

-G

Well if you need help testing,rendering, encoding, I have time to help you just like we did in the past, because all I have left to do on the other 2 is encode then put it all together, the lossless avi's are done on both now.

 

What? You mean like on the project where you flaked out and erased all the files off your hard drive because you found out that I had offered some advise to someone else who was doing a similar project? Or are you talking about your previous encode where you took my code and changed a couple of things and called it your own with no acknowledgments? Or maybe you're referring to the time I gave you a script for color correction and told you just to evaluate it to see if it had promise and specifically asked you not to post anything because it was too preliminary, and then within a couple hours you had screenshots posted. Or like on my last encode where you were suspicious that the script was any better and slammed me for the encoder I was using?

Yeah, I'll be just like old times! Where do I sign up?

-G

Oldschooljedi said:

Thank you for posting these, I'll make final encodes when I find the time (can't actually wait).

You deleted the line

'PAL = false       # <-----Set to false for NTSC, true for PAL'

in you final scripts so are these only for NTSC-DVDs?

 Yeah, sorry man, there are just too many things that are specific to the NTSC version that I don't have the time to change them all for the PAL (hey, I don't even have time to do ROTJ and it would be about the same amount of work)

-G

Sorry guys, I still haven't had time to do ROTJ. I decided to just post the scripts for ANH and ESB (see the first page).

ROTJ will get done eventually, and a few people around here already have ESB, so hopefully this will make some others happy in the mean time.

-G

dark_jedi said:

...supposedly G's DVD5 version with freeware encoder is WAAAAAY better,sure would love to see this mysterious script enhancement myself...

Hey man, what's with the hostility? Yeah, they're DVD5, but all the menus and extras are stripped off, and these are not nearly DVD quality anyway, so they really don't need two layers.

And yeah, it's a freeware encoder, but stacks up with (and blows away a few of) the big boys, especially when you tweak the settings and play around with various matrices.

And the only reason I'm holding back the scripts is because ROTJ is not done yet. Some of the settings had to be different between SW and ESB, and I'm sure some will be different with ROTJ. I just didn't want the SW or ESB script applied to ROTJ, because I know it can be better. I just need some more time to work on it.

-G

It sounds like poeple are having difficulty with the ****** version. That's too bad. I have had some people PM me for SW and ESB, with the promise of ****** ** *******, but their circumstances are preventing them from doing it at this time. Maybe some more people will have to PM me ;)

-G

msycamore said:

I understand what you mean, but the footage we all have of the Original films in the form of the GOUT isn't even DVD resolution. Of course we want a HD version but you can never make the HD footage blend with the GOUT footage even with g-force's superb script to look any good. We don't even have the trilogy in laserdisc quality without flaws.

 Well put msycamore! Especially the last sentance.

-G

Oldschooljedi said:

Hey, g-force!

Beautiful news!

Congratulations to your new daughter!!!!

 

 

 

 Thanks OSJ! Don't worry, I'll get ROTJ done eventually, and then I'll post all of the scripts.

-G

Oldschooljedi said:

I must say that I've lost track here the last few weeks.

G-force, how are your encodings for ESB and ROTJ going?

Please let us know!

 ESB is as good as I can get it. I'm not completely happy with it, but hey, it's the GOUT after all.

ROTJ may take a while, as life has become much more hectic (in an awesome new daughter kinda way!) lately.

I'll let everyone know once ESB is up in the usual place.

-G

dark_jedi said:

Jonno said:

Hi DJ, I'm a bit behind - did you use the menus on this latest version? If not I can put the links to the materials up on here in case folks want to do it themselves...

Not yet but I intend to give it a whirl,version 2 has not been uploaded yet.

 DJ,

might want to wait until I get all the scripts worked out, otherwise you will not have de-ringing, improved anti-aliasing, and improved image stabilization.

ESB is coming along nicely.

-G

gltaylor74 said:

Thanks G-force.  Is 6.64 on page 1 the new script?

 

 

 No, the new script will be posted once I'm done with all 3. I'm making some minor tweaks for ESB, and I expect the same for ROTJ. Once all are encoded and checked I'll update the script. In the mean time, my encode will be the only way to get this.

-G

Darth Mallwalker said:

Looks great g-force!
But don't take my word for it--head over to ****** and see for yourself

 Thanks, DM!

-G

I have completed a new version of the cleaned up GOUT. This represents a noticeable improvement over all previous attempts in anti-aliasing, detail retention, de-ringing and image stability. I won't be posting the updated script until all movies are complete and checked, but ANH is encoded.

-G

vaderios said:

I think it looks too magenta.

-Angel

Agreed. DJ, I sent you the wrong numbers (from memory).

So you talked me into sending you the script from memory for "just for playin around", and then go and post the results. Oh well. ;)

Here is the actual script that seemed to work well for ANH:

Levels(0,1.5,255,0,255)
yv12 = last
ConvertToRGB32
r=ShowRed.  Levels(0,.75,255,10,255)
g=ShowGreen.Levels(0,.80,255,10,200)
b=ShowBlue. Levels(0,.75,255,10,190)
MergeRGB(r,g,b)
ConvertToYV12
MergeChroma(yv12,last)

dark_jedi said:

But keep in mind we are working with lossless avi files.

Oh, they won't be lossless when I get my hands on them. ;)

Anyway, what is the colorspace you are using? There may have to be some conversion in between, and I know I had to convert to RGB to get enough control over the color channels. I'll PM you the script tonight and if it's something that looks promising, I'll tell you how to tweak it for your purposes.

-G

DVD-BOY said:

Still working out the best approach to make on these releases - I don't have the skill, or patience to colour correct these releases like Adywan.

 

I've been casually (no Blu-Ray playback capabilities here) following this thread. I have worked up an avisynth script that gets rid of the blue and green push on the SE ANH DVD. If the colors are the same on the HD version, it may be of some benefit to this project.

Just though I'd throw it out there. Adywan did a nice job of color correcting the SE, but I thought some of the skin tones were a little off in some shots. Just reducing the blue and green push gets pretty close to what I remember Adywan's looking like but does a better job on the skin tones IMHO. But since the same reduction in color push is applied to the whole film, it may not be quite as good as Adywan's for some scenes.

Let me know if you're interested.

-G

zee944 said:

zee944 said:

g-force said:I've also found out how to repair a TON of detail by collecting pixels from the same object in different frames. The results are a little preliminary (like as of before work this morning) but the results so far are breathtaking (I literally couldn't believe my eyes!).

What did you find out? Please share :)

 

You haven't replied. Is it intentional?

It was maybe a little intentional. Old stuff really, and not to dissimilar to the current discussion about sharpness and detail.

At the time I was playing around with an the idea (not mine) of sharpening, but limiting the amount by the min and the max of the last and next mo-comped frames. Looks pretty good most of the time, but really aweful in some cases. Turns out that there are other ways of limiting the sharpening that get close, but don't cause as many problems.

Filters are really like drugs in that way. This one makes one ailment better, but has symptoms that are worse than the original problem.

-G

I don't think there will be any updates any time soon. The ringing appears to be different for different scenes. I can get rid of it on one scene, but then on another scene a completely different set of convolution coefficients are required. (the ringing and halos vary in that they appear at different distances from objects in different scenes)

Some of this work may have been well suited for your JSC capture DJ. :(

-G