Questions about MA_RAMP_LINEAR #98856

New Issue

zitong · 2022-06-13T18:54:17+02:00

zitong commented

2022-06-13 18:54:17 +02:00

In source/blender/blenkernel/intern/material.c 1821 MA_RAMP_LINEAR
I'm curious if this

if (col[0] > 0.5f) {
    r_col[0] = r_col[0] + fac * (2.0f * (col[0] - 0.5f));
}
else {
    r_col[0] = r_col[0] + fac * (2.0f * (col[0]) - 1.0f);
}

is same as

 r_col[0] = r_col[0] + fac * (2.0f * (col[0] - 0.5f));

?
If not, then for what? Thanks!

In source/blender/blenkernel/intern/material.c 1821 MA_RAMP_LINEAR I'm curious if this ``` if (col[0] > 0.5f) { r_col[0] = r_col[0] + fac * (2.0f * (col[0] - 0.5f)); } else { r_col[0] = r_col[0] + fac * (2.0f * (col[0]) - 1.0f); } ``` is same as ``` r_col[0] = r_col[0] + fac * (2.0f * (col[0] - 0.5f)); ``` ? If not, then for what? Thanks!

zitong commented

2022-06-13 18:54:17 +02:00

Added subscriber: @zitong

Scurest commented

2022-06-13 19:07:07 +02:00

Added subscriber: @scurest

tempdevnova commented

2022-06-13 20:09:22 +02:00

Added subscriber: @tempdevnova

tempdevnova commented

2022-06-13 20:09:22 +02:00

1243c2bdae/source/blender/blenkernel/intern/material.c (L1821)

I am not sure but ramp_blend(), which is the function containing MA_RAMP_LINEAR, seems to be some implementation of the mixRGB node or something similar (what is it exactly?).
But what I can tell you is that

"if (col- [x] > 0.5f) {

  r_col[0] = r_col[0] + fac * (2.0f * (col[0] - 0.5f));

}
else {

  r_col[0] = r_col[0] + fac * (2.0f * (col[0]) - 1.0f);

}"

Is an implemention of the linear light blending mode.
Basically linear light is a combination of linear dodge (called add in Blender) and linear burn and is in the group called contrast blend modes, which is why it has the if...else statement.

"is same as

r_col- [x] = r_col- [x] + fac * (2.0f * (col- [x] - 0.5f));
?
If not, then for what? Thanks"

I did some testing:
Color_blend_test_file.blend

And yes... it is so.

In fact (2.0f * (col- [x] - 0.5f))=(2.0f * (col- [x]) - 1.0f) is a mathematical identity so we are calculating the same function in both statements.
As for the reason the top version is implemented in Blender is probably historical.
Back when Adobe first invented those blend modes all contrast blend modes including linear light came with this same logic.
See: https://www.deepskycolors.com/archivo/2010/04/21/formulas-for-Photoshop-blending-modes.html
The thought was that as we have an 'overlay' operation, which is a combination of multiply and screen, we should also have one for linear burn and linear dodge(add), which is what they called linear light.

After that no one ever bothered to optimize the implementations, which is why to this day Blender and all other software use the top version as linear light.

By the way what does the f in e.g. (col- [x] > 0.5f) mean and why does sit exist?

https://github.com/blender/blender/blob/1243c2bdae39a0d3f7644e39e24d2cc3e901e03b/source/blender/blenkernel/intern/material.c#L1821 I am not sure but ramp_blend(), which is the function containing MA_RAMP_LINEAR, seems to be some implementation of the mixRGB node or something similar (what is it exactly?). But what I can tell you is that "if (col- [x] > 0.5f) { ``` r_col[0] = r_col[0] + fac * (2.0f * (col[0] - 0.5f)); ``` } else { ``` r_col[0] = r_col[0] + fac * (2.0f * (col[0]) - 1.0f); ``` }" Is an implemention of the `linear light` blending mode. Basically `linear light` is a combination of `linear dodge` (called add in Blender) and `linear burn` and is in the group called `contrast blend modes`, which is why it has the if...else statement. "is same as r_col- [x] = r_col- [x] + fac * (2.0f * (col- [x] - 0.5f)); ? If not, then for what? Thanks" I did some testing: [Color_blend_test_file.blend](https://archive.blender.org/developer/F13157246/Color_blend_test_file.blend) And yes... it is so. In fact (2.0f * (col- [x] - 0.5f))=(2.0f * (col- [x]) - 1.0f) is a mathematical identity so we are calculating the same function in both statements. As for the reason the top version is implemented in Blender is probably historical. Back when Adobe first invented those blend modes all `contrast blend modes` including `linear light` came with this same logic. See: https://www.deepskycolors.com/archivo/2010/04/21/formulas-for-Photoshop-blending-modes.html The thought was that as we have an 'overlay' operation, which is a combination of multiply and screen, we should also have one for `linear burn` and `linear dodge(add)`, which is what they called `linear light`. After that no one ever bothered to optimize the implementations, which is why to this day Blender and all other software use the top version as `linear light`. By the way what does the `f` in e.g. `(col- [x] > 0.5f)` mean and why does sit exist?

zitong commented

2022-06-14 05:20:43 +02:00

Thank you very much @tempdevnova！

And about the f is mean the constant before is a float type and not a double.

See: https://docs.microsoft.com/en-us/cpp/c-language/c-floating-point-constants?view=msvc-170

A floating-point constant without an f, F, l, or L suffix has type double. If the letter f or F is the suffix, the constant has type float. If suffixed by the letter l or L, it has type long double.

Thank you very much @tempdevnova！ And about the `f` is mean the constant before is a float type and not a double. See: https://docs.microsoft.com/en-us/cpp/c-language/c-floating-point-constants?view=msvc-170 ``` A floating-point constant without an f, F, l, or L suffix has type double. If the letter f or F is the suffix, the constant has type float. If suffixed by the letter l or L, it has type long double. ```

Ray molenkamp commented

2022-06-14 06:19:01 +02:00

Added subscriber: @LazyDodo

Ray molenkamp commented

2022-06-14 06:19:01 +02:00

Changed status from 'Needs Triage' to: 'Archived'

Ray molenkamp closed this issue

2022-06-14 06:19:01 +02:00

Ray molenkamp commented

2022-06-14 06:19:01 +02:00

The bug tracker is just for bugs, if you have questions about the code, please use devtalk.blender.org

Sign in to join this conversation.

No Label

Download

What's New

Blender Studio

Manual

Developers Blog

Documentation

Benchmark

Blender Conference

Development Fund

One-time Donations

Questions about MA_RAMP_LINEAR #98856