Totally sumptuous!

I've got me thinking cap on. This has definitely throw the cat amongst the Vegas Poulets!!

Yup! Count me in Jonathan!

Okay, I see now. That was from their web site, duhh...

Yes, my question would be how would one make the color separations in Vegas, then put it all back together?

Jay, it was this that gave me the clue: "Rather, the challenge is to do this process in Vegas"

But there again is it to create the actual 6 frame Track Motion? Can do that.

It's the re-assemnbly thing that is intriguing me. Oh yes!!! Look at that Blue sky and that lushes skin tone? Straight outta the '50s - huh Jay?

Oh I can feel myself wanning to break out into yestful, joyous projectile song,

"There is nothing like a dame!"

"Oh what a beautiful Mornin'!"

"OOOoooooohklahoma . . where the .. . "

"The Deadwood Stage .. "


.. oh yes!!!

LOL -- Yes, the old Technicolor movies were something to shout about.

Also, the look of the 2-color process reminds of me of the early Kodachrome days--beautiful!

My immediate thoughts are somehow manipulating Colour Curve masks? Then somehow bringing them back together. Dunno. . . .

Excellent puzzle. How DO we lay one colour over another in Vegas? I can mix layer colours within Paint Shop Pro. I can manipulate layers and all that. In Vegas? Layers? The colours are, well opaque? Aren't they?

Where is Glenn Chan when you need him?

I think you'd use the Channel Blend option to get your RGB layers. The question I have is using them to create the final 3-color image.

Yeah. I think the "mattes" block-out areas, becoming the "stencil" to allow the "other" mixes to come together. So, what we have in actual fact is a series of see-through stencils that combine allowing one to see the whole again .. oh yes . . SO, unlike colour mixing, we would be creating colour=stencils of each layer.. It is the these layered back up stencils that combine.

I'm still not convinced by my own theory .. to many "holes" . .well, actually, not ENOUGH holes . . . hmmm...

In terms of adding tracks, you might be able to use the same method that farss recommended ages ago in the thread on my "ultimate" VHS restoration. I was capturing the same tape multiple times and then adding the captures together. The noise averaged out, providing noise reduction. To make it work, I needed to add two (and eventually three or more) copies of the identical footage, but have the luma values all come out correctly.

Here's the link to the recommendation:

Adding Multiple Instances of the Same Source

By using the Channel Blend "Distribute" to create the 3 B&W layers then adding the color via "Color" (RGB) Only I get three layers. Then I blend them using "Add" in the Compositing Mode", but all I get is a dingy version of the original.

I'm missing a step somewhere.

Okay, I'm stuck on how to create the individual R-G-B matte filters!

I seem to recall a preset for the CCs that simulated 3 strip Technicolor.

However to do exactly what is done in this demo, piece of cake in Vegas.

Use CCs to create a R only, G only and B only output, add the B&W FX. You'd need to render this to 3 files. Each one will contain values such that the luma = the R or G or B values of the original. You now have a tri separation.

Render out these applying the MinMix FX and Mask Generator as uncompressed with alpha to create the masks.

You then composite the masks, B&W triseparations with say generated R,G,B. You then composite the new R,G,B back together using ADD as the compositing mode.

Job done.

Tweaking values in the MinMax FX will change the look in interesting ways in the final composite.

Bob.

And that's a piece of cake? Who does the baking in your household? I give up . . it's late here on London . . I'll get back on it in the morning.

Ah, so there is a "simulation" CC - my guess that this rather a saturation FX?

But then there is "Bob's Home Bake Recipe No2341" .. . I'll need to do some meditation before I take THAT one on.

Piece of cake?!? My eye!

A huge thank you to everybody for the thick replies, wow, we've really got something going here.

Let me point you to some images that would help us all with this experimentation. These are the same images that were used in the videos I already posted to demonstrate how the Effects team were able to create the look.

Images of the 2 Strip Technicolor process: before & after
Original Photography
2 Strip Technicolor

Images of the 3 Strip Technicolor process: before & after
Original Photography
3 Strip Technicolor

Perhaps we can experiment taking each original photo and getting it to look like the final Technicolor outcome that they did. For this project I recommend using a resolution of 706x395 progressive scanned with a pixel aspect ratio of 1.000 (Square).

Disclaimer: The images of the 3 Strip Technicolor process are not cropped the same, but that is the fault of the source, not Vegas. To correct the problem when using the project settings I already suggested, crop both images to the preset "1.85:1 Academy Aspect Ratio", then take the 'original photography' image and change the Y Center from 197.5 to 200.3. Now they will align perfectly (I think).

