You guys are right, it all looked slightly too green.
I made some adjustments, and here's compared old to new:
I'll update with more screenshots later, and pittrek, I'll upload screenshots of those as well. :)
By the way, here's a comparison of Greedo from untouched GOUT vs my new settings:
Thank you so much dark_jedi, what you said means a lot.
Now it's going to be very interesting to hear what everyone else thinks about it!
Here's how that shot looks with this script:
You have to keep in mind though, that those technicolor pics are photos from a camera, and can just give hints on how it should look.
Also, if I matched this scene completely to that photo, some other parts of the movie would look worse.
By the way, here's a screenshot comparison which shows how I tried to restore the yellows, like 3PO's golden color.
bkev said:
I like this a lot more than I did your blu color corrections. It might boil down to personal taste, though - plus, you have a different goal with the GOUT than you do with the 2011. On the GOUT you're just going for the most pleasing look, yeah?
Actually, the goal is quite the same. I'm trying to fix the most obvious color errors and of course make it look more pleasing and balanced overall.
Just like the blu-ray, this one would require a scene by scene correction if aiming for complete balance and accuracy, but I think even here it makes a great improvement even with these settings overall.
One setting in particular that was very useful was SatMask, the long in-script function. It works like vibrance in photoshop: It increases the saturation of the least saturated colors in the picture, and since the GOUT is very desaturated overall, it did wonders to it.
bkev said:
This generally reminds me of LeeThorogoods earlier project, which I never did get to watch due to newsgroups...
Actually I never got to watch that one either, but from the screenshots I saw of the final version, that one looked much brighter, maybe to the point where whites got clipped.
Darth Tater said:
I think it looks good so far, You_too! Is there any way a newbie like me with a digital copy of the GOUT discs can apply these settings and watch at home, either in advance or on the fly? Free is the key here. :)
You could apply the settings and watch through an avisynth script in your computer, but it requires some plugins as well.
Plugins used are MVTools or MVTools 2, (I don't remember which one but you could get both to be sure it will work) Gicocu and Selective color. (Once again a huge thanks to Darth Editous for the selective color plugin!)
If you use the raw video of Star Wars, then it's in mpeg2 and needs to be indexed, which can be done in file indexer in Megui. That will create the d2v file that the script uses.
For Gicocu you'll need my curves files which you put in the same folder as the movie, d2v file and script.
And finally, here's the script which is for the PAL-version. For NTSC you'd have to use different values for crop.
