TDR Nova spectrum analyzer lag (still happening in Nova 2.0)

[thebutler]

In all cases, we have major updates in the works.

Interesting...

I have some problems running TDR plugins in WaveLab Pro. Mainly keyboard focussing quirks; other reports of 'issues' on Steinbergs forum from other folk too (blaming WL of course..!). I also posted there, pointing the finger at WL and asking them to examine the behaviour - since there were no such issues in other apps. Yet to see a response.

Are you aware of any similar reports yourselves.?

In any case, as I said, no problems seen in Cubase Pro for example - it seems just WL that is very 'picky'.

Anyways, looking forward to results of your recent hard work - all TDR plugins I have are fab.!

Cheers,
Bob

PS:- any VST3 love just yet.?

[FabienTDR]

About Wavelab, we helped steinberg fix the year old "sticky knob" bug, and they rolled it out one, two versions ago! Consider updating Wavelab, it should fix the issue.

[thebutler]

About Wavelab, we helped steinberg fix the year old "sticky knob" bug, and they rolled it out one, two versions ago!

Wow - sounds good.! Though I was talking about 'keyboard focus' problems (using computer keyboard shortcuts). Not sure how that relates to a 'sticky knob' (which I would immediately assume to be a mouse/UI issue)..?

Consider updating Wavelab, it should fix the issue.

You'll see in my sig that I am pretty much up to date with WL Pro editions.

Apologies to the OP - getting well off topic here

[Reefius]

I recently noticed that in Cubase (9.5 Pro) the analyzer works perfectly in sync, while in other DAW's it lags behind.

[Reefius]

Version 2.0 analyzer is still lagging behind. It's better than 1.x but still a big difference with other spectrum analyzers like SPAN or Fabfilter Pro 2.

Also I think I found a bug in 2.0: whenever I enable the THRES button for any band (with the knob still at 0 dB so nothing changes to the sound), the analyzer suddenly drops down a lot when set to IN. It actually shows the same graph as when it's set to SC (which should not show anything because there is no sidechain audio coming into Nova).
Only when the analyzer is set to OUT it shows normal values.

I used to think it was my graphics card back in the day, because I was using onboard Intel 2500 graphics. But I recently got a new PC with a Geforce 1060 and the issue is still there. I can run games with all settings maxed so the graphics card is not the problem anymore.

[Tappistry]

Version 2.0 analyzer is still lagging behind. It's better than 1.x but still a big difference with other spectrum analyzers like SPAN or Fabfilter Pro 2.

Also I think I found a bug in 2.0: whenever I enable the THRES button for any band (with the knob still at 0 dB so nothing changes to the sound), the analyzer suddenly drops down a lot when set to IN. It actually shows the same graph as when it's set to SC (which should not show anything because there is no sidechain audio coming into Nova).
Only when the analyzer is set to OUT it shows normal values.

I used to think it was my graphics card back in the day, because I was using onboard Intel 2500 graphics. But I recently got a new PC with a Geforce 1060 and the issue is still there. I can run games with all settings maxed so the graphics card is not the problem anymore.

Also seeing the analyzer lag here on the new version. Tried it in two different hosts: Reaper and Patchwork.

I'm running Windows 10, 64-bit everything if that matters. I'm currently trying to get a video with sound to show the difference between Pro-Q2 and Nova but I need to play with input/output sources in Reaper or something to get the audio to come through.

[FabienTDR]

It's on our TODO list, but I am not sure wether we can improve it or not.

If you need a fast/tight analyzer, you are likely better served by a dedicated analyzer plugin. Or maybe a comparably low cpu static EQ (like pro Q and similar).

SPAN and EQs can analyze the input without having to wait for the output to come out of the "club". A dynamic EQ (like any other nonlinear device), must be measured at the output. This alone gives a head start, a tool like Nova on the other hand must "wait" for any processing latency related to lookahead or resampling.

Linear processors can be predicted, so only input has to be analyzed. This allows to walk around most sources of delay.

Nonlinear processors can't be predicted, so only output can be analyzed. This wastes a lot of time.

A super high quality dynamic EQ resampling to 400kHz internally, both asking for heavy DSP cpu load and heavy screen-repaints (constantly recalculating and repainting curves on screen) is slightly oversized for the task. A bit like joining a car race with a tank.

Horses for courses!