Here's a VEG file of the project where I have separated the channels as farss described. It looks exactly the same as the Scorcee tutorial begins, however, when I applied the Sony Black and White filter, the image was far darker than I believe it should be; this may be an issue with luminosity.

Vegas Project

and you will need these images...

4.jpg
6.jpg

After that, I don't understand what farss means by applying the MinMix FX and Mask Generator, but I do know how to render as uncompressed with alpha. Can anyone help? We're 1 of 4 steps there.

Step 1: Separate the colors of the original image.
Step 2: Separate the mattes of the colors.
Step 3. Combine matte layers into color layers to create new color layers.
Step 4. Combine new color layers to create the new image.

OK,
actually I don't think you need to create mattes (or masks) IF you're going to use Add as the compositing mode later on.

The point was to use the MinMax FX followed by the mask generator to create a mask with alpha channel. As I said above I don't think you need one.

Basically the process is to create 3 B&W videos of the source such that the luminance values match the corresponding R,G,B channels in the source. This gets you to what would have come out of the three strip Technicolor process.

Each one of those was then basically printed onto full color stock. The film was exposed by the B&W 'Red' roll through a red filter etc.

You can convert the 3 B&W strips to R,G and B either by dropping out the other two colors using CC or else compositing using multiply with pure R, G and B. What you now are back to is the same as you had in the first step.

Then you composite the three back together again with composite mode set to ADD.

If eveything is working 100% you should get back exactly what you had in the first place. That's not very exciting. However in the process you've created three separate channels in B&W. Applying FXs to those before compositing back together again, either alone or in combination can create some interesting FXs.


Of course you may just end up duplicating some of the already existing FXs in Vegas.

Bob.

You really aren't getting anything even remotely resembling technicolor unless you actually photograph using three cameras (through colored filters), or something similar. Perhaps you could modify a 3-chip camera to get some of the raw data from the chips.

What I think you're doing here is simply creating a very interesting fX, but all it really does is let you manipulate each color channel, something you can do in many other ways without "separating" each image into three B&W pictures and then recombining.

Ok my brain is exploding with math here so if I make no sense that's why.

I believe this is how the 3-strip Technicolor process works:
Image is split up into R, G, and B and captured onto seperate pieces of film.

This is then matrixed into the complementary colors, so subtractive dyes can be put onto the final film.

*Ok, somewhere in there 50% green was put onto some of the film. I can't remember which step this was.

Aviator:
Aviator was shot on modern film stock, where the film layers are stacked on top of each other.

Key concept: In film, the light passing through the film is determined by multiplying the respective transmission %. So what happens in the first layer affects what happens in the subsequent layer.

Output light = Input light * transmission % of first layer * transmission % of second layer * transmission % of third layer [...] etc.

---

The goal is to take this film and get at the original R, G, and B- as if you split the light and recorded that onto seperate pieces of film.

But because of the effect of the key concept mentioned above, there is "cross talk" between the colors. This is what the matte layer is for- to undo this effect.

Now we have a guess of what 3-strip technicolor R, G, and B layers would have recorded. From there, we composite those layers together in complementary colors, then a final print with complementary dyes on it. Again, we have to pay attention to the cumulative transmission.

2- If you want *video* to look like Technicolor, then you also need to pay attention to film's non-linear s-shaped transfer characteristic.


Ok. Whew. I don't want to touch this with a 10-foot pole ;)
Well actually, if you had the correct maths, this shouldn't be hard to emulate in a plug-in. MBE does this already I believe... although I haven't really looked deeply into whether its algorithm is correct.

They don't say how they combine the matte frames to make the new color mask. If you notice in the QT, when they combine the G and B mattes, they have to be using some strange type of compositing. It appears that the B channel was inverted and then they added the G and B mattes together (comparably a 'darker' composite).

I wish they had explained more about how the mattes are combined into the new channel masks.

Well, if it can't be done then it can't be done. If it needs to be done with a plugin that might already exist, then so be it too. But why not find out?

I tried the technique of B&W'ing the colorized layers, then importing them back together , but the image turns out a little funky. When I add the red and green b&w mask layers (as described in the video) onto the blue color layer, I suddenly get these wild red and green colors, yet there is NOTHING with the red or green color available to produce them. I b&w the whole thing and then color curve it to only show blue, and they reappear! It's wild, man.

im trying to make the "color matte filters" , but, witch kind of matte is that? does not look like a color separated image, looks more like a "secondary color corrector" picked selection.

this part is missing in my brain... :(

You're missing the whole point of the exercise, John.

Well with enough layers, Vegas might be able to do it. I'll have to wrap my brain around this one.