UPDATE: Two scripts, first one for SW, second for ESB and ROTJ.
Mpeg2Source("GOUT.d2v")
crop( 0, 124, 0, -124)
ConvertToYV12()
function SatMask(clip c, "Sat", float "bias", string "mode", clip "Rc", bool "lsb", bool "show"){
bias = string(default(bias, 5))
mode = string(default(mode, "Vibrance"))
lsb = default(lsb, false)
show = default(show, false)
lsb ? c.ditherpost(mode=6) : cw=width()
h=height()
Assert(isclip(Sat) || isint(Sat) || isfloat(Sat), "'Sat' only accepts float, integers or clip inputs")
Assert(w%4 == 0, "width and height must be mod 4 resolution")
Assert(h%4 == 0, "width and height must be mod 4 resolution")fmode = isclip(Sat) ? false : (Sat>0.0 ? true : false)
a= mode=="Recover" ? mt_lutxy(Vtoy(Rc),Utoy(Rc),"x 128 - 2 ^ y 128 - 2 ^ + .5 ^ 128 > 255 x 128 - 2 ^ y 128 - 2 ^ + .5 ^ "+bias+" * ?",u=1,v=1) : nop
b= !fmode || show || mode=="Recover" ? mt_lutxy(Vtoy(),Utoy(), "x 128 - 2 ^ y 128 - 2 ^ + .5 ^ 128 > 255 x 128 - 2 ^ y 128 - 2 ^ + .5 ^ "+bias+" * ?",u=1,v=1) : nop
msk = !fmode || show ? ((mode=="Vibrance") ? mt_lutxy(Vtoy(),Utoy(),"x 128 - 2 ^ y 128 - 2 ^ + .5 ^ 128 > 0 128 x 128 - 2 ^ y 128 - 2 ^ + .5 ^ - 128 / "+bias+" ^ 255 * ? ",u=1,v=1) : \
(mode=="Dullness") ? b : \
(mode=="Recover") ? mt_lutxy(a,b,"x y - abs 2 *",u=-128,v=-128) : \
Assert(false, "Wrong Saturation Mode")) : nop
show ? (lsb ? msk.Spline36Resize(w,h).Dither_convert_8_to_16() :\
msk.Spline36Resize(w,h)) :\
(Defined(Sat) || fmode ? \
((mode=="Vibrance") ? \
(lsb ? (fmode ? mergeluma(Dither_lutxy8(SwapUV(),"x 128 - 2 ^ y 128 - 2 ^ + .5 ^ 256 * 32768 > x 256 * "+
\+" x 128 - 128 x 128 - 2 ^ y 128 - 2 ^ + .5 ^ - 128 / "+bias+" ^ "+string(sat-1)+
\+" * 1 + * 128 + 256 * ?",y=2,u=3,v=3),c) :\
Dither_merge16(c,Sat,YToUV(msk,msk).Dither_convert_8_to_16,y=4)) :\
(fmode ? mt_lutxy(SwapUV(),"x 128 - 2 ^ y 128 - 2 ^ + .5 ^ 128 > x "+
\+" x 128 - 128 x 128 - 2 ^ y 128 - 2 ^ + .5 ^ - 128 / "+bias+" ^ "+string(sat-1)+
\+" * 1 + * 128 + ?",y=2,u=3,v=3) :\
mt_merge(Sat,YToUV(msk,msk),y=4,u=3,v=3))) :\
(mode=="Dullness") ? \
(lsb ? (fmode ? mergeluma(Dither_lutxy8(SwapUV(),"x 128 - 2 ^ y 128 - 2 ^ + .5 ^ 256 * 32768 > x 256 * "+
\+" x 128 - x 128 - 2 ^ y 128 - 2 ^ + .5 ^ "+string(sat-1)+
\+" * "+bias+" / 1 + / 128 + 256 * ?",y=2,u=3,v=3),c) :\
Dither_merge16(c,Sat,YToUV(msk,msk).Dither_convert_8_to_16,y=4)) :\
(fmode ? mt_lutxy(SwapUV(),"x 128 - 2 ^ y 128 - 2 ^ + .5 ^ 128 > x "+
\+" x 128 - x 128 - 2 ^ y 128 - 2 ^ + .5 ^ "+string(sat-1)+
\+" * "+bias+" / 1 + / 128 + ?",y=2,u=3,v=3) :\
mt_merge(Sat,YToUV(msk,msk),y=4,u=3,v=3))) :\
(mode=="Recover") ? \
(lsb ? (fmode ? mergeluma(Dither_lutxy8(SwapUV(),"x 128 - 2 ^ y 128 - 2 ^ + .5 ^ 256 * 32768 > x 256 * "+
\+" x 128 - 128 x 128 - 2 ^ y 128 - 2 ^ + .5 ^ - 128 / "+bias+" ^ "+string(sat-1)+
\+" * 1 + * 128 + 256 * ?",y=2,u=3,v=3),c) :\
Dither_merge16(c,Sat,YToUV(msk,msk).Dither_convert_8_to_16,y=4)) :\
(fmode ? mt_lutxy(SwapUV(),"x 128 - 2 ^ y 128 - 2 ^ + .5 ^ 128 > x "+
