Question about jit.gl.textmult

jit.gl.textmult basically creates individual jit.gl.text for each item of a list prepended by the word "text". So if you send "text hello world" it will create two instances of jit.gl.text, one for "hello" and the other for "world".
The thing is, you need to manually set the position of each jit.gl.text, and you have no idea of the size of the rendered word's mesh. The only way I know to get that size is to use [jit.gl.text @mode 3D] and send a "getbounds" message to it. From that you could compute the word's size, and so be able to place each word relatively to each other to recreate your sentence. However, from my understanding, jit.gl.textmult doesn't allow to send messages to its inner jit.gl.text. So what you ask for would require to either:
- Modify the existing js code of jit.gl.text so that you can get the size of each word and place them accordingly
- Recreate your own jit.gl.textmult tailored for your very own use (you have multiple ways of programmatically creating several objects)
- Use a completely different approach such as jit.gl.sketch in combination with jit.gl.text2D or jit.gl.text3D, or if you want to stay in 2D, you could use the MGraphics API which allows to draw text and has the convenient text_measure(text) method which returns the computed text size with defined font, and font size.

Thanks TFL!

Yes, I need to interact with each jit.gl.text independently so I can control all the various parameters for each. I'm interested in being able to pull a word or words apart by the letters, and then put it all back together again. In other words, a collage of letters might enter in, randomly distributed, but then coalesce to form the original word(s). Or perhaps with a longer text, the words enter individually, and then coalesce into a sentence. This would require complete control over each word/letter with all the various jit.gl.text parameters. Maybe its too much to ask, but then it seems like anything is possible with jitter.

I think the idea of creating a custom jit.gl.textmult is the best idea, though I'm not sure I am have the chops to do that. Perhaps if I had a simple example I could take it from there...

If you're comfortable with js, modifying the existing jit.g.textmult should not be too hard.
For the first step (getting size of each jit.gl.text), I would add

this.GetBounds = function()
    {   
       return this.glText.bounds;
    }

into the TextObj() function, and modify the SetTextAndDrawBang() function to call g_textObj.GetBounds for each created g_textObj. From here, depending on how you want to handle each words position, you could compute each word's size from its bounds and store that size either in js or output it in Max using secondary outlet.

Another simpler solution could be to use a mono font that that any character will always be the same width, so the letter/word position calculation could be pretty straightforward.

Thanks again! Unfortunately my javascript skills are fairly non-existent, however if there is a straightforward Jitter solution I could mostly likely implement that.

I managed to modify jit.gl.textmult to get each inner jit.gl.text bounds. From that I managed to get a matrix with as much cells as there is words, and each cell is the dimension of the word.
Now, the hard part is to set the position of each of these words accordingly. Hard to do in jitter, as it is a more sequential problem (you want to accumulate the x dimension of each previous word, + adding a space between each of them, and the result of that big sum is the position of the current word).

I tried something for the positionning (see below) but it doesn't work as expected, don't know why (maybe i get the sampling coordinates wrong?). Anyway, you might be able to do something from that starting point! Curious to see the/any result!

//Genexpr code to compute position of each word from a matrix with dimensions of each of them.
//DOES NOT WORK AS EXPECTED - here just to get the idea

cursor = 0;

if (cell.x >0) {
    for (i=0; i<cell.x; i+=1) {
        prev_item_coord = vec( i/dim.x , 0 );
        cursor += sample(in1, prev_item_coord).x;
        cursor += space;
    }
}

out1 = vec(cursor, 0, 0);

From the beginning I speak about words, but the logic is the same if you want your words to get independant letters.

Great work TFL! This is definitely a start. However for me, what is most important is having control over each word so that they can be split apart and independently manipulated. Right now, if I change the position for example, all the words move together. How is it possible to change the position of each fragment of text? It may be that what I am looking for is fairly straightforward, since the goal isn't perfect alignment of the sentence, but rather creatively positioning and animating its parts.

I will provide an example of what I am doing with one word (I'll have to prepare the patch first) so you can imagine the same thing with multiple words that all belong to the same phrase or sentence.

Right now, if I change the position for example, all the words move together.

Yes, this is because of the jit.gen and my weird code to try to position words one after the other, but you still have complete control over the absolute position of each word. Oh and maybe that the named matrix "dims" that might be confusing. (as I'm filling it at the bottom of the patch, but using its values to compute words positions.
But you can send the positionning matrix into the first inlet of jit.gl.textmult. This matrix has to be of 3 planes (for x, y and z position), and of dim "N 1" where N is the number of inidivual text elements (hence the zl.len after the message object with the text to display). First cell is the position of the first word, and so on.

Here is a stripped down example allowing you to set the position of the first to words independently (first you need to save the patch into the same folder as the other files I sent you in my previous post, close the patcher, then re-open it)

We're definitely making progress! As you can see below, when I added additional setcell's, I was able to move each of the words around. It looks like for each word, no matter how many, it inhabits its own jit.gl.text? You have XY positioning working for each word, but what about all the other jit.gl.text paramaters (I am using 3D mode) so that would include using a jit.anim.drive object for animation. How would you connect the animation object to each individual instance? Also, I ran into an error, see below.

When I set the window to floating, something was broken in the patch, looks like the binding was disabled. I wasn't able to control the words so they are still collapsed.

jit.gl.textmult functions similarly to jit.gl.multiple, so if you want additional transform parameters you provide those as arguments to the glparams attribute in the object box, preceded by a number indicating the number of matrix inputs you want.

If you want to control these transforms via anim.drive, you should create an anim.drive and anim.node for each word you want control over, and send their tranform values to the corresponding matrix cell every frame. This is perfectly suited for something like poly~, where each instance of the poly contains the anim.node and anim.drive combo, and the poly~ index is used to determine which cell of the matrix to set.

The context errors when floating is toggled are unfortunate side effects of the underlying tranform-feedback. The workaround is to re-initialize the text after toggling floating.

Basic example of controlling 3 words below.