Erratic curve behavior in the tone equalizer when curve smoothing is set to high values.

[michelklaasen]

Describe the bug

I am in the tone equalizer in the advanced tab and I have a curve set up.

Setting the curve smoothing setting to low values, -1 to 0, works as expected.

However, when I pull the curve smoothing setting to higher values, from about +0.58, the transitions in the curve become a bit clunky. At 0.83 specifically, the curve becomes clunky, and at 0.84 it continues at normal again.

Then at certain specific values, 0.90, 0.96, 0.98, 0.99, the curve will max out and set all values to EV values to +2, forming a straight line at the top of the graph. At 1.00, the curve behaves as expected again.

Steps to reproduce

1. Open image in darkroom
2. Make changes to the tone equalizer graph in the advanced tab
3. Set curve smoothing to high values, specifically to 0.83, 0.98 or 0.99.

Expected behavior

Make the curve smoother, not crazy

Logfile | Screenshot | Screencast

Screenshot showing several values of the curve smoothing curve. Some normal (0, 0.84, 1.0), one that's just weird (0.83) and some that are maxed out (0.98, 0.99).

I don't know where to find logfiles.

Commit

I don't know

Where did you obtain darktable from?

downloaded from www.darktable.org

darktable version

darktable 4.8.1

What OS are you using?

Windows

What is the version of your OS?

Windows 11 Pro 22H2 Build 22621.4317

Describe your system?

This PC: Gigabyte Aorus Ultra Gaming x470 AMD Ryzen 9 5900X 32 GB RAM AMD Radeon 6900XT

Are you using OpenCL GPU in darktable?

Yes

If yes, what is the GPU card and driver?

AMD Radeon 6900XT Driver Version: 24.20.19.01-241011a-408380C-AMD-Software-Adrenalin-Edition AMD Windows Driver Version: 32.0.12019.1028 16GB of memory

Please provide additional context if applicable. You can attach files too, but might need to rename to .txt or .zip

I'm running script manager in darktable with only the OpenInExplorer script enabled. This happens on every image I tried it upon (I only have RAW images from Canon R6 Mark II) at the exact values I mentioned.

[jenshannoschwalm]

There check for bad solving is not correctly reported back.