\+" x 128 - 128 x 128 - 2 ^ y 128 - 2 ^ + .5 ^ - 128 / "+bias+" ^ "+string(sat-1)+
\+" * 1 + * 128 + ?",y=2,u=3,v=3) :\
mt_merge(Sat,YToUV(msk,msk),y=4,u=3,v=3))): nop ) : c)}Tweak(hue=-5)
SatMask(Sat=2.25)ConvertToRGB32()
GiCoCu("GOUT.cur")
selectivecolour(last,\
0.65,0,0,\
-0.35,0.10,-0.05,\
0,0.75,-0.05,\
0,0,0,\
0,0,0.75,\
-0.25,0,0,\
false)GiCoCu("GOUT3.cur")
GiCoCu("GOUT5.cur")
GiCoCu("GOUT8.cur")
GiCoCu("GOUT9.cur")
ConvertToYV12()
crop( 0, 0, 0, -2)
BicubicResize(768,326,0,0.5) # Bicubic (Neutral)
Mpeg2Source("ESB.d2v")
crop( 0, 124, 0, -124)
ConvertToYV12()
function SatMask(clip c, "Sat", float "bias", string "mode", clip "Rc", bool "lsb", bool "show"){
bias = string(default(bias, 5))
mode = string(default(mode, "Vibrance"))
lsb = default(lsb, false)
show = default(show, false)
lsb ? c.ditherpost(mode=6) : cw=width()
h=height()
Assert(isclip(Sat) || isint(Sat) || isfloat(Sat), "'Sat' only accepts float, integers or clip inputs")
Assert(w%4 == 0, "width and height must be mod 4 resolution")
Assert(h%4 == 0, "width and height must be mod 4 resolution")fmode = isclip(Sat) ? false : (Sat>0.0 ? true : false)
a= mode=="Recover" ? mt_lutxy(Vtoy(Rc),Utoy(Rc),"x 128 - 2 ^ y 128 - 2 ^ + .5 ^ 128 > 255 x 128 - 2 ^ y 128 - 2 ^ + .5 ^ "+bias+" * ?",u=1,v=1) : nop
b= !fmode || show || mode=="Recover" ? mt_lutxy(Vtoy(),Utoy(), "x 128 - 2 ^ y 128 - 2 ^ + .5 ^ 128 > 255 x 128 - 2 ^ y 128 - 2 ^ + .5 ^ "+bias+" * ?",u=1,v=1) : nop
msk = !fmode || show ? ((mode=="Vibrance") ? mt_lutxy(Vtoy(),Utoy(),"x 128 - 2 ^ y 128 - 2 ^ + .5 ^ 128 > 0 128 x 128 - 2 ^ y 128 - 2 ^ + .5 ^ - 128 / "+bias+" ^ 255 * ? ",u=1,v=1) : \
(mode=="Dullness") ? b : \
(mode=="Recover") ? mt_lutxy(a,b,"x y - abs 2 *",u=-128,v=-128) : \
Assert(false, "Wrong Saturation Mode")) : nop
show ? (lsb ? msk.Spline36Resize(w,h).Dither_convert_8_to_16() :\
msk.Spline36Resize(w,h)) :\
(Defined(Sat) || fmode ? \
((mode=="Vibrance") ? \
(lsb ? (fmode ? mergeluma(Dither_lutxy8(SwapUV(),"x 128 - 2 ^ y 128 - 2 ^ + .5 ^ 256 * 32768 > x 256 * "+
\+" x 128 - 128 x 128 - 2 ^ y 128 - 2 ^ + .5 ^ - 128 / "+bias+" ^ "+string(sat-1)+
\+" * 1 + * 128 + 256 * ?",y=2,u=3,v=3),c) :\
Dither_merge16(c,Sat,YToUV(msk,msk).Dither_convert_8_to_16,y=4)) :\
(fmode ? mt_lutxy(SwapUV(),"x 128 - 2 ^ y 128 - 2 ^ + .5 ^ 128 > x "+
\+" x 128 - 128 x 128 - 2 ^ y 128 - 2 ^ + .5 ^ - 128 / "+bias+" ^ "+string(sat-1)+
\+" * 1 + * 128 + ?",y=2,u=3,v=3) :\
mt_merge(Sat,YToUV(msk,msk),y=4,u=3,v=3))) :\
(mode=="Dullness") ? \
(lsb ? (fmode ? mergeluma(Dither_lutxy8(SwapUV(),"x 128 - 2 ^ y 128 - 2 ^ + .5 ^ 256 * 32768 > x 256 * "+
\+" x 128 - x 128 - 2 ^ y 128 - 2 ^ + .5 ^ "+string(sat-1)+
\+" * "+bias+" / 1 + / 128 + 256 * ?",y=2,u=3,v=3),c) :\
Dither_merge16(c,Sat,YToUV(msk,msk).Dither_convert_8_to_16,y=4)) :\
(fmode ? mt_lutxy(SwapUV(),"x 128 - 2 ^ y 128 - 2 ^ + .5 ^ 128 > x "+
\+" x 128 - x 128 - 2 ^ y 128 - 2 ^ + .5 ^ "+string(sat-1)+
\+" * "+bias+" / 1 + / 128 + ?",y=2,u=3,v=3) :\
mt_merge(Sat,YToUV(msk,msk),y=4,u=3,v=3))) :\
