Sunday, March 30, 2008

1 Prim Interface

1 prim interfaces aren't possible at this time, but has been something long wished. Basically, the idea is to just have one texture that has the whole interface, and put that on one (side of a) prim. This would save prims and would be less complex script wise.

The reason it isn't possible at the moment, is because we can't determine where is being clicked on a prim, only just that it is being clicked. Qarl Linden is working on new LSL functions that will give us the needed information, and more! Qarl talked about it on the latest Mono Office Hour. Qarl is thinking of a rough timeframe of about a month, until we can test it on the beta grid.

Colourised transcript for the parts about the new functions:
Periapse Linden: Ok, so our special guest today is Qarl.
Periapse Linden: The Linden who brought you sculpties
Periapse Linden: He's back with a great new idea he'd like to get feedback on.

Qarl Linden: so i have to give credit where it's due -
Qarl Linden: Seifert originally asked for this feature a year ago.
Qarl Linden: i got sidetracked and didn't get to it 'til now.
Qarl Linden:

Seifert Surface: i think people have been asking for it from waaay back in the day

Qarl Linden: yeah, true.

Vincent Nacon: yeah I've been wondering how that will work

Qarl Linden: what we've got here is the ability to get more information about touch events.

Vincent Nacon: by polygon?

Qarl Linden: you get to know where on the surface, and where in world the touch occurs.
Qarl Linden: you get the face number as well.

Vincent Nacon: err... how many faces are there?

Qarl Linden: it's continually updated as the mouse is held down - so you can implement dragging.
Qarl Linden: 6 faces on a cube - etc.

Seifert Surface: vincent: faces as in faces on a prim

Chance Abattoir: Where on the surface, that's rad!

Seifert Surface: 1 face on a default torus

Qarl Linden: and the POSITION, as well.

Vincent Nacon: oh was thinking about sculpty part

Mbrb Rau: If our AV was a prim it's probably a sculptie.

Qarl Linden: texture coordinates, etc.

Mbrb Rau: and uses llTakeControls

Qarl Linden: right now we're asking for feedback - make sure we don't miss anything important.

Seifert Surface: presumably if we ever get 2 sided planar sculpties, theyd have 2 faces

Qarl Linden: presumably. :P

Seifert Surface: :P

Vincent Nacon: wait... didn't we already have something like that
Vincent Nacon: the drag I think

Qarl Linden: there's a similar call: llDetectedGrab().

Vincent Nacon: yeah

Qarl Linden: but it's not so good for a variety of reasons...

Vincent Nacon: yeah, tried to make a joystick imput by using that
Vincent Nacon: input*

Qarl Linden: it reports movement relative to the camera position...

Vincent Nacon: yeah

Qarl Linden: not so useful.

Vincent Nacon: it was neat at first but meh

Qarl Linden: now you'll be able to put an entire GUI into a texture, and determine with the position info which button was clicked.
Qarl Linden: chess boards, for instance - won't need 64 prims.

Seifert Surface: the continual updating works with the touch event firing lots right?

Siann Beck: That's awesome. A lot of vendors can be re-designed.

Qarl Linden: exactly.

Vincent Nacon: so... if we're going to have touch_position on surface, what does it means for html-on-prim?

Seifert Surface: right

Mbrb Rau: 1 prim vendors...

Qarl Linden: yes - that's one of the motivations for this work -
Qarl Linden: so that prim web browsing will be easier for us to implement.

Vincent Nacon: yeah, figured
Vincent Nacon: well that's good, hope it won't be much of a problem with flash

Rex Cronon: could it be possible for a script to load a notecard that has html code on a specific face?

Qarl Linden: flash is hard because mozilla has a truly horrible plugin interface.

Vincent Nacon: now that's a thought

Qarl Linden: we're looking at webkit and others for that...
Qarl Linden: maybe notecard->html ... but that's not REALLY useful...

Vincent Nacon: so how the position will work? gobally or local?

Seifert Surface: rex: really you want to be able to alter the webpage on the fly from a script

Qarl Linden: only if you can't get your own webpage.

Seifert Surface: say for a scoreboard or something

Qarl Linden: yes. exactly.
Qarl Linden: but you can do a LOT by passing parameters to a webpage in the URL.

Rex Cronon: i want to have display that is not connect to a web page

Vincent Nacon: and for non-flat surface... how will that set the local position? based on UV map maybe?

Seifert Surface: html file in a string?

Qarl Linden: i've seen an example where HTML is in the parameter...
Qarl Linden: yes - it's the UV map.

Vincent Nacon: ah ok perfect

Chance Abattoir: When will this script go into testing?
Chance Abattoir: function

Qarl Linden: soon - i've got the implementation finished - just a couple more tweaks.
Qarl Linden: a month timeframe, i think.
Qarl Linden: but no promises.

Vincent Nacon: oh about the touch position.... how will you set the unit

Qarl Linden: for llDetectedTouchPos() - the units are region coordinates.
Qarl Linden: for llDetectedTouchUV() - the units are 0 to 1, in UV space.

Vincent Nacon: yeah... for UV one I mean

Qarl Linden: we're also adding a function to return the TEXTURE coordinates...
Qarl Linden: its units will depend on how many repeats are in the texture.

Seifert Surface: i was just reading that bit on the wiki qarl
Seifert Surface: seems useful.

Qarl Linden: yeah - there's been a bit of discussion about it.

Seifert Surface: err, jira i mean

Qarl Linden: clicking on a moving monkey.
Qarl Linden: again - it's a case where it could be implemented by a resident-written function...
Qarl Linden: but it's so useful, we kinda have to put it in.

Seifert Surface: interactions with texture animation?

Qarl Linden: Seifert - yes - that's an important factor. the new call will properly get the texture coordinate from an animating texture.

Seifert Surface: whats the precision like on the returned data from the new functions? presumably its at pixel size?

Qarl Linden: it will be pixel level, yes.
Qarl Linden: we're using a raycast to compute the values.

Seifert Surface: so zoom in closer and you get better numbers

Qarl Linden: yes, because the mouse position is quantized...

Seifert Surface: good thing uv coords are floats

Periapse Linden: We aren't adding functions yet that are Mono only

Simil Miles: Except the touch thing ?

Qarl Linden: no no - the touch stuff is in old LSL.

1 comment: