I can confirm this issue.  Appeared with the update to Plasma 5.20.4.

I am using an Nvidia card and have a native install (not virtual)

I am using Arch as well.  This happens with the panel at top as well and persists into the second screen with the same effect.

Plasma 5.20.4
Frameworks 5.77.0
Qt 5.15.2

Not sure if it is related or if this needs to be another bug report, but some programs are no longer properly suspending composting.

I have several games in my steam library that do not suspend compositing like they should and did prior.  These are Linux native games. 

None of the games that run via proton 5.13-4 will suspend compositing when all did before.  This part may be related to a proton bug as previous versions suspend compositing properly.

Until this point, all native games did suspend compositing properly when the option to allow programs to suspend compositing was selected, so I think at least the native game issue might be related.  Maybe something was changed in the compositing settings that caused both these issues?

As I said above this might need to be a separate bug report but as it seems to appeared at the same time as the "ugly panel issue" I thought I would mention it here in case it will help you to track down the issue

*** Bug 430367 has been marked as a duplicate of this bug. ***

I did some further testing today and it looks like it was just an update that was preventing compositing from suspending on all my native games, or a failure to load something properly on boot temporarily as the issue went away during reboot so please disregard my comment about that as a possible direction to investigate.  

All but a few of my native games suspend compositing properly after a reboot.  The few that remain are just not coded properly to report a true full screen application it would appear.  

I apologize for a possible false trail, I should have rebooted and checked again before posting my comment....

Bisected.

Caused by c709076d9c52cbd177de18b3f199917f32d416dc in plasma-frameworks

Easier way to test:

diff --git a/tests/testborders.qml b/tests/testborders.qml
index 62c9c6ed7..10eea79cf 100644
--- a/tests/testborders.qml
+++ b/tests/testborders.qml
@@ -20,7 +20,7 @@ Item
     PlasmaCore.FrameSvgItem {
         id: theItem
 
-        imagePath: "widgets/background"
+        imagePath: "opaque/widgets/panel-background"
         anchors {
             fill: parent
             margins: 10

A possibly relevant merge request was started @ https://invent.kde.org/frameworks/plasma-framework/-/merge_requests/171

Git commit b1364fa601866792ecf69cf61b8ab48732150857 by David Edmundson.
Committed on 17/12/2020 at 12:41.
Pushed by mart into branch 'master'.

Fix broken background

In commit c709076d9c52cbd177de18b3f199917f32d416dc the SVG got somehow
broken leading to black rectangles in all cases where it was used.

This commit reverts to the last working version, then reapplies the
"thick margin" from the change.

A test is supplied to show all our frameSVGs to hopefully make this
easier to spot if it happens again.

M  +420  -80   src/desktoptheme/breeze/opaque/widgets/panel-background.svg
A  +41   -0    tests/frames.qml     [License: LGPL(v2.0+)]

https://invent.kde.org/frameworks/plasma-framework/commit/b1364fa601866792ecf69cf61b8ab48732150857

*** Bug 430482 has been marked as a duplicate of this bug. ***

As a curious thing, if I set a new user in kde and I login with it, the error is not visible.

(In reply to Roberto Orden from comment #10)
> As a curious thing, if I set a new user in kde and I login with it, the
> error is not visible.

That is because if you are using an older version of Plasma you don't have the bug, kubuntu for example.  If you using the newer versions like in Plasma Neon, Arch, etc the bug has been fixed

*** Bug 431339 has been marked as a duplicate of this bug. ***