originaltrilogy.com
Calling all Color Correctors: Can this source yield a different set of results to Gout?
Calling all Color Correctors: Can this source yield a different set of results to Gout? :: 1 < 4 > 5
russs15
This user is offline.
Where am I????There is a 1985 UK PAL LaserDisc but I have never, ever seen it or any transfer from it.
4 - 5 - 3 - 1 - 6 - 2
Discuss..........
frank678
This user is offline.
This is trying out the idea of taking a snapshot of a still in vlc, then changing the settings and taking a second snapshot, then merging the two together. The idea behind it is to increase the range/depth of a single image by adding additional information to it. Using the 82 LD which does not have a lot of image depth it does seem to work to a degree, the only problem was it added too much darkness where I just was aiming for just adding depth. I wonder has anyone here ever tried merging moving images (i.e. clips/or the whole films) with slightly different settings (i am unable to try this out).
original 82 ld:
settings changed:
the above two merged:
The following stills are the end results only (all using the same settings - i.e. this is meant to be a one step/automated process which either improves on the original in one 'fix' or not.
frank678
This user is offline.
Duplicate layering technique, with the 1984 tv broadcast. 
Dunedain
This user is offline.
Padawan LearnerVery nice work, Frank. :) I hadn't thought of that, color correct a frame, then overlay it on an uncorrected frame, thus improving the colors, but retaining some of the desirable qualities of the unaltered frame.
The shot of Han at the top is an excellent example. By combining the two, you have greatly reduced the overly bright look of the first and also improved the colors, but at the same time you have not given Han that weird darkish orange/red look that is so hard to avoid in everyone's skin tones when trying to boost the saturation in the home video sources of the Star Wars trilogy, and which does not look good when it happens. Even though it might have made some areas a bit darker, that's a price well worth paying so the people look natural, which is far more noticeable.
You can see that oh so clearly on the first page of this thread. Luke's face looks like he was covered in red mud or something on the Blu-ray, looks terrible and unnatural and definitely very incorrect.
I wonder what your method would yield when applied as a color correction method to the best quality sources available, like darkjedi's 720p upscaled video, or negative1's 35mm film frames. Would be *great* to see what that would look like, it might give some pleasing qualities to the final result. Maybe some test samples can be shown by those involved. :)
Although, being that the 35mm frames need so much correction in the first place, I'm not sure if that would help in that case. But darkjedi's 720p video is the best video (non-film) source I know of, and this method should be applicable to it. And who knows, the end results might be nice.
frank678
This user is offline.
Thanks for the feedback Dunedain, unfortunately I havnt been able to systematize what I'm doing. What I do is push one picture darker and one lighter and then find the right percentage to overlay them (so the darkness and lighness from one picture 'adds on' to the darkness and lightness of the other - the trick seems to be to get it so the two ends align with the mid range so you get a seemless transition in the picture from top to bottom). It works either by having two distinct layers or adding a duplicate layer in another measure.
I have been working on one universal setting for each source, but you could tailor each overlay to each scene to get the most information of whats there at any given time (variables over constants, but establishing a constant first might be a good idea to give a starting base line). I have only tried it with two layers as I am still operating theory-blind - someone who knew how to adjust highlights, midtones and shadows separately might be able to stack 3 or more subtler layers. The thing however with soft undetailed sources it starts to look too "painted" if I start adding too many layers.
As far as I can tell even if the Gout/Blu Ray is damaged it would be possible to paint whats missing back in (if the information is missing it can't be recovered). I tried this with the Luke in the Judland Wastes picture a number of posts back , this was done through thinning out the original picture to thin midtones, duplicating it, then adjusting the new duplicate layer with a different colour curve (so you are not adding back in a linear way but reshaping the boundaries of the original picture - I think [?]). The Luke picture had several layers, I got the 3D effect I was after but it looks too painted.
Jetrell Fo
This user is offline.
Padawan LearnerI am going to hazard a guess by saying what you're doing is similar to cap averaging except your going frame by frame and using opposite techniques to find a middle ground?
The combined shot of Han above looks nice but it also loses the detail from the second shot while retaining the dark of the first shot.
Maybe with some tweaking and playing you'll come up with something that can retain the best qualities of both and not be blown out.
Cheers for your effort.
"They assimilate entire worlds and we fall back...Not this time, The line must be drawn here, this far...NO further"
frank678
This user is offline.
Jetrell Fo said:
I am going to hazard a guess by saying what you're doing is similar to cap averaging except your going frame by frame and using opposite techniques to find a middle ground?
The combined shot of Han above looks nice but it also loses the detail from the second shot while retaining the dark of the first shot.
Maybe with some tweaking and playing you'll come up with something that can retain the best qualities of both and not be blown out.
Cheers for your effort.
Agreed this is pretty much the same as averaging +this technique does seem to submerge details. The plus side is it can give added depth to what hasnt been submerged. The trade off is between colour and detail so far, maybe that might be resolved by subtler moves. maybe.
frank678
This user is offline.
