Divide Blend Mode for VEGAS Pro

alifftudm95 wrote on 3/28/2020, 10:47 AM

I learn something amazing today about the math behind "Divide Blending" mode from and how it can be used to remove color cast on the video. (Maybe this can fix VEGAS Pro sluggish inaccurate WB tools)

This should be universal across entire editing software, as devide blending only devided the RGB value of the color

underneath the "devide" layer.

The way this method work is to create a solid color and pick a color from the images/videos that have a cast on it. (Color that supposed to be white but arent), then simply blend the solid layer with devide blending mode.

256 " 8 bit color " ( A ÷ B )

= R G B Value

This straight away remove color cast on the images/videos & make what supposed to be white, to be an actual white (WB corrected I shall say?). Sadly, VEGAS Pro don't have any divide blending mode and this should be add on future VEGAS Pro 18.

Luckily, I own a VEGAS POST product and I remembered VEGAS Effects have bunches of blending mode option to choose from. It does works, but sadly I dont have the ability to tweak Brightness to regain back the highlight detail as in videos. I tried to use Fill color FX in VE but the FX is very simple and it doesnt offer any small tweak control for hue, saturation & brightness.

This small control (Hue/brightness/Saturation) can be founded on After Effects (Both Solid layer & Fill FX) & in Premiere Pro too.

Back to post

Comments

Marco. wrote on 6/25/2020, 6:32 PM

@alifftudm95
I think I can get (very close to the) same result as if using the divide blend mode when using an inverted solid color sample composited with the overlay blend mode and fine tuned via luminance (and saturation) of the HSL FX applied before the Invert FX.
I know – the math behind is different. But I tried with various different sources and all finally pretty much matched the results of using the divide blend mode.

Left is source, right is composited.

adis-a3097 wrote on 6/25/2020, 7:04 PM

Now the shirt, that looked black before, looks reddish. Who knows how it's supposed to look, I don't.

Let's not forget: white is only white under white light! Which is the reason why I posted the frame shot under red light in the first place.

Yes, Mr. Dinda is a smart guy, but he also is a bit of a show off. As am I, you know, it takes one to know one... 😆

Marco. wrote on 6/25/2020, 7:46 PM

I personally don't find the divide blend mode being the better approach for correcting a color cast, also I'm not a friend of auto white balance modes. Finally it all depends on how the colors got biased, what's left to be used as reference and what then is defined to be the overall goal. Then choose the appropriate tools and get the craft done.

But I don't evaluate these methods. I just was curious if in Vegas Pro we can use something similar like that divide blend mode. It seems we can, what I think is the topic here. So we have another option.

alifftudm95 wrote on 6/26/2020, 12:58 AM

Now the shirt, that looked black before, looks reddish. Who knows how it's supposed to look, I don't.

Let's not forget: white is only white under white light! Which is the reason why I posted the frame shot under red light in the first place.

Yes, Mr. Dinda is a smart guy, but he also is a bit of a show off. As am I, you know, it takes one to know one... 😆

hahahahahahahahha, very funny guy. I was right tho in the first place haha

alifftudm95 wrote on 6/26/2020, 1:00 AM

@alifftudm95
I think I can get (very close to the) same result as if using the divide blend mode when using an inverted solid color sample composited with the overlay blend mode and fine tuned via luminance (and saturation) of the HSL FX applied before the Invert FX.
I know – the math behind is different. But I tried with various different sources and all finally pretty much matched the results of using the divide blend mode.

Left is source, right is composited.

Interesting, How many FX and steps do you need to recreate this?

Yelandkeil wrote on 6/26/2020, 2:14 AM

More interesting is, how much brain that Mr. Einstein did take to divide?

walter-i. wrote on 6/26/2020, 2:19 AM

hahahahahahahahha, very funny guy. I was right tho in the first place haha

There are people who are always right, and that is what makes them so charming.

Yelandkeil wrote on 6/26/2020, 2:22 AM

And can always find reason to beat you down.

walter-i. wrote on 6/26/2020, 2:30 AM

There is a saying, (unfortunately I don't know who) that says: Who makes others big is big himself, who makes others small is small himself.

Marco. wrote on 6/26/2020, 5:15 AM

"How many FX and steps do you need to recreate this?"

1. Put your source clip into a video track.
2. Put a solid color event into the track above.
3. Apply HSL FX and Invert FX onto the event (or track).
4. Pick the neutral color.
5. Set the track blend mode of the upper track to overlay.
5. Adjust brightness in HSL of the solid color event.

I think step 2., 3. and 5. could be automated via scripting thus it could be optimzed to 3 steps.

Actually you don't even need the HSL FX but I found it's easier to tune the luma via a separate FX instead of via the solid color event. Both works.

Marco. wrote on 6/26/2020, 10:54 AM

Already optimized it a bit. If I start from an empty project, I …

- Run first script to automatically create and label two video tracks.
- Run a second script which sets the selected track to "Overlay" blend mode, insert a grey Solid Color Event and select the Solid Color Event.
- Run a third script which applies HSL FX and Invert FX onto the selected Solid Color Event.

Having these three scripts added into the tool bar it's only three mouse-clicks to have it done and once you got your tracks you only need two clicks to be ready to insert your source media, pick the neutral color and fine tune the luminance.

Should be at least as fast now as it is in other Divide-Blend-Mode-Apps. Though it seems like this workflow is only suitable for more or less slighter casts.

john-brown wrote on 6/26/2020, 11:34 AM

What does White Balance do? Does it not do a divide blend?

Yelandkeil wrote on 6/26/2020, 12:57 PM

White Balance does only one job: If your Machine got the wrong temperature of the light, then it corrects the temperature in the post-edit, nothing else.

The Principle of RGB for display is, a trichotomy by adding/dividing values of the other channel leads you getting gray scale toward white/black.

And I dare say, it has nothing to do with "small color cast" in this topic.

Robert Johnston wrote on 6/27/2020, 4:19 AM

I believe that the Dodge blend mode in Vegas gets very similar results to the Divide blend mode in APP if you invert the solid color event in Vegas. So if you sample a yellowish area, for example, put the Vegas Invert FX on the yellowish solid color event, which changes it to blueish color, and set the blend mode on that track to Dodge. Your video would be on the track underneath.

That method I got from one of the Photoshop Restoration and Retouching books by Katrin Eismann.

Marco. wrote on 6/27/2020, 5:07 AM

@Robert Johnston
You are right. Dodge seem to work even better than Overlay if you tune the brightness differently.

Yelandkeil wrote on 6/27/2020, 5:31 AM

Thanks Robert, that's just one way in Vegaspro and ways are still there.

The problem of us the modern human is, we picked some fragmentary knowledge and believe we found the truth. Ok, till then it's just your problem.

But if we try hijack with this "truth", that's the bad thing. A suggestion of Divide Blend for Vegaspro is good, even without any studying this NLE. But the example and description confused me so much.

From the video show we see clearly that the boy has false color on his face, but the Mr. Einstein goes correcting the ear ring: an operation White-Balance?

The topic OP then introduces the term "small color cast". I gave my "big color cast" as water surface and try to say, change one certain value means changing all pixels in the distribution in Histogram that have this value, it doesn't matter whether you add/subtract/divide this value.

My understanding of Blend-Mode: make a gradient map (not choose one single value) and compose it to the origins - that's why we generally need two tracks.

My understanding: "small color cast" is floating pixel values that belong nowhere or illegal, for example --