mverta

Thanks - so glad you are enjoying the work!

Yes it's an artifact, but it's usually triggered in proportion to saturation.  Not always, but often.  

Disney would much more likely take poor sales as an indication to abandon effort on the originals and focus on new stuff, rather than "learn their lesson," eat crow, and put out originals.  That's just corporation 101.

I don't even...

That's sRGB not playing well with DCI-P3.  The actual image is more saturated, definitely pure red, not magenta, etc.  

Happy Easter.

Speaking of Lord Christ Almighty... take a look at this shot:

R6S001

_Mike

The prints are cleaned ultrasonically, but still you have to be gentle.  With the IR dirtmap from the scan and multiple prints, the digital clean-up is tedious, but seamless.

_Mike

X-Wings Dive towards the Death Star

_Mike

Well that makes sense, inasmuch as the whole thing is a joke.

In 25 years as a professional, I've found labeling people "professional" means very little.

Most likely.  I don't see mixing of sources; certainly not combining of sources.

We know the ON is incomplete at best, and I do not believe they have intact 3-strip masters, at all.  There are WAY too many lack-of-detail and/or suspect internal color problem shots present in every official restoration for me to believe that.  If they have them, then somebody should be shot, because you'd have to be trying to fuck things up that badly.

Another shot:

Falcon Peels Off

It's got a luminance error in it, since corrected.  Something invisible in P3 which popped right out as soon as I put it online, but I didn't feel like remaking the video. :)

Well certainly the technique and process would be the best way to go sure; recruiting data from several of the best available print sources.  

Splicing/glue marks are a problem inherent to anamorphic film because it uses so much of the horizontal area.  To combat the flashes often seen at edits, projectionists began using a projector aperture gate that was narrower in the vertical, meaning that when projected Star Wars was actually 2.39-2.4. 

As for the other damage in those frames, normally I'd remove them!  Also, this sort of thing is precisely what the Legacy forum is for, so no worries there!

_Mike

g-force said:

mverta said:

Tie on Luke

Careful Mike. On this one on the upper left strut of the canopy, the original has a couple white dots that appear for one frame each, and on your restoration they now fade in and then fade out. That's always a risk with temporal averaging, and you might want to look into a temporal median instead.

-G

 Good catch. I don't use strictly median or averaging, but a dynamic recruit based on motion and comparative analysis across sources.  In this case, I had both limited sources and misjudged the threshold setting.  Sometimes when there are areas of low motion and areas of high motion in the same frame, I have to be more careful.  Notice there are no artifacts on the fast-moving X-Wing, for example.  So it got one end of the spectrum right, and not the other.  It's been repaired.  Always appreciate the eyes on it!

Yeah, and actually this is the first time I've routinely put DCI-P3 material online in sRGB, and I realize I hadn't appreciated just how different they are - and in funny sort of ways.  There's this area in the low-mids that just gets boosted in the conversion, which takes a lot of the contrast/saturation with it, but it's non-linear.  Like, in many of these space shots, you can see tone in the black but the actual image is pure black - I see stuff show up in there when I post it online that is not on my monitor, reference monitor, or projector.  It's gotten so that I set up an extra node which boosts things in that range just so I can make sure there's nothing ugly to be revealed later.  Weird.  But still, yes, these images have basically always looked like shit, so even these compromised restorations look tons better.

Believe it or not, it's actually dead-sexier than it looks online.  The DCI-P3 > sRGB conversion just kills me.  It's actually contrasty-er and more saturated for real than it appears.  When I'm done, I'm going to do a Rec.709 grade which I can pull stills from, which will look more like they actually do.  But still...!

Another iconic shot:

S-Foils in Attack Position

_Mike

Cool.  Yes, that is looking familiar!

I can't tell you the timecode (mine is based on reels, frames, 24fps) but it's just after the Falcon has been sucked in by the tractor beam.

The Look of Star Wars - Color Comparison #2

_Mike

BuddhaMaster said:

@ MVerta. Your restore looks phenomenal as usual. First I thought 3CPO is a bit bright and blunt, but looking at it for a moment. Phenomenal. I love how you always manage to make the pictures look 3D. Even more depth than actual 3D.

 Thanks! This shot is WIP, and even though it's obviously a step up, I can feel it's got another level in it.  You'll see what I mean in the final, but there's a flat spot in its luminance near the mid-lows which when fixed will bring out some nice things. There is a touch of cyan in it, too, which I can see in R2s blue, and a few other things.  It's amazing how delicate the adjustments can/need to be to really nail a shot.  But I believe in this sort-of "70%" spot where if you can hit it 70% right, quickly, you're on the right track, and then it's just finesse after that - as much a difference as it may make. It's just this vibe/flavor thing, which if you don't have it, all the details in the world don't make it.  It's actually fun to feel it come together!

The Look of Star Wars - Color Comparison #2

A comparison of grading on the 2006 DVD, Blu-Ray, Reliance showreel, and Legacy.

_Mike

Yep!

EDIT: Well, wait.. are you saying the channel alignment of the stars (which I've done) or the bleed through onto the cockpit window frame, which has always been part of the image as a composite limitation?  If the latter, then no.  I don't improve composites in Legacy.  I don't paint out matte lines, etc. Wherever possible, I am not looking to do another Special Edition.  If the stars bleed through the foreground cockpit element, then that stays. If the individual channels are misaligned due to print anomalies, they are corrected. If channels are misaligned due to lens aberration, then they stay.  When I'm not sure where the channel alignment issue is coming from, they're corrected just to be safe.  This isn't an alignment issue, it's a compositing issue inherent to limitations in the process in 1977.

_Mike

Not only that - look closely and you'll see that they are totally misaligned with the engine by the end of the shot!