GR meter and inline display have a fixed release time about dpl.lv2

[x42]

The plugin uses the exact same value for both the GUI and the inline display, there is no separate value calculated. Both GUI and inline display use the gain-factor that is used to attenuate the signal as is.

Compare the falloff from -20 to 0. It matches in both custom and inline GUI:

[x42]

The same value (self->_peak) is used for inline display and the output port:

dpl.lv2/src/lv2.cc

Line 292 in 3ad9b0e

const float display_lvl = enable ? fmaxf (-10.f, self->_peak) : -100.f;

dpl.lv2/src/lv2.cc

Line 303 in 3ad9b0e

*self->_port[PLIM_LEVEL] = enable ? fmaxf (-10.f, self->_peak) : -10;

The latter (PLIM_LEVEL), is what is used in the GUI to draw the gain-reduction meter.

If there is a difference it could be related to display updates, redraw FPS (Ardour's inline display is updated during idle), or perhaps both are delayed respective to each other which may be noticeable with large buffer-sizes?

Either way there is nothing that can be done on the plugin's side.

[magnetophon]

Thanks.

The video you made actually shows the problem really well.
Turns out both the GR meter in the plugin and the inline display are wrong, imho.
Only the time graph shows the value that we are hearing, iow the correct value.
After the GR has dropped to 0dB in the graph (and the audio), the meters are still slowly dropping off.

Do you agree?

[x42]

Yes, this is intentional.
The history display can show instant changes, however you cannot see that on the meter, that would just flicker (if at all), since an audio peak is shorter than a video-frame, you could even miss reduction.

The gain-reduction meter has dedicated ballistics: it falls off 6dB/sec. This is quite common for limiter redux meters.
Just like a needle meter, the needle cannot jump.

[magnetophon]

Yes, this is intentional. The history display can show instant changes, however you cannot see that on the meter, that would just flicker (if at all), since an audio peak is shorter than a video-frame, you could even miss reduction.

Right, I understand your point of view.
For me, the actual GR would still be much more informative, for 2 reasons:

1. Currently, a big spike in the GR will mask following smaller GR's, also causing you to miss them.
2. When I read a GR meter, I am not only interested in the amount of reduction, but also in how long it lasts, especially in a plugin with automatic release time variation, like this one. When there is only a short moment of GR, I am more interested in knowing it was short than I am in knowing how many dB it was.

I see two possible solutions:

1. Add a switch to choose between the real and the slow GR.
2. Speed up the fall off rate.

I strongly prefer solution 1, because solution 2 does not solve the 2nd problem.