cap10subtext

You need a mouse with a scroll wheel to test this. The object is to click from a location near to your near clipping plane then extend a line using the scroll key to just about the far clipping plane. Anything that intersects the line drawn between the two should report a collision.

1) the screentoworld message is killing me. I give it screen coords and a value between 0. and 1. (0. being the near, 1. being the far). It doesn't give me anything useful. Basically it reports the z position to 10 then jumps to -88. ???? I realize the jump is probably garbage from when the z value is 1 and therefore exceeds the far clip, but why isn't it reporting from 10 to 100?

2) Does jit.phys.world not with jit.gl.sketch?

The Phys body for jit.gl.sketch which draws the line between them just sits in the middle of the screen, I can't seem to get it to attach to the jit.gl.sketch object. I realize it probably doesn't handle resizing, but shouldn't it still report once the linesegment is drawn, if even only for the one frame? Or does it just not work at all?

3) tips on how to then collect the appropriate collisions (ignoring when the near and far crosshairs are too close together for example) would be appreciated.

You need a mouse with a scroll wheel to test this.  The object is to click from a location near to your near clipping plane then extend a line using the scroll key to just about the far clipping plane.  Anything that intersects the line drawn between the two should report a collision.

1) the screentoworld message is killing me.  I give it screen coords and a value between 0. and 1. (0. being the near, 1. being the far).  It doesn't give me anything useful.  Basically it reports the z position to 10 then jumps to -88.  ????  I realize the jump is probably garbage from when the z value is 1 and therefore exceeds the far clip, but why isn't it reporting from 10 to 100?

2) Does jit.phys.world not with jit.gl.sketch?
The Phys body for jit.gl.sketch which draws the line between them just sits in the middle of the screen, I can't seem to get it to attach to the jit.gl.sketch object.  I realize it probably doesn't handle resizing, but shouldn't it still report once the linesegment is drawn, if even only for the one frame?  Or does it just not work at all?

jit-phys-body-and-world-issues

i'm a little unclear on what you are trying to achieve and what's not working for you.

1 - if your camera is at zpos 10, and your far_clip is 100 (the default) then screentoworld with a z val of 0 will return a world position with a z val around 10 (cam pos z), and screentoworld with a z val of 1 will return a world position with a z val around 90 (100 units from the camera's near plane). so the values you are getting sound correct to me.

2 - seems to work fine with gl.sketch. check the patch below.

3 - again, not entirely sure what you want and what you are not getting.

could you break your patch down into smaller pieces exemplifying the specific problems?

i've pasted another patch showing how to look for specific collisions between named objects.

3 - again, not entirely sure what you want and what you are not getting.
could you break your patch down into smaller pieces exemplifying the specific problems? 

Hi Rob, Thanks for the reply. Looking into your response.

With point 1, I see what's happening now, but I'm trying to figure out the math on the distance, basically going from 0-9 went from 9-1 but then 0. to -99. was when screentoworld was between 0.992 to 0.999. So it was jumping drastically. So I'm scaling but I think a logarithmic (?) scaling would work better.

Regarding point 2) jit.phys.body/jit.gl.sketch: Okay so I think the problem is not that it doesn't work but it certainly doesn't work as I had expected. I think I understand why, but it leaves me stuck. Perhaps there's a better way to go about it?

Look at the following patch and I've provided 2 example under which collision "fails" to accurately detect what's in the space with jit.gl.sketch, because the rigid body is calculated around a centre point/radius defined by jit.phys.body but what's drawn by jit.gl.sketch fails to fill or does not exist within that space.

So if I draw a line using linesegment, is there any way to use jit.phys.body or another object to detect if that line passes through another jit.phys.body?

Is this where concave/convexhull or compound needs to be applied?

Thanks so much for your patience, just trying to wrap my head around these new objects. :)

Hi Rob, Thanks for the reply.  Looking into your response.  

With point 1, I see what's happening now, but I'm trying to figure out the math on the distance, basically going from 0-9 went from 9-1 but then 0. to -99. was when screentoworld was between 0.992 to 0.999.  So it was jumping drastically.  So I'm scaling but I think a logarithmic (?) scaling would work better.  

Regarding point 2) jit.phys.body/jit.gl.sketch:  Okay so I think the problem is not that it doesn't work but it certainly doesn't work as I had expected.  I think I understand why, but it leaves me stuck.  Perhaps there's a better way to go about it?

Thanks so much for your patience, just trying to wrap my head around these new objects. :)


you have to tell the phys.body what kind of geometry to use for its shape, it doesn't automatically infer this from the attached gl shape.

it does however automatically send proper scale values to attached shapes (if the @send_scale attribute is on, which it is by default), so you want to scale the phys.body instead of the gl object.

i'm not sure if you want a line shape that interacts dynamically with the scene, or simply the ability to do collision ray testing.

if the former, you'll need to create a cube shape and scale it appropriately, to approximate a line segment. the patch below shows how.

if it's the latter, you will have to wait for 6.0.2 update, and utilize the new ray-testing features of phys.world.

you can send collision queries with either screen coords, or world-space line segment end points.

i'm not sure if you want a line shape that interacts dynamically with the scene, or simply the ability to do collision ray testing.
if the former, you'll need to create a cube shape and scale it appropriately, to approximate a line segment. the patch below shows how.

if it's the latter, you will have to wait for 6.0.2 update, and utilize the new ray-testing features of phys.world.
you can send collision queries with either screen coords, or world-space line segment end points.

Hmmmm... Probably need ray testing after all. Will probably wait for update. Thanks for the feedback! Really appreciate it!

Hmmmm... Probably need ray testing after all.  Will probably wait for update.  Thanks for the feedback!  Really appreciate it!


Hi there, sorry to bump, but I've been working on other projects and have just started to re-visit this one.

I've got the pickray feature working but does the pickray feature only output the first object? Is it possible to output all of the objects the ray intersects with in order (front to back or back to front?) I think this would be the missing feature I need to finally put this project to bed.

(below is just the sort of patch I'll be working with, I'd like to know that the ray is hitting the red ball behind the grey ball but for any number of objects)

I've got the pickray feature working but does the pickray feature only output the first object?  Is it possible to output all of the objects the ray intersects with in order (front to back or back to front?)  I think this would be the missing feature I need to finally put this project to bed.

(below is just the sort of patch I'll be working with, I'd like to know that the ray is hitting the red ball behind the grey ball but for any number of objects)


thanks for the request and the patch. it is registered and will be in a future update.

unfortunately 6.0.5 was frozen before this feature was added.

jit.phys.body and world issues