DrDre

JawsTDS said:

That Tech IB match looks amazing. I would love to see the Blu-ray matched to that, even with the changes. Would you ever consider releasing the Blu-ray rips matched to these sources for the basis for other projects?

That's certainly a nice idea, but it would be a 6 month endeavour to color correct the bluray, and some scenes are impossible to correct without separating the elements. I'm afraid with the projects I've got lined up, it won't happen any time soon. I will be sharing the user interface I've created for color correcting, so maybe someone else would like to do it?

Another set of comparisons between the various home video releases, and the Tech IB:

Bluray/2004 DVD:

1997 SE:

GOUT/1993 LD:

JSC:

Tech IB:

Here's a frame that looks massively different on the Tech IB, than on the bluray.

Before:

After:

camroncamera said:

DrDre said:

Thanks! Building a color matching model takes quite a long time. About 10 min on my Intel Core i5 laptop. So, if you want to match each frame to a reference, it would take forever. However, if you use 1 reference, and correct between 20-50 frames with this model, it takes 10 min to make the model, and then 5-20 sec for correcting each frame, depending on the resolution. It's about 10 sec for a 1080p frame. On the 3.5GHz 4 core it will probably be a lot faster, especially if you can use parallel computing. 

 It sounds to me like this is a full-blown render operation. What about, also/instead, a plug-in for something like DaVinci Resolve? The plug-in could analyze a reference (like your algorithm does) but generates a "correction" or LUT that can be applied in a new node. Basically a filter or "look" that can be cut, copied, pasted, blended to varying strengths, etc. to any clip in the timeline, with live preview. Rendering would take place using the Resolve engine once the project is finalized, as any normal Resolve render job.

There would be quite an advantage to have this tool built-in to an established colorists' workflow. I think you'd have commercial hit on your hands. Perhaps you already have a goal like this in mind.

No goals yet, just did it for the fun of it. 

I want to compile the GUI, such that it can be used as a standalone application.

Harmy said:

I know very little about this stuff but would it be possible to connect it to some cut detecting algorithm, so that it references one frame in each shot (or by setting the cut detection to much lower sensitivity possibly just each scene) and then apply that to the corresponding shot/scene in the file you want to correct?

I've made a user interface, that makes it very easy to select, crop frames, and subsequently build a model. You can then select any number of frames, that will be color corrected and saved in a separate directory.

Zyrother said:

Wow, that is really impressive. Looks very natural.

I have not watched the Blu Ray set, but I assume the consensus was that the colors were over saturated? Or what was the problem regarding the set color wise?

The bluray colors are truly awful. You can see some screenshots in my color matching thread. 

Thanks! Building a color matching model takes quite a long time. About 10 min on my Intel Core i5 laptop. So, if you want to match each frame to a reference, it would take forever. However, if you use 1 reference, and correct between 20-50 frames with this model, it takes 10 min to make the model, and then 5-20 sec for correcting each frame, depending on the resolution. It's about 10 sec for a 1080p frame. On the 3.5GHz 4 core it will probably be a lot faster, especially if you can use parallel computing. 

Here's SRV15, a huge improvement:

Just kidding, this is the bluray adjusted to match the GOUT colors. 

Here's a funny experiment. I took the frame of the Tantive IV soldier, and regraded the bluray to each of the home video releases over the years: 1997 SE, GOUT/1993 LD, and JSC. Also shown is the Technicolor IB print. The differences are huge.

Bluray/2004 DVD:

1997 SE:

GOUT/1993 LD:

JSC:

Tech IB:

...and another one of Obi-Wan aboard the Deathstar. 

Before:

After:

Here's a frame with Han Solo and Chewie in the Cantina.

Before:

After:

towne32 said:

Hmm, definitely some weird magenta artifacts going on around the horizon. 

Where'd the other shot go? :)

 The magenta artifacts were indeed another cropping issue, here's the correct result:

towne32 said:

Hmm, definitely some weird magenta artifacts going on around the horizon. 

Where'd the other shot go? :)

The magenta artifacts are also in the bluray, to a certain extend. The bluray is really full of weird color artifacts. Blue smears, yellow smears, etc. I will check the cropping, because that can sometimes cause problems, as was evident from before. The greenish discoloration at the edges are also visible in the print, interestingly enough. 

towne32 said:

Looks great.

How do the shots of 3PO and R2 first arriving on Tatooine look?

 Like this :-).

Before:

After:

What about this great shot of the greatest villain of all time.

Before:

After:

The binary sunset and the green lightsaber have been digitally manipulated, and are therefore very difficult to correct with an algorithm. Here's another frame that I love. Peter Cushing was such a great villian. It's a subtle difference, but I think the original is so much better.

Before:

After:

towne32 said:

You mean the second set is using the specific IB frame to match, while the first did several matches based on a single frame comparison, right?

The second set looks more natural. The darker regions don't have the weird green blobs going on. The sky is quite white, and it looks a bit strange against the very light desert landscape. But would you say that's how the reference prints looked? Do you think the luminescence is correct? Is the bluer sky in the frame 1 of the first set due to the reference frame used (i.e., leftover blue from the BR)?

Here is a set of comparisons from the two sets:

http://screenshotcomparison.com/comparison/142387/picture:0

The Tatooine scenes are indeed bright on the reference print. The difference in the first set is probably due to a slightly different color gradings on the bluray. 

NeverarGreat said:

DrDre said:

 Those last frames look spectacular!

I think this method will be the way to go when doing color corrections of the Blu-ray, as it automates what would be (and was) hours and hours of work trying to rescue these shots using manual color correction. There will still be problems with the Blu-ray that need to be corrected by hand of course, such as R2's dome in the above shot, which is oddly green with purple panels. That's just the nature of the blu-ray, that some shots are missing color channels and gradients and nothing can be done to fix that without pulling in color information from another source.

Actually, the purple panels were due to inaccurate cropping, before color correcting. R2's dome is actually slightly green tinted from the dirt on his dome. Here is the correct one:

Here's the same sequence of frames, now color matched directly to a reference:

Notice, how there is a blue cast/blue patches on everything in the bluray: the sandcrawler, the walls of the homestead, Luke's shirt, etc. It's awful!

Hahaha! 

The results are extremely close to the actual Technicolor IB print references. 

This example of a sequence on Tatooine shows the power of the color transfer method I developed. First I calibrated the color transfer model on a frame from to bluray matched to the Technicolor IB print reference:

Next I used the color transfer model to correct the following sequence of frames from the bluray:

The results are...

I extended the method, where you match histograms in multiple color spaces. I did this in three ways:

1) Replace the histogram with an estimated probability distribution, using kernel fitting, which is easier to invert.

2) Invert the estimated cumulative probability distribution to obtain a direct mapping between the colors in the source frame and the reference in multiple color spaces. This can then be used to predict other frames.

3) Stabilize the solution.