Here's a cross pollination with the 2 questions just posed in Anctufaalb's thread:
http://originaltrilogy.com/forum/topic.cfm/Three-strip-Technicolor-Please-help/post/598218/#TopicPost598218
Here's then an attempt at changing the blue and green curves to dovetail with the red (with absolutely no science behind it).
frank678
This user is offline.
*deleted*
frank678
This user is offline.
The above GOUT adjustment layered = 
frank678
This user is offline.
Several layers on, all method went out the window and its wrong/too painted but the potential is there to get a more 3d effect from GOUT:
frank678
This user is offline.
Looking back at these I might not have got the original adjustment right - so I took it forward to early. Even so there seems to be a limit on the relationship between the amount of bits/softness in the image and amount of depth that SHOULD appear with those bits/that softness. Otherwise the layers look too phoney. maybe the answer is to strip it back and only build it back to the original density??
snicker
This user is offline.
I like the look of that last frame.
frank678
This user is offline.
*deleted*
frank678
This user is offline.
Anyone know how to do red, green, blue analysis? Is there such a thing even?
snicker
This user is offline.
If you mean an RGB or Hue map/cycle, which displays a "map" of R/G/B/y/c/m clusters representing the location of R/G/B/y/c/m within an image then, yes, it exists. The Colorama tool in After Effects allows for mapping hue or RGB. Wouldn't have a clue how to do it with any other software.
frank678
This user is offline.
Layering the 1982 PAL VHS:
frank678
This user is offline.
La Guerra de las Galaxias
frank678
This user is offline.
/
frank678
This user is offline.
/
frank678
This user is offline.
/
frank678
This user is offline.
/
Spaced Ranger
This user is offline.
"Your love of the halflings' leaf has clearly slowed your mind"frank678 said:
This is trying out the idea of taking a snapshot of a still in vlc, then changing the settings and taking a second snapshot, then merging the two together. The idea behind it is to increase the range/depth of a single image by adding additional information to it.
Unless you are merging frames from different sources (for a particular reason), you will get nothing more than if you adjust that single frame. The problem is ... adjusting it by eye will not produce anything definitive. You must have a reference to guide you. For example, here are some numbers for luminance / red / green / blue (graphed, of course) for this capture:
The capture looks good, the graphs look good (nothing crushed or blown-out) ... but it's wrong. (BTW, has this snapshot been pre-edited? If not, the gapping in the R/G/B graphs indicates a capture problem.) Compare it to the 2006 DVD (or any other version of your preference) and let's start with the luminance:
'82 LaserDisc 2006 DVD
The LD capture is significantly brighter. From the graphs, it looks like a gamma adjustment is needed:
Sure enough, here, after using a 0.6 gamma adjustment to pull the brightness down (note it's new graph), it looks amazingly good. Further work (color correction -- green looks a little strong) should be much easier now.
Suggestion -- it would be better if you could start your capture with luminance closer to your target image (if possible). That means a lot of sample caps (from all over the movie), graphing, settings-twiddling, and recapping to find the sweet-spot(s) for your capture settings.
frank678
This user is offline.
Spaced Ranger said:
Unless you are merging frames from different sources (for a particular reason), you will get nothing more than if you adjust that single frame.
Hey Spaced Ranger I've found making one frame with the contrast set far apart and one with the contrast in the mid range and layering as transparencies then I get a composite spread of the different information I could'nt get from one set of settings on one frame. Not entirely clear on how it works - I only know that I can't duplicate the results without layering with the program I have. I agree using references and matching the data would be a much less haphazard and more true way of approaching this to my guesswork.
Spaced Ranger
This user is offline.
"Your love of the halflings' leaf has clearly slowed your mind"frank678 said:
... I've found making one frame with the contrast set far apart and one with the contrast in the mid range and layering as transparencies then I get a composite spread of the different information I could'nt get from one set of settings on one frame.
I, too, have tried using such "presets" (for lack of a better term) -- techniques for area-targeted adjustments. snicker mentioned the Colorama tool in After Effects for such adjustments. As I don't have that, my paint program must suffice with it's Layers settings:
They are fun and produce interesting effects. But, ultimately, it's all manipulation of the R/G/B either individually and/or locked together (luminance). I've found that Historgram settings does a man's job of correction or just plain alteration:
One really learns how picture colors work with these controls -- compress / expand; progressive increasing / decreasing [gamma]; min / max / low / high limits -- and, judiciously used across both locked & unlocked R/G/B, can produce almost everything you'd like to do.
Once you know what's going on there, then Curves adjusting does "the impossible":
As you can see, one can do anything to the curve (with enough control points) to do anything to a picture or to parts of a picture. Dangerous territory. :)
For an example, see my post over in the THX 1138 preservations (Italian Cut available, see 1st post) thread -- http://originaltrilogy.com/forum/topic.cfm/THX-1138-preservations-Italian-Cut-available-see-1st-post/post/590208/#TopicPost590208
The curves are not pictured but the control-point numbers are there for an amazing result in color restoration.
Calling all Color Correctors: Can this source yield a different set of results to Gout?