(mode=="Recover") ? \
(lsb ? (fmode ? mergeluma(Dither_lutxy8(SwapUV(),"x 128 - 2 ^ y 128 - 2 ^ + .5 ^ 256 * 32768 > x 256 * "+
\+" x 128 - 128 x 128 - 2 ^ y 128 - 2 ^ + .5 ^ - 128 / "+bias+" ^ "+string(sat-1)+
\+" * 1 + * 128 + 256 * ?",y=2,u=3,v=3),c) :\
Dither_merge16(c,Sat,YToUV(msk,msk).Dither_convert_8_to_16,y=4)) :\
(fmode ? mt_lutxy(SwapUV(),"x 128 - 2 ^ y 128 - 2 ^ + .5 ^ 128 > x "+
\+" x 128 - 128 x 128 - 2 ^ y 128 - 2 ^ + .5 ^ - 128 / "+bias+" ^ "+string(sat-1)+
\+" * 1 + * 128 + ?",y=2,u=3,v=3) :\
mt_merge(Sat,YToUV(msk,msk),y=4,u=3,v=3))): nop ) : c)}
Tweak(hue=-5)
SatMask(Sat=1.75)ConvertToRGB32()
GiCoCu("GOUT.cur")
selectivecolour(last,\
0.65,0,0,\
-0.35,0.10,-0.05,\
0,0.75,-0.05,\
0,0,0,\
0,0,0.75,\
-0.25,0,0,\
false)GiCoCu("GOUT3.cur")
GiCoCu("GOUT5.cur")
GiCoCu("GOUT8.cur")
GiCoCu("GOUT9.cur")
ConvertToYV12()
crop( 0, 0, 0, -2)
BicubicResize(768,326,0,0.5) # Bicubic (Neutral)
I thought I’d make a new thread for this, since it’s not about the blu-rays like my other thread.
What I’ve done is I tried the same thing I did with the Star Wars blu-ray, but with the GOUT-version instead, that is, I tried to make a good “one-setting” for the whole movie.
I know the GOUT is not a very good source when it comes to picture quality, but it also has a lot of things the blu-rays don’t have, like intact shadow detail and much less tweaked colors.
I’m using the PAL-version by the way.
Now you guys will be the judges whether I should continue with this or not. If people are interested, I might continue to try and find the best overall settings for the GOUT. Keep in mind that the colors are inconsistent throughout the movie, so before you say “this looks too yellow” or “that looks too blue”, consider how much that changes with the scenes. Of course you’re still free to point things out though.
My current result looks like this. (Click for full-size)
UPDATE: Now all three movies!
http://img840.imageshack.us/img840/8121/esb3red2thumbs.png](http://img820.imageshack.us/img820/5592/esb3red2.jpg “img820.imageshack.us/img820/5592/esb3red2.jpg”)
http://img225.imageshack.us/img225/9316/rotj3red2thumbs.png](http://img221.imageshack.us/img221/4451/rotj3red2.jpg “img221.imageshack.us/img221/4451/rotj3red2.jpg”)
Harmy said:
Yeah, the xmas release is going to be impossible now, sorry. I clearly underestimated the amount of work that was still to be done. I don't even want to give another estimate, since it would really be just a wild guess at this point.
It's better to not give an estimate, but take your time and keep polishing it until you're 100% satisfied, and I think most people here agree on that.
Harmy said:
On a more positive note, I will hopefully soon post the next workprint within the next few days. It will actually be the first hour or so, from the beginning, because I made a lot of changes based on some of the observations made by you guys, so it's possible I could have made some new mistakes.
