Trash 2: strange DC Offset
- KVRist
- Topic Starter
- 180 posts since 18 Nov, 2017
Hi people,
I'm using the awesome Trash 2 plugin to get some fancy distortion fxs.
I've a problem with a preset I'm working on.
Here's the screenshot:
As you can see, its "bipolar", with the center point 0/0. But if I disable DC Offset, I have a constant DC signal coming out from the plugin.
I'd like to understand why this, learning waveshaping. at 0.0 it should be 0.0. Why that offset?
Thanks
I'm using the awesome Trash 2 plugin to get some fancy distortion fxs.
I've a problem with a preset I'm working on.
Here's the screenshot:
As you can see, its "bipolar", with the center point 0/0. But if I disable DC Offset, I have a constant DC signal coming out from the plugin.
I'd like to understand why this, learning waveshaping. at 0.0 it should be 0.0. Why that offset?
Thanks
- KVRAF
- 8823 posts since 6 Jan, 2017 from Outer Space
Its not point symmetric around 0/0. So if your input signal has no dc, the output will get one. As in your example the negative high values will go back to zero and thus you will weight the negative values less than the positive ones...
Without a dc blocker you’ll never find the good spot for you center, as it also changes with the input level. In your example on maybe -8 dB you will instead get a negative dc...
Feed a sine wave and play with the input level...
Without a dc blocker you’ll never find the good spot for you center, as it also changes with the input level. In your example on maybe -8 dB you will instead get a negative dc...
Feed a sine wave and play with the input level...
- KVRist
- Topic Starter
- 180 posts since 18 Nov, 2017
Thanks for the fast reply dude!Tj Shredder wrote: ↑Mon Jan 20, 2020 2:36 pmSo if your input signal has no dc, the output will get one.
Not sure I got the point though
NO DC input = 0.0 as input value.
Thus, having x=0.0 means y=0.0 in the graph above:
That's why I don't get a +DC offset value...
-
- KVRian
- 778 posts since 21 Sep, 2008
Yeah, this should not be happening. Why it happens may range from 0 signal being not truly 0 in your DAW (maybe the dither signal is only +1 bit) or some glitch with the math on the plugin.
- KVRist
- Topic Starter
- 180 posts since 18 Nov, 2017
DAW input is 0.0 (i've analysed the samples).
Even if it was 1e-6, the added DC offset is very huge (somethings like -4DB).
Probably a Trash 2 bug. Where can I report it? Can't see any official iZotope forum...
- KVRAF
- 4590 posts since 7 Jun, 2012 from Warsaw
Tj Shredder already gave correct and detailed explanation of this one, so learn from smarter ones instead of weaving your own nonsense...
DC means the mean value of signal is not 0. If you feed a symmetric signal through asymmetric effect, the resulting signal will also be asymmetric and thus likely have mean value different than 0. This is what every non-linear effect does to a signal. it's not a bug, it's simple maths.
DC means the mean value of signal is not 0. If you feed a symmetric signal through asymmetric effect, the resulting signal will also be asymmetric and thus likely have mean value different than 0. This is what every non-linear effect does to a signal. it's not a bug, it's simple maths.
Blog ------------- YouTube channel
Tricky-Loops wrote: (...)someone like Armin van Buuren who claims to make a track in half an hour and all his songs sound somewhat boring(...)
Tricky-Loops wrote: (...)someone like Armin van Buuren who claims to make a track in half an hour and all his songs sound somewhat boring(...)
- KVRist
- Topic Starter
- 180 posts since 18 Nov, 2017
But... did you read the post? I don't feed any signal to it (I'm talking about this, not processing a signal). Thus "all" 0 value samples.DJ Warmonger wrote: ↑Tue Jan 21, 2020 8:20 am DC means the mean value of signal is not 0. If you feed a symmetric signal through asymmetric effect, the resulting signal will also be asymmetric and thus likely have mean value different than 0. This is what every non-linear effect does to a signal. it's not a bug, it's simple maths.
What's the "mean" of a all 0 data, my hero?
- KVRAF
- 4590 posts since 7 Jun, 2012 from Warsaw
How about "drive" control? Depending on algorithm, it might introduce some bias.
Blog ------------- YouTube channel
Tricky-Loops wrote: (...)someone like Armin van Buuren who claims to make a track in half an hour and all his songs sound somewhat boring(...)
Tricky-Loops wrote: (...)someone like Armin van Buuren who claims to make a track in half an hour and all his songs sound somewhat boring(...)
- KVRist
- Topic Starter
- 180 posts since 18 Nov, 2017
Tried all settings in the right panel.DJ Warmonger wrote: ↑Tue Jan 21, 2020 9:14 am How about "drive" control? Depending on algorithm, it might introduce some bias.
Drive/Pre 0/100... changes nothing.
Of course Mix to 0 remove the DC, but because it won't mix the processing
-
- KVRian
- 778 posts since 21 Sep, 2008
- KVRist
- Topic Starter
- 180 posts since 18 Nov, 2017
Here's the reply from the support:
"This is a normal behavior of Trash 2.
When enabling BiPolar, on an asymmetrical waveform, you will get 'noises' from it, because there is some type of phasing that is happening when the BiPolar feature is engaged"
Not really sure about this answer. Phasing on zero-input signal? Uhm...
"This is a normal behavior of Trash 2.
When enabling BiPolar, on an asymmetrical waveform, you will get 'noises' from it, because there is some type of phasing that is happening when the BiPolar feature is engaged"
Not really sure about this answer. Phasing on zero-input signal? Uhm...
-
- KVRian
- 986 posts since 8 Mar, 2009
I got a similar answer when i asked about this years ago and another similar bollocks answer about why when you move the centre axis from zero even the tiniest amount you get a rush of noise when the sound ends(check the crispy and capacity algorithms for a demonstration of it)they said that is how all waveshapers behave if you was to do the same thing(also bollocks)don't hold your breath that they will do anything about either of these issues
I
-
- KVRian
- 642 posts since 28 Oct, 2014
There's a similar feature in Melda's MWaveShaper and you get exactly the same behaviour. It's just the way these things work. You can place a high pass filter afterwards to mitigate the effect somewhat.