jit.gl.pix - simperimposing textures/matrices - how to

layering gl-textures is done in one of two ways. either with fixed-function blend modes, or with shaders (gl.slab, gl.pix).
for ff blending, you use gl.videoplane (or gl.cornerpin, or...), set @depth_enable 0 @blend_enable 1, control the draw order with @layer, and control the blend function with @blend.

for shaders, you can use gl.pix (or gl.slab, but that requires writing a shader or using one of the built-in shaders - check out all the co.*.jxs shaders for examples). the default gl.pix gen patch does a simple additive blend. changing the + to a * makes it a multiplicative blend. loading gl.pix @gen alphablend does an alpha blend. once you start editing the gen patch, the possibilities are endless.

your link links to a movie file, not patches.

HI rob thanks for taking the time for the insight,

Although I knew everything about what you just wrote (didn't know the term ff blending), it's never a bad idea to review.

When I say I'm a noob, I mean everything having to do with gen
All the objects inside gen, except

swiz , vec, mix, >,?/switch

concat

dot
the dot : product of (dot64 ) tells me nothing.

but more the logic inside demonstrates one needs to have a better
underlying understanding of shader building.
---------
Link:

I don't know how that happened.
My mistake

here is the good one

https://www.dropbox.com/s/6nj63afdevyd27h/vignette%20%26%20layers%20jig.gl.pix.zip?dl=0

You'll understand what I'm trying to do.

thanks a lot Rob

phiol

What about multiplying the inputs?
Have a look at this:

you are using the red channel of the vignette as the alpha channel of your composition, but doing so, you're translating the "black" pixels into transparent (as alpha values go from 0 to 1, where 0 is transparent and 1 full opacity)

To avoid this, you can either disable the blend_enable attribute in the videoplane, or use the alpha channel of your jit.grab matrix in the composition, and multiply the rgb channels with the r, g, or b channel of your vignette.

my only suggestion is to convert the mask to luminance value, and multiply with your input texture.

Both of you produced 2 patches that gave the exact results I wanted.

Thing is I am still quite a noob @ the way this works (Color mapping)
I see GL as 2 categories:
THE PIXELS & COLOR APPROACH VS 3D OB3D object creation and positioning (transforms)
(textures would be dressing up these shapes.)
How you play with the texture's content, is pixels and colors.

Could almost say to sum up,
there are 2 styles/topics in Jitter and Jitter users:
1. geometry transform manipulation : Jit.gen
2. pixel and color manipulation: jit.gl.pix

-------
(slightly off topic)
Funny observation , (more a hypothesis though),
about the C74 Jitter guys and their interest in these categories
(I obviously could be totally wrong :-)

Andrew Benson is more a pixel/shader guy -playing with colors
Rob is more more of the gl object manipulation in 3d space.
Wes I've seen a bit of both.
Not c74 :
Matmat/silicat more like Rob's style.
Same with pedro’s work I seen.
------

Anyways, back on topic:

I had read up on (even in wikipedia) last spring
what dot and concat was. Took appart patches and tried many combos.

And for these patches I posted yesterday I had even tried implementing those two [dot] & [concat]
but could not work it correctly.

Finally, Helping you help me (if possible)

Color/pixels:

Normal Jitter: order a r g b
Gl: r g b a

luma : has anything to do with black and white and alpha luminance.
alpha: has to do with transparency of your pixels

This combo [vec r g b (r or g or b)] :

Another trick:

- use jit.rgb2luma: to turn an image into black and white

- use [> ]: to then make the colors of image to become say "binary"

you can then flip the colors using the
[!- 1.]

to then play with the colors
jit.gl.slab @file cc.scalebias.jxs
or built it jit.gl.pix

[swiz x] [swiz y] [swiz z]
[* ] [* ] [* ] <
[+ ] [+ ] [+ ] <
[vec 0 0 0 1 ] <

---
Now, [dot] and [concat]:

Hypothesis:

[concat]: is sorta like an [append ].
[dot]: Have no clue what the dot product of ..... ???

THanks so much for any help on this and sorry for the long post,
But as I get better @ this, the more I can help others myself :-)

thanks

phiol

https://cycling74.com/tutorials/your-first-shader/
and
https://www.khanacademy.org/math/linear-algebra/vectors_and_spaces/dot_cross_products/v/vector-dot-product-and-vector-length

are very informative

@LSKA Thanks for the patch.

I have a question though:
When I'm putting the r value in the a channel, isn't my red a Value of 1?
Why would my red be a value of 0 ?

thanks a lot

@PHIOL, consider you're working with vectors (the corresponding of planes in a Jitter matrix), so with an array of values. Each pixel has a different value. Basically, what you see as black is value 0, and what you see as white is value 1.
Hence, if put your vignette's r channel in the composition's alpha channel, the corresponding values will set transparency.
Instead, the alpha plane coming from jit.grab has all cell values set at 1.

HTH

Thanks for the response LSKA!

I guess i'm not explaining what I know clearly. Although a long post (3 responses ago), I thought I had done so.
I don't consider myself a noob with Jitter (will post my work when I can) but a noob with jit.gl.pix

As mentioned I did know
Jitter char 0-255 argb
and glcolor float32 0.-1. rgba.

>>Instead, the alpha plane coming from jit.grab has all cell values set at 1.

shouldn't the swiz r be ever moving as such??

Endless thanks for time and knowledge LSKA