logo Sign In

Three-strip Technicolor: Please help! — Page 2

Author
Time
 (Edited)

I just noticed this. WOW!

Comparison

Edit: It seems like the contrast is definitely too high now.

A picture is worth a thousand words. Post 102 is worth more.

I’m late to the party, but I think this is the best song. Enjoy!

—Teams Jetrell Fo 1, Jetrell Fo 2, and Jetrell Fo 3

Author
Time

Also, I just experimented and determined that applying the contrast-fix after Technicolor() preserves details in very bright scenes, but still has the desired effect.

A picture is worth a thousand words. Post 102 is worth more.

I’m late to the party, but I think this is the best song. Enjoy!

—Teams Jetrell Fo 1, Jetrell Fo 2, and Jetrell Fo 3

Author
Time

To me it looks like a split between the old and new is appropriate. Old was a bit too warm, new is definitiely a bit too cold. The Alderaan pic has a few tiny artifacts because of the video source but I would say it actually looks more pleasing because of the massive amount of detail added. Again though this is pretty subjective. Either way is an improvement on the GOUT.

Author
Time

Here's a GOUT-related question: Assuming that we don't make any color changes beforehand, how much do the reds need to be lowered to fit in with the rest of the colors? That is, what change would result in them no longer being over-represented? Is it just the reds?

A picture is worth a thousand words. Post 102 is worth more.

I’m late to the party, but I think this is the best song. Enjoy!

—Teams Jetrell Fo 1, Jetrell Fo 2, and Jetrell Fo 3

Author
Time
 (Edited)

AntcuFaalb said:

Here's a GOUT-related question: Assuming that we don't make any color changes beforehand, how much do the reds need to be lowered to fit in with the rest of the colors? That is, what change would result in them no longer being over-represented? Is it just the reds?

 

This is pretty much THE question for me. I saw some posts where someone indicated they could repair the red channel out of the RGB channels. My understanding is that the red shift is an organic shift in the film - its not the red channel is damaged but the greens and blues have faded leaving the red over represented. Anyone know how correct that is? If so can the green and blue be repaired?

edit: had a go at a blind ajustment to blue and green curves to find a result = http://originaltrilogy.com/forum/topic.cfm/Calling-all-Color-Correctors-Can-this-source-yield-a-different-set-of-results-to-Gout/post/598240/#TopicPost598240

 

Author
Time

frank678 said:

AntcuFaalb said:

Here's a GOUT-related question: Assuming that we don't make any color changes beforehand, how much do the reds need to be lowered to fit in with the rest of the colors? That is, what change would result in them no longer being over-represented? Is it just the reds?

 

This is pretty much THE question for me. I saw some posts where someone indicated they could repair the red channel out of the RGB channels. My understanding is that the red shift is an organic shift in the film - its not the red channel is damaged but the greens and blues have faded leaving the red over represented. Anyone know how correct that is? If so can the green and blue be repaired?

edit: had a go at a blind ajustment to blue and green curves to find a result = http://originaltrilogy.com/forum/topic.cfm/Calling-all-Color-Correctors-Can-this-source-yield-a-different-set-of-results-to-Gout/post/598240/#TopicPost598240

 

I asked You_Too to provide some examples of the over-represented reds (if possible) in my GOUT Bugs thread.

A picture is worth a thousand words. Post 102 is worth more.

I’m late to the party, but I think this is the best song. Enjoy!

—Teams Jetrell Fo 1, Jetrell Fo 2, and Jetrell Fo 3

Author
Time
 (Edited)

OK, so GOUT has over-represented reds. Challenge accepted!

Enter YellowMagic™ (a joke on Laserman's BlackMagic and _,,,^..^,,,_'s WhiteMagic).

I'm not claiming that these colors are correct, but I believe this may be right approach to de-emphasizing the reds in GOUT.

gout = gout.Tweak(cont=1.1, sat=1.3).ConvertToRGB24(matrix="Rec601")

gout.Overlay(gout.BlankClip(color=$D29628), mask=gout.ShowRed(), opacity=0.1, mode="lighten")

The $D29628 color was chosen because (1) it's yellowish, (2) it's close to -1's crawl text color, and (3) I didn't know what else to choose.

Pay close attention to skin-tones. Also, note how high I'm setting the saturation. This approach works for even higher saturation values.

Comparison1

Comparison2

Comparison3

Comparison4

Comparison5

Comparison6

A picture is worth a thousand words. Post 102 is worth more.

I’m late to the party, but I think this is the best song. Enjoy!

—Teams Jetrell Fo 1, Jetrell Fo 2, and Jetrell Fo 3

Author
Time

Those are some nice skintones man.
Best part is that it doesn't seem to have fucked up the reds in the picture or anything.