Looking forward to the next WP! :)
It's also worth noting how much better everything looks when in motion compared to when looking at a still. :)
OMEN!-_-! said:
Point taken, You_Too. I stand corrected. :) I often mix up the effects of too much green and too much yellow to be honest. I should use photoshop more to get an indicator of balanced colours like you did here.
Photoshop is usually the best tool there is to be able to find such things. At least for me.
Then of course, colors are and will always be a matter of taste too. But from what I've seen so far, Harmy always tries to create the best balance using the few references we have.
Harmy said:
Yeah, sure, it looks a bit green when you compare it to that but now the walls are a bit too purple.
I loosely based my colour timing of that scene on this picture:
I know it may not be entirely correct but it seems to be the best reference we've got right now.
Yeah, when one thing is right the others look wrong and vice versa. I think you're doing a great job on balancing it anyway. I just posted that pic to show people that his skin had too much green, not too much yellow in it.
Actually, the color that needs to be just slightly reduced to make Tarkin closer to natural would be green.
Look here, I used curves in photoshop and put a point in the green curve at 127/135 and here's what came out:
I'll never understand why the movie studios think they improve the picture with DNR and EE. It's horrible.
Found a subtle color inconsistency which is on the raw blu-ray as well, but I thought I'd post it anyway.
Look at the blue on R2:
Is it like this in the HDTV version too?
Harmy said:
Yeah, unfortunately, at the same time this resulted in even more white clipping (look at the reflection in R2's dome). But it's just a little and I feel that comparatively it is a small price to pay for all the detail recovered in the dark areas.
I'm glad you also think it was a small price. And actually it's not that much clipping, what I did was to push the highs closer to white to make the movie look closer to the original. The shades of white being closer to white sometimes creates an illusion of them being clipped. The raw blu-ray is badly underexposed so I thought it was necessary.
Actually, when examining that brightest part of those screenshots in photoshop, the only color which peaks at 255 is blue, so if something was clipped it was some blues in the middle of the brightest part. The rest was just pushed up towards white and that's why it looks clipped when it's not.
I agree with vbangle, V1 is darker and also less saturated. V2 looks better.
You've edited that one nicely. I would never have guessed it was cut together from two sources if I didn't know.
I continued watching the WP a bit today, and found some more stuff.
First there was some inconsistency, the first one being a bit brighter:
Not extremely important though, since the shots doesn't come right after each other.
Then there was this frame which looks kinda weird:
Is this the same kind of thing that I helped you with in Greedo's scene? Like some frame missing and needed to be blended? In that case I could render a new frame for you if you want.
Harmy said:
Oh, yeah it's really hard to align the SE and the GOUT perfectly, because the GOUT is warped, so these things happen sometimes. You'd never notice it in movement (unless you were specifically looking for it of course).
Ah, yes I understand that now. And you're right I would never notice it when it's moving, and I guess that's what really counts. I'd rather have such small unnoticeable things here and there than a big Ronto anyway. :)
Harmy, when looking closer on that shot-comparison there, I noticed something weird when looking at the mid-left part of the speeder.
It's like the image doesn't line up with itself in the second one.
That shot looks great! A bit more detailed, more color-correct feeling overall, and the stormtrooper's backpack is not crushed black with blue/cyan tint anymore. Good job! :)
Btw, I haven't had the time yet to watch the whole WP but I did watch the end just because you said there was a surprise. Obi-Wan sure doesn't like Mos Eisley...
Anyway, something I noticed was that on the Mos Eisley matte painting, the color of the mountains doesn't match the shot just before it. I don't know how it is in the GOUT and other versions but just thought I'd let you know.
Harmy said:
Also, at first I only saw that square as white but when I focused and took a closer look, I can clearly see those tints. Should it be obvious at first sight on a properly calibrated monitor?
Even on a calibrated monitor it can sometimes be hard for the human eye to see the difference between shades of white like that, and that's one reason why color correction can be difficult when it comes to shades of white, especially if trying to find a neutral look.
Oh, and good news about the lasers!
Harmy said:
Well, I never had any professional calibration done, since it's hard to do anything like that on a laptop but it seems to generally show what most people agree on.
A true professional calibration is probably impossible yes, since most laptops lack advanced setup properties for the screen itself, but a normal fast auto-calibration using a colorimeter works fine, I've done it on a laptop myself.
Harmy said:
EDIT: Well, according to this article, my monitor is quite well calibrated but it has terrible viewing angle and bad response time.
Actually that's basic setup, not really calibration. Calibration is when you have a colorimeter or equivalent on your screen which measures the greyscale and colors and sets them up so they display as close as possible to a reference, in this case Rec.709 since it's HD film material we're working with.
If you're interested, google for X-rite i1 Display, it's one of the best price vs performance brands.
Harmy said:
I think the way you see colours also greatly depends at how long have you already been staring at your monitor, sometimes I colour-grade something in the evening and it looks fine and then I look again in the morning and it looks off.
I agree. If I work on a photo or video with changing or correcting colors for too long, I get slightly "colorblind". It's always good to leave it for a while and take a look again.
Harmy said:
Another thing is that letting photoshop balance the image will probably make it look more natural, which isn't always necessarily correct for the film.
Yeah, I just did that to fix the earlier "mistake" of posting a photo that was not color balanced. Just like have been stated before, even the original prints are inconsistent with color, so naturally it's up to you as the "director" to choose how you want it to look. I just thought maybe the photo would be good for seeing how the walls and skintones really looked.
OMEN!-_-! said:
Your whites are still a little yellow there You_Too (at least on my monitor) but I like the flesh tones here the most of the three so it definately gets a thumbs up from me. Has your monitor been colour calibrated as well?
I calibrated my monitor using an X-rite i1 Display 2, not too long ago, which I bought used, mainly to calibrate my TV but I also did the computer monitor and laptop while I was at it. Anyway, like I said I didn't do that color correction by eye but let photoshop auto-balance it. I took a point sample on the wall just in front of Vader's eyes and it came out RGB 247/247/247. And I think it looks pretty neutral overall now too.
Either way, the eye can have a hard time seeing a difference between shades of white in a picture. Look here:
This picture is divided into 4 squares, the upper left being tinted red, upper right green, lower left blue and lower right pure white. I bet on some monitors it's even impossible to see the difference because of the monitor itself.
Harmy said:
And Yoo_Too, you had me believe that those pictures I reposted above were from two different shots, whereas actually they're from one and the colour-timing of the one on the left is only there for the first two frames just before a huge blinding flash, so it is virtually unnoticeable unless viewed frame by frame.
Actually, the tint in the first one begins with this shot:
Then it changes at the shot I showed you and continues when 3PO and R2 run across the screen.
OMEN!-_-! said:
You_Too, I personally think the walls, the white light on the ceiling (yellow light in this case) and flesh tones are too yellow in this still and don't look 'natural' IMHO.
EDIT:
Just regraded this shot quickly so it looks more natural, at least to me. The walls/light look fairly white to me now, but the yellow tint is still noticable, annoyingly. Their faces could perhaps use a bit more red to make the flesh tones more natural. Saturation might also need to be boosted slightly. Anyway, definate improvement IMHO:
After a closer look and point sampling in photoshop, I can only agree. It was slightly too yellow. Your regraded version is slightly too blue though. So anyway, I did a final version, letting photoshop auto-balance the levels and now the RGB levels for white are very close. I also increased saturation slightly. So this will probably be the cleanest look for this image.
And Harmy, sorry for spamming your thread with pictures!
I still think you're doing a great job, and that you probably put a lot of thought behind your color-decisions. I just point out things to make sure you'll be able to make this the optimal restoration of the movie using the sources we have.
Harmy said:
@You_Too: Thank you for these observations, I must say that most of thm differences are fairly minor - I mean, I do see them but just barely and considering the time it took to balance these shots out to their current look, I think I'll leave most (not all) of the as they are now. But I'm definitely not saying that they're out of place and I'll sure look into what I can do about them.
First, just a question, is your monitor calibrated? I mean, you say you see the things I pointed out but just barely. No offense!
Some of the things are less important of course, but make sure you at least fix the part where the picture goes darker in an eyeblink after R2 and 3PO runs across the screen, since it's a very visible and fast change.
And in my opinion the change in this point is very visible too, especially on a calibrated screen:
By the way, about wall and skin tones, I'd say this production photo shows almost what it probably looked like, if aiming for natural tones:
