Stock Table Upgrade?

DanH

@Oli-Ullmann yes backwards compatibility is the main issue here. The table is everywhere in Hise. Perhaps the existing table can have a switch to choose between types of table, 0 for the stock Hise table, 1 for the new 'flex' table or whatever...

griffinboy

@Oli-Ullmann

Indeed, that's my thinking. It would be nice for table 2.0!

Although, I'm likely to do it custom no matter what, as is my preference for high control!

ustk

@dannytaurus said in Stock Table Upgrade?:

I expect we all want slightly different sets of interactions with tables

Yes that's why I went for panel, if you need more behaviour like smoothing or different cubic/quadratic splines.

HISEnberg

@ustk Are you controlling the table with a panel in your case? I didn't know that was possible. If you are then it sounds like the simplest solution for a Table 2.0 just exists in HISEscript + a bit of math.

This article peaked my interest however!

https://signalsmith-audio.co.uk/writing/2021/monotonic-smooth-interpolation/

ustk

@HISEnberg You could, but there's no point interfacing a table when your panel addresses everything already. I just send the data to C++ node.
That being said, the panel is quite a beast...

griffinboy

@ustk

Idk if you have used a different strategy, but the way I'm doing it, is sending only the control points and tensions into c++
Since the GUI code only contructs the graph at the resolution it needs to for each segment, there is no point calculating a high resolution graph on the hise side of things.

The c++ then constructs the actual modulation shape in high resolution to be used for audio processing.

ustk

@HISEnberg Yep this article is nice. In the end it all depends on the interaction you want with the curve, because there would be as many spline constructions as desired interactions... I needed mine with curve passing through the points, and no possible overshoot on the y axis (when 2 points have same x, it's forbidden for the curve to pass those points (like in standard table) which isn't possible with standard cubic unless all control points are permanently resizing to follow their neighbours

ustk

@griffinboy Yeah same here
I just send the nodes and tension information (called smoothing here) to C++ and build the curve in there...
We should have worked together on this lol

griffinboy

@ustk

haha indeed, we have the same system.

HISEnberg

@griffinboy @ustk Thanks for sharing! I think I am going to build my own system now and we can see who has the smoothest curves

ustk

@HISEnberg sick! a contest! You have to know mine is as smooth as baby skin...

griffinboy

@ustk @HISEnberg

jokingly
If we make this a contest, I'll just be putting my table through my super strong antialiasing resampler. Can't beat antialised tables...

That said, we could probably pump out a hell of a table, if we pooled our resources lol

Oli Ullmann

Here's some info for you that I noticed recently:
If you use the content callback of a table as follows:

Table.setContentCallback(tableCallback);

inline function tableCallback(i)
{
	// Your code
}

i returns the index of the point that you control from outside, for example from a panel or a slider.
This was very useful for my use case.

Christoph Hart

Another issue is only being able to move points by a pixel at a time. This came up a few weeks ago and is apparently a Juce issue but it's a very poor resolution by modern standards so would love to investigate further.

What's that?

DanH

@Christoph-Hart https://forum.hise.audio/topic/13439/mouse-cb-drag-resolution?_=1763023869984

Christoph Hart

@DanH ah that looks very fringe, how is that a real problem? I never felt the urge of having a sub-pixel mouse position and while JUCE theoretically has float resolution, it's only used on macOS because the Windows API only returns the pixel position as integer.

DanH

@Christoph-Hart well accuracy, but really it's about the feel of it. We've been spoilt by much smoother resolutions in other plugins. I also find my ruler line moves in pixels as well?

Christoph Hart

Alright, I've played around with the table component a bit. There was already a function that allowed you to customize the drag behaviour to a certain extend, but I thought it would be super funny to not add any documentation to that thing whatsoever.

https://docs.hise.audio/scripting/scripting-api/scripttable/index.html#setmousehandlingproperties

This can be extended to allow pretty detailed customizations, so any input for more stuff is welcome. Here are the currently supported properties (I'll hold off with the docs for this method until we included all popular requests):

Table1.setMouseHandlingProperties({
	margin: 10, // adds a margin around the table so that points at the edges are not cut off
	fixLeftEdge: 0.5, // sets a normalized fixed y-position for the left edge
	fixRightEdge: 0.5, // sets a normalized fixed y-position for the right edge
    syncStartEnd: false, // if true, then the start and end y value will always be in sync
	endPointSize: 12, // changes the size of the points at the edges
	dragPointSize: 18, // changes the size of the draggable points
    midPointSize: 12, // changes the size of the curve node points (default = 0)
    allowSwap: true, // if true, then points can swap their position when dragged
    numSteps: 12, // this will enable a grid for snapping to the x-axis
    snapWidth: 10, // if numSteps > 0 is on, this will define the tolerance for the snapping in pixels
    useMouseWheelForCurve: false, // if true, it will use the mouse wheel for setting the curve. 
	closePath: false, // if false, it will not create a boxed and filled version of the path (like above)
});

I've also made a few minor changes (remove that nasty green outline, why???) Note that the useMouseWheelForCurve property replaces the compile time constant HISE_USE_MOUSE_WHEEL_FOR_TABLE_CURVE but it's not a breaking change because the default value for this property will be picked up by the constant.

Any ideas for more customizations are welcome now. Let's do this once right and then not think about that component for another 5 years...

Oh and it's not pushed yet, so you can't try it out yourself yet.

HISEnberg

@Christoph-Hart syncStartEnd is missing the boolean in your example!

Christoph Hart

@HISEnberg true.

Stock Table Upgrade?

15

2.0k

12.8k

111.4k