Author
Time

Can someone please try my method with $CDAA96 instead of $D29628? A screenshot or two would be greatly appreciated. (I don't have access to my computer right now.)

A picture is worth a thousand words. Post 102 is worth more.

I’m late to the party, but I think this is the best song. Enjoy!

—Teams Jetrell Fo 1, Jetrell Fo 2, and Jetrell Fo 3

Author
Time
 (Edited)

Great job.  I'd say it's definitely an improvement. 

Here's your latest method with a couple extra steps for the other 2 channels. The saturation is over the top just to give a better idea of what's happening.

EDIT: Just to be clear the first image is unsaturated.  The bottom 2 contain the same saturation tweak.

source=avisource()
rgb= source.Tweak(cont=1.1, sat=1.1).ConvertToRGB24(matrix="Rec601")
fixr=source.Overlay(rgb.BlankClip(color=$FFFF00 ), mask=rgb.ShowRed(), opacity=0.1, mode="lighten")
fixg=fixr.overlay(rgb.blankclip(color=$00ffff),mask=rgb.showgreen(),opacity=0.15,mode="chroma")
fixb=fixg.overlay(rgb.blankclip(color=$ff00ff),mask=rgb.showblue(),opacity=0.15,mode="chroma")
test=source.tweak(sat=2)
comp=stackhorizontal(source.subtitle("original",x=-1),fixb.tweak(sat=2).subtitle("adjusted",x=-1),test.subtitle("original (saturated)",x=-1))
comp

Luke threw twice…maybe.

Author
Time

althor1138 said:

Great job.  I'd say it's definitely an improvement. 

Here's your latest method with a couple extra steps for the other 2 channels. The saturation is over the top just to give a better idea of what's happening.

EDIT: Just to be clear the first image is unsaturated.  The bottom 2 contain the same saturation tweak.

source=avisource()
rgb= source.Tweak(cont=1.1, sat=1.1).ConvertToRGB24(matrix="Rec601")
fixr=source.Overlay(rgb.BlankClip(color=$FFFF00 ), mask=rgb.ShowRed(), opacity=0.1, mode="lighten")
fixg=fixr.overlay(rgb.blankclip(color=$00ffff),mask=rgb.showgreen(),opacity=0.15,mode="chroma")
fixb=fixg.overlay(rgb.blankclip(color=$ff00ff),mask=rgb.showblue(),opacity=0.15,mode="chroma")
test=source.tweak(sat=2)
comp=stackhorizontal(source.subtitle("original",x=-1),fixb.tweak(sat=2).subtitle("adjusted",x=-1),test.subtitle("original (saturated)",x=-1))
comp

looks pretty good to me..

 

later

-1

[no GOUT in CED?-> GOUT CED]

Author
Time

-1: Agreed. It does look pretty good.

With that being said, I think this approach deserves more exploration. Would anyone like to meet on IRC to brainstorm?

A picture is worth a thousand words. Post 102 is worth more.

I’m late to the party, but I think this is the best song. Enjoy!

—Teams Jetrell Fo 1, Jetrell Fo 2, and Jetrell Fo 3

Author
Time

I'd be willing to meet up and chat about it.  I haven't used IRC in like 10 years though so I'd have to brush up on it.  Where would you want to meet?

Luke threw twice…maybe.

Author
Time

I use IRC on a daily basis, wouldn't mind.

Can we use efnet? That happens to be conivenent for me.

Author
Time
 (Edited)

There should be a place for OT'ers to hang out and chat realtime.  Is there one already?

 

EDIT: can somebody point me to a good(free) irc program?  Apparently MIRC costs money now. 

Luke threw twice…maybe.

Author
Time

althor1138 said:

There should be a place for OT'ers to hang out and chat realtime.  Is there one already?

 

EDIT: can somebody point me to a good(free) irc program?  Apparently MIRC costs money now. 

There are plenty, but I'd suggest trying mibbit.com's web-client first.

A picture is worth a thousand words. Post 102 is worth more.

I’m late to the party, but I think this is the best song. Enjoy!

—Teams Jetrell Fo 1, Jetrell Fo 2, and Jetrell Fo 3

Author
Time

Mirc is stil free, but it just "nags" you whilst starting up.
Click continue, wait click continue again.

