[Released] Harry Potter saga - open matte

[spoRv]

New Nagini scene without subs:
https://mega.nz/#!VS5VhRQI!1-irgaH4E6CXd...hUOVhwcpgA
this time I can barely see the traces of the deleted subtitles, but only if I crank up the brightness A LOT!
Watching it with "normal" settings (read: without the possibility to have retina burned by intense whites), I think it would not be noticeable; plus, subtitles in various language would be placed just over there.

@Tagg: the forced subs included in the BD are placed on top of the image; what I want is to have them carefully placed just over the deleted ones, and also with similar font and color. I have no problem to generate them, what I have to test yet is a way to have these subs forced, so that for any given language, the related subs would be automatically selected. I've read something about BDsup2sub, that is able to set a subtitle track as forced, but still have to try it.

[deleted user]

Here you go. This is the one as I used it for a Blu Ray source, just copy pasted. Might need to tweak the opacity value.

Might also want to look at the degrained only and tune thSAD until the artifacts/grain are gone before you use the combined one. The source I used was rather grainy, so you may need a different value.

Code:

src = FFmpegSource2("yourfile.mkv")

loadplugin("C:\Program Files (x86)\AviSynth 2.5\plugins\mvtools2.dll")
srcfordegrain=src
src = ConvertToRGB(src)

src

assumefieldbased()
bla = FQRestore(line=false,fr=10,x=2,y=0,wn=0.15,scale=0.44)

super = MSuper(srcfordegrain,pel=2, sharp=1)
backward_vec1 = MAnalyse(super, isb = true, delta = 1, overlap=4)
backward_vec2 = MAnalyse(super, isb = true, delta = 2, overlap=4)
# backward_vec3 = MAnalyse(super, isb = true, delta = 3, overlap=4)
forward_vec1 = MAnalyse(super, isb = false, delta = 1, overlap=4)
forward_vec2 = MAnalyse(super, isb = false, delta = 2, overlap=4)
# forward_vec3 = MAnalyse(super, isb = false, delta = 3, overlap=4)
degrained = MDegrain2(srcfordegrain, super, backward_vec1,forward_vec1,backward_vec2,forward_vec2,thSAD=400,limit=255)
#degrained = MDegrain1(super, backward_vec1,forward_vec1,thSAD=600,limit=255)

final = Overlay(degrained, bla,opacity=0.3).ConvertToRGB
#degrained
#src

#This line is just for direct comparison
StackHorizontal(Crop(src,0,200,800,0),Crop(final, 0,200,800,0))


#Crop(500,200,0,0)

[spoRv]

Thanks for the script, Tom!

FQRestore is not useful in this case; your MVDegrain script is indeed close to mine, and the result is really really close; they both improve the quality of a single source, still the combination of both sources (with just deblock_qed applied to each) improves the quality a lot; adding the MVDegrain to each source or to the combination of them lead to almost imperceptible quality improvement, versus a longer encoding time (probably three days more!), and I'm pretty sure that, when encoded to x264, this little improvement would be vanished.

So, at the end, I would combine the two sources (TVA and Space) whenever possible, and eventually use MVDegrain when forced to use just one source.

UPDATE
comparison of TVA, Space, and a combination of both (adding deblock to each, and a 35mm grain plate at the end):
http://www.framecompare.com/screenshotco...n/09FNNN8U
to download each image:
https://postimg.cc/gallery/2xb9smxay
note: this is just a test, of course the TVA logo would not be present in the final version!

[Freddy]

So, at the end, I would combine the two sources (TVA and Space) whenever possible, and eventually use MVDegrain when forced to use just one source.

Hello again,

I've compared those three screenshots and I see what you're trying to do, but... with this combination and the fuzzy character of the Space shot, the combined shot still loses a lot of sharpness compared to the original TVA one. Of course it's a bit cleaner, but at the expense of finer detail.

[spoRv]

I've compared those three screenshots and I see what you're trying to do, but... with this combination and the fuzzy character of the Space shot, the combined shot still loses a lot of sharpness compared to the original TVA one. Of course it's a bit cleaner, but at the expense of finer detail.

Freddy, following your hint, I've sharpened both TVA and Space:

http://screenshotcomparison.com/comparison/125704

now TVA and the combined version has comparable sharpness, but as you can see, TVA perceived detail is reached thanks to MPEG-2 artefacts and coarse grain; try to magnify the image to see what I'm talking about.

I think that, encoding this in x264 at 30mbps average should preserve all the details and grain. Of course, each film would be released as BD-50 instead of BD-25. Hope this will make everyone happy - I know it will mean double download time, double size onto HDD and eventually more cost on BD-R, but I think this project would deserve all of this, right?

[Freddy]

Freddy, following your hint, I've sharpened both TVA and Space:

http://screenshotcomparison.com/comparison/125704

now TVA and the combined version has comparable sharpness, but as you can see, TVA perceived detail is reached thanks to MPEG-2 artefacts and coarse grain; try to magnify the image to see what I'm talking about.

Yes, I can see the artifacts in TVA, but the sharpened combined one shows edge enhancement that isn't in the TVA one. I'm just saying that when the second source (Space in this case) is that much worse than the first one (TVA), it could be better to try one of those filters you were talking about on just TVA, preserving more detail.

[spoRv]

TVA+deblock+degrain+grainplate Vs combined (TVA+deblock & Space+deblock+sharp)+grainplate:

http://screenshotcomparison.com/comparison/125709

I think the latter is a tiny bit better, as the degrain wipe out smallest details; without counting that it will take a lot less time to encode! But, again, difference is negligible, and could be appeciated only examining a single frame, magnified; when in motion, it would be inperceptible.

Going to finish to put TVA in sync with the BD now!

EDIT: added new comparisons
http://screenshotcomparison.com/comparison/125744
http://screenshotcomparison.com/comparison/125746
as you can see especially in the second one, using two sources helps more with the macroblocking problem. And not, adding degrain to one or both sources doesn't improve blocking, but only removes small details, and adds a lot of encoding time...

[spoRv]

UPDATES

HP8 plans: use TVA and Space combined to get the best picture (expected to be more than 90%, I guess around 95% of the whole movie except credits); where TVA has those huge and "shapeshifting" pop-ups, only Space (denoised) would be used; where both have missing frames/shots, Nova or Freeform would be used, using each other to delete logos.

• Things done: Nagini scene with subs; TVA, Nova, Freeform sync
  
• Things to do: Space sync, beginning scene to be improved using TVA
  
• ETA: few days
  

HP7 new plans: as now TVA and Nova 1080p are available, to get better result TVA would be used as main source, while the old final version (final no more, at this point!) would be used to remove TVA logo, and Nova 1080p would replace the old Nova 720p to get finer details; as usual, where there are the annoying huge pop-ups, the old final would be used.
All in all, a noticeable improvement is expected, of course only in the open matte shots (note: the cropped BD used for the rest of the movie is still better than TVA, and has also more images on all sides).

• Things to do: TVA and Nova sync
  
• ETA: few days (after HP8 end)
  

Still, new HDD is needed - the release versions would be around 350/450GB, while each temp file is around 150/200GB; if someone would like to donate few bucks, this is the right moment!

[spoRv]

UPDATE

A new first scene is now ready; I think it's very good now!
comparison: http://screenshotcomparison.com/comparison/125796
test clip: https://mega.nz/#!VWhVQDRK!XXDXJDWEhtYiA...rK_Lw8ZLXg

It was hard, one of the most complex projects so far, but it's done, it seems....
As expected, the amount of shots with both TVA and Space combined is around 93.5%.
Right now a test version of HP8 is encoding - at ultrafast it will take more than 18 hours!!!


Tomorrow evening I'll check it, to find out eventual problems - I hope it would not be the case, but who knows!

Question: Nagini scene with subs, would you prefer English hardsubs with foreign soft subs placed on top of them (with a black box as background), or only soft subs?

HP7 fix would not take so much time *I guess* and I'll start to work on it on monday.

[Beber]

The only worst thing than 2 sets of subtitles at the same time would be subtitles in a blackbox covering the original subtitles. So please, no.