We don't delay anything explicitly, beside the unavoidable audio buffer, FFT buffer, and of course the plugin's own latency (resampling, lookahead). But using a dynamics processor as analyzer, and expecting it to behave like or better than a dedicated analyzer is not realistic.

[pumafred]

Mix with your ears, not your eyes... (Do I get a free plugin for this support, Fabien? )

[2NDMOUSE]

I've had the Nova GE a while now, but the thing that's always stopped me using it is the visual lag.
I set up a kick as a sidechain signal, adjust the threshold / ratio to where I can see the signal affect the EQ curve, but there is latency between what I hear and what I see the EQ curve doing, and it moves too slow.

It doesn't seem to be an issue with other plugins that give visual readouts such as Fabfilter Pro EQ 2.

What I'd like is an immediate response that is timed exactly with what the sidechain signal is doing in real time.
I have downloaded the new update today as this issue was apparently addressed, and I had high hopes that it would be but I still have visual lag. Nice that I can resize the GUI though so thanks for that.
Any way to fix this lag issue ??

Im using FL 20.0.5 ( build 681 ) 64bit, with Windows 8.1 ( 64bit) and 8GB of ram.
Cheers

[Reefius]

It's on our TODO list, but I am not sure wether we can improve it or not.

If you need a fast/tight analyzer, you are likely better served by a dedicated analyzer plugin. Or maybe a comparably low cpu static EQ (like pro Q and similar).

SPAN and EQs can analyze the input without having to wait for the output to come out of the "club". A dynamic EQ (like any other nonlinear device), must be measured at the output. This alone gives a head start, a tool like Nova on the other hand must "wait" for any processing latency related to lookahead or resampling.

Linear processors can be predicted, so only input has to be analyzed. This allows to walk around most sources of delay.

Nonlinear processors can't be predicted, so only output can be analyzed. This wastes a lot of time.

A super high quality dynamic EQ resampling to 400kHz internally, both asking for heavy DSP cpu load and heavy screen-repaints (constantly recalculating and repainting curves on screen) is slightly oversized for the task. A bit like joining a car race with a tank.

Horses for courses!

We don't delay anything explicitly, beside the unavoidable audio buffer, FFT buffer, and of course the plugin's own latency (resampling, lookahead). But using a dynamics processor as analyzer, and expecting it to behave like or better than a dedicated analyzer is not realistic.

I understand there could be a delay when the analyzer is set to show the output, but why is there also a delay when it's set to only show the input? I would be quite happy with a visualizaton of just the unprocessed audio, as long as it's not lagging behind.

[FabienTDR]

It seems that side-chaining is a factor? If anybody finds the time, I'd greatly appreciate to see a video of two dynamic EQs, sidechained, showing Nova's the SC spectrum lagging behind another dynamic EQ's analyzer.

Do quality modes or the OpenGL option maybe make any difference for you?

I personally experience no visual lag from the analyzer in Reaper or any other host. I can wildly automate a sine and get visual representation "in time", perfectly following the automation curves.

[Reefius]

Sidechaining is not even needed, the analyzer in Nova always lags for me.
Here is a video of Fabfilter Pro Q3 and Nova next to each other with a simple kick drum beat:

https://www.youtube.com/watch?v=8DxaXxztRIk

As you can see, Pro Q3 is right on the beat, while Nova is between the beats.

[FabienTDR]

Ah thanks for this. Looks like a difference in smoothing, we could try to offer smaller FFT windows.

[heavymetalmixer]

Ah thanks for this. Looks like a difference in smoothing, we could try to offer smaller FFT windows.

How much time in average would it take you to fix this?

[neverbeeninariot]

It's not just Nova that is laggy, it's also SlickEQ and SlickEQ M. If you put them side by side with the same source material, you see the same reaction times, even though their display resolutions are different (Nova shows much more detail in the analyser window than the Slicks).

If you put MEqualizer side by side with them, by default it has near the same reaction time as the TDR EQs (ie. it lags). However, you can tweak MEqualizer's analyser to (almost) match Pro-Q3 by setting Averaging to ~75ms, Decay to ~50%, Time Resolution to ~15%, Hamming type windowing, 4x overlap, and FFT to 4096. Smoothing is at 1% (default).

Not sure if this is useful information, or I just have too much time on my hands this evening, but it would be nice to have a more responsive display in the TDR EQ's.