Author
Time
 (Edited)

Well I think that adding yellow to the red is a bad idea.  After a little bit of research it seems that it should be something like this:

RED-Add cyan

Green-Add magenta

Blue-Add yellow

source=avisource()
 rgb= source.ConvertToRGB()
 red=rgb.showred().overlay(source.blankclip(color=$00FFFF),mask=rgb.showred(),mode="chroma",opacity=.5)
 green=rgb.showgreen().overlay(source.blankclip(color=$FF00FF),mask=rgb.showgreen(),mode="chroma",opacity=.9)
 blue=rgb.showblue().overlay(source.blankclip(color=$FFFF00),mask=rgb.showblue(),mode="chroma",opacity=.1)
newrgb=mergergb(red,green,blue)
mix=source.overlay(newrgb,mode="chroma",opacity=1.0)

return(mix)

This seems to work a lot better imo. The opacity for each channel has to be tweaked scene by scene still.  Is anybody still pursuing this?

EDIT: Forgot to add tweak to the last variable so you might wanna do that and tweak it up really high. Somewhere between 3 and 5.

Luke threw twice…maybe.

Author
Time

althor1138 said:


Well I think that adding yellow to the red is a bad idea.  After a little bit of research it seems that it should be something like this:

RED-Add cyan

Green-Add magenta

Blue-Add yellow


source=avisource()
 rgb= source.ConvertToRGB()
 red=rgb.showred().overlay(source.blankclip(color=$00FFFF),mask=rgb.showred(),mode="chroma",opacity=.5)
 green=rgb.showgreen().overlay(source.blankclip(color=$FF00FF),mask=rgb.showgreen(),mode="chroma",opacity=.9)
 blue=rgb.showblue().overlay(source.blankclip(color=$FFFF00),mask=rgb.showblue(),mode="chroma",opacity=.1)
newrgb=mergergb(red,green,blue)
mix=source.overlay(newrgb,mode="chroma",opacity=1.0)

return(mix)


This seems to work a lot better imo. The opacity for each channel has to be tweaked scene by scene still.  Is anybody still pursuing this?

Wow! That looks very similar to something I was toying with the other night.

I'll take a look at this tonight when I get home.

A picture is worth a thousand words. Post 102 is worth more.

I’m late to the party, but I think this is the best song. Enjoy!

—Teams Jetrell Fo 1, Jetrell Fo 2, and Jetrell Fo 3

Author
Time

AntcuFaalb said:

Here's a GOUT-related question: Assuming that we don't make any color changes beforehand, how much do the reds need to be lowered to fit in with the rest of the colors? That is, what change would result in them no longer being over-represented? Is it just the reds?

frank678 said:

This is pretty much THE question for me. I saw some posts where someone indicated they could repair the red channel out of the RGB channels. My understanding is that the red shift is an organic shift in the film - its not the red channel is damaged but the greens and blues have faded leaving the red over represented. Anyone know how correct that is? If so can the green and blue be repaired?

edit: had a go at a blind ajustment to blue and green curves to find a result = http://originaltrilogy.com/forum/topic.cfm/Calling-all-Color-Correctors-Can-this-source-yield-a-different-set-of-results-to-Gout/post/598240/#TopicPost598240

 

This theme, from the "Three-strip Technicolor: Please help!" and "Calling all Color Correctors: Can this source yield a different set of results to Gout?" threads, is one that echoes throughout the OriginalTrilogy forum -- that of restoring faded source material.

Restoring "by eye" (usually using hit-or-miss, narrow-fix filters that don't address the full problem) or referring to other "restorations" (accomplished with their own "by eye" fixes) is well intentioned but flawed. The science of film is what's needed and we already have the tools to examine and implement a scientifically guided approach.

Being in the middle of such research, I have done enough to formulate a workable approach, demonstrated in the "THX 1138 preservations (Italian Cut available, see 1st post)" thread (starting at page 48):
http://originaltrilogy.com/forum/topic.cfm/THX-1138-preservations-Italian-Cut-available-see-1st-post/post/585765/#TopicPost585765
http://originaltrilogy.com/forum/topic.cfm/THX-1138-preservations-Italian-Cut-available-see-1st-post/post/585922/#TopicPost585922
http://originaltrilogy.com/forum/topic.cfm/THX-1138-preservations-Italian-Cut-available-see-1st-post/post/589761/#TopicPost589761

Therein is a more extreme example of "bad colors" (faded color) from a 16mm film source:

And the way to look at it is by it's color layers -- this one is the faded red channel:

The way to correct it is knowing that fade is ultimately contrast contraction, and the fix is to expand that condensed contrast back to it's original size:

The other 2 channels (green and blue) usually need such correction, too, and the result is the way it was before it faded:

Of course, the above explanation is simplistic and the "THX" thread goes into more detail about the non-uniformity of fade -- across the condensed spectrum and through the length of the source. But this is a starting point that should address all kinds of issues (skin too red, other reds weakened, etc.) in a uniform way. Hope this helps.