[Test] Aliens open matte

[spoRv]

Maybe the one you are referring to is simply the WOWOW - that often was referred simply as HDTV - and it's big indeed, almost 25GB Vs 8.5GB of the Channel 1 (don't know exact name, it has a "1 HD" logo on top right corner); this showed up lately - at least in my vault!

[AntcuFaalb]

I haven't been able to find one larger than ~10GB. If you happen to come across a hyperlink to it I'd greatly appreciate if you could PM it to me. Thanks!

[Chewtobacca]

No, I don't think it's the WOWOW, but it is big. I had it in the same place as the Alien d-theater, but it's probably gone. I normally rar stuff and archive it, but it takes ages to look through all my drives. I'll look.

EDIT: Nope! Gone for good...

[spoRv]

AntcuFaalb: I'll do!

Chewtobacca: will be great to have even another source - the more, the merrier!

[BronzeTitan]

... explain the advantage of degraining the three channels separately... would you use different settings for each channel?

Yes, working on each channel has the advantage of targeting specific problems to particular channel(s), which generally works best for standard film scans. In the case of Aliens, their obviously artificial regrain is not too different from channel to channel, so adjustments will be close. (Every little bit helps.) However, the blue channel picture generally looks worse than the other two (typical for old/worn film). It can be more heavily smoothed to compensate for that problem without over-processing the other channels to do it.

For this test, I used an Edge Preserving Smooth (in a paint program) as the representative effect of de-emphasizing Aliens's "grain" while minimizing general blurriness (hard to do). Same with the sharpen -- Unsharp Mask is just one type available in the paint program that has adjustments. But you should use whatever processes you know work best for these types of fixes.

[spoRv]

Avisynth filters, please...

[BronzeTitan]

Well, for sharpening, I think this is the best (based on it's theory of operation) ..

Doom9 forum: Here is LimitedSharpen()  http://forum.doom9.org/showthread.php?threadid=84196
This is the development thread of his script that runs as a function. He has some updates added on the first page.

Here's the graphics now missing from the page, but still on Internet Archive's 2005 capture, with context:

Internet Archive (2005 record): Doom9 forum - Here is LimitedSharpen() - page 1

Let's have a look at a simple transition from dark to bright, and what our standard sharpeners will do to it:

The blue line represents the original edge, the dark side on the left, the bright side on the right, and in-between the gradient that builds "the edge". Red is (basically) the result of sharpen() or UnsharpMask(), pink is the result of XSharpen().

...

Shortly, LimitedSharpen() applies one out of three different sharpeners (two domain sharpeners or a windowed range sharpener) to the source, but will limit the oversharpening (either "hard" or "soft") IF it exceeds a defined "overshoot".

In reference to the graphs above, the script's results look like that (basically) :

(The proportions are not "real" - both graphs were constructed free-handed.)

As you see, LimitedSharpen always takes the enhanced edge steepyness from normal sharpening, but avoids oversharpening in the same way as XSharpen, as long as overshoot is kept at zero, and limiting mode 1 is used.
In many cases, this comes out really nice, and is sufficient.

Avisynth has an official page:

Avisynth: LimitedSharpen  http://avisynth.nl/index.php/LimitedSharpen
This is the up-to-date version (now called "faster") so follow it's instructions for setup.

- - -

Regarding denoisers & smoothers, I'm not up to date on those. Previously I used a spacial/temporal (with tracking) smoother but I'm sure things have improved since then. The support-downloads for LimitedSharpenFast are packages of sub-function denoisers, which may be independently called. But I've never tried them that way. So I can't recommend anything just now.

However, whatever you use, you can independently process R/G/B separations using Avisynth's ..

ShowBlue (clip, string pixel_type)
ShowGreen (clip, string pixel_type)
ShowRed (clip, string pixel_type)

.. and rejoin them with ..

MergeRGB (clipR, clipG, clipB [, string "pixel_type"])

(see Avisynth docs for detailed usage)

 I hope this helps!

[spoRv]

Aliens open matte test clip #1: https://we.tl/WqiNF5IuW6 (906MB, link expiration date 2017-06-23)

Known problems:
first part of the shots inside the rescue ship has top slice darker
horizontal pan over the Earth is stuttering - my fault

Please download and review; apart the previous noted one, the other slices seems consistent with the rest of the picture, in colors, brightness, grain - I've used the new grain plate from the very same film.

Waiting for your opinions!

EDIT: forgot to mention, the color grading is provisional.

[spoRv]

Further 23m30s test clip produced - up to about 40m in the film, or roughly 1/5th.

Apart one shot where open matte was shifted laterally, and an handful of occasions where brightness difference between slices and central parts are noticeable - I remember three - the rest is practically perfect!

Very proud of this project, not because it would gain 4% of the image - even if it's nice to get it - but because it would be the first one using the overlap technique (AFAIK).

[spoRv]

Tests continue, quality improved.... http://screenshotcomparison.com/comparison/213025