filter() incompatible with high pixel density on P2D and P3D

[processing-bot]

Created by: HectorCarral

Description

Using filter() (for example, filter(GRAY), filter(INVERT)) in combination with a higher pixel density (set for example with pixelDensity(displayDensity())) behaves weirdly when using P2D or P3D.

Expected Behavior

The filter would be applied with P2D/P3D and high pixel density in the same way as with the default renderer or with default pixel density.

Current Behavior

The frame is drawn zoomed in on the top left corner. If the default renderer is used, everything works as expected. If no higher pixel density is used, everything works as expected. Interestingly, if we use saveFrame(), the saved image looks fine even though it is rendered wrong in Processing. Therefore, I'm attaching here screenshots instead of saved frames.

Default renderer or default pixel density:

High pixel density and P2D/P3D:

Steps to Reproduce

1. Set P2D or P3D as the renderer.
2. Set a high pixel density.
3. Use filter().

void setup() {
  size(500, 500, P3D);
  pixelDensity(displayDensity());
  rectMode(CENTER);
  noLoop();
  
  background(0);
  noStroke();
  fill(255, 0, 0);
}

void draw() {
  rect(width*0.5, height*0.5, width*0.5, height*0.5);
  filter(GRAY);
}

Your Environment

• Processing version: Processing 4.0b6 (x64)
• Operating System and OS version: macOS 12.2.1
• Other information: Apple M1 chip

[processing-bot]

Created by: codeanticode

@benfry just came across this, my quick take on how to fix it is in this PR #429 but I may be missing something.

[processing-bot]

Created by: codeanticode

I went through the filter examples in the reference, even tried a few more involved cases (rendering into a PGraphics and then applying the filter on the PGraphics object), and so far so good!

[processing-bot]

Created by: benfry

Ok, shall I merge or are you still checking?

[processing-bot]

Created by: codeanticode

I'm done checking, you can merge.

[processing-bot]

Created by: github-actions[bot]

This issue has been automatically locked. To avoid confusion with reports that have already been resolved, closed issues are automatically locked 30 days after the last comment. Please open a new issue for related bugs.