Comments on: Trying out flash player 10 3d features http://www.sakri.net/blog/2008/10/16/trying-out-flash-player-10-3d-features/ I Flash und I Flex Jaaaa Fri, 12 Mar 2010 15:21:46 +0000 http://wordpress.org/?v=2.6.2 By: AP http://www.sakri.net/blog/2008/10/16/trying-out-flash-player-10-3d-features/#comment-786 AP Thu, 01 Jan 2009 23:38:20 +0000 http://www.sakri.net/blog/?p=98#comment-786 On the Performance3dTest1 it says if you run over 5000+ objects your browser will crash...I ran 66 x 66 x 66 which means i got 287,496 objects...anything above that made my browser crash...i was on Firefox 3.0.5 on Leopard 10.5.6...Flash 10...2 GB ram...2.5 GHz Intel Core 2 Duo...GeForce 8600M GT 512 MB vram On the Performance3dTest1 it says if you run over 5000+ objects your browser will crash…I ran 66 x 66 x 66 which means i got 287,496 objects…anything above that made my browser crash…i was on Firefox 3.0.5 on Leopard 10.5.6…Flash 10…2 GB ram…2.5 GHz Intel Core 2 Duo…GeForce 8600M GT 512 MB vram

]]> By: The Flash Art of War » Blog Archive » Two Sided Plane - FP 10 http://www.sakri.net/blog/2008/10/16/trying-out-flash-player-10-3d-features/#comment-681 The Flash Art of War » Blog Archive » Two Sided Plane - FP 10 Sun, 16 Nov 2008 16:55:12 +0000 http://www.sakri.net/blog/?p=98#comment-681 [...] 10 there is no depth sorting of 3d objects. You are on your own to create that logic. Here is a post that does a better job of demonstrating what I am talking about. So with the limitations in mind [...] [...] 10 there is no depth sorting of 3d objects. You are on your own to create that logic. Here is a post that does a better job of demonstrating what I am talking about. So with the limitations in mind [...]

]]> By: sakri http://www.sakri.net/blog/2008/10/16/trying-out-flash-player-10-3d-features/#comment-593 sakri Fri, 17 Oct 2008 16:10:34 +0000 http://www.sakri.net/blog/?p=98#comment-593 Thanks for commenting jabek. I'll admit this post was a bit of a rush job before going travelling. The idea was to test the 'out of the box' features, just to see what the API looks like and how things behave by default. Nothing fancy here. You're right about the 'ton', I guess I was a bit giddy at the ease of use implemented by the adobe guys. Damn it's hard to write this shit on a gsm ;) Thanks for commenting jabek. I’ll admit this post was a bit of a rush job before going travelling. The idea was to test the ‘out of the box’ features, just to see what the API looks like and how things behave by default. Nothing fancy here. You’re right about the ‘ton’, I guess I was a bit giddy at the ease of use implemented by the adobe guys. Damn it’s hard to write this shit on a gsm ;)

]]> By: Jabek http://www.sakri.net/blog/2008/10/16/trying-out-flash-player-10-3d-features/#comment-591 Jabek Fri, 17 Oct 2008 11:33:15 +0000 http://www.sakri.net/blog/?p=98#comment-591 Where do I start... With regards to tests with multiple faces (Performance3DTest1 and CubeTest): First you need to add a sorting mechanic in order to remove the 'optical illusion' as you call it. Flash doesn't sort the scene for you. The order that the objects are drawn in the scene is based on the initial order that you add them. Any time you change a 3D position or rotation of an object (by rotation, I mean with respect to the scene... not the individual objects) you have to resort the scene. That means you need to change the index (or depth as it used to be called) of each object since one may have moved behind another (in this case I mean the Z value). Here's the catch... although it's true that Z is positive into the screen and negative away from the screen... the index work opposite of that. Smaller index values are into the screen and higher index values are away from the screen. I'm not sure if and how you're keeping track of the objects, but say you use an array, you can write a custom sort function and call yourArray.sort(customSortFunction) to do that for you. With regards to the CubeTest: Apply what the previous comment mentions and only half of the cube sides need to be shown on the scene. The other half that doesn't need to be shown, (assuming you add the sorting mechanic) you won't be able to see anyways since the normal will be facing away from the screen so there's no point of showing them anyways. Are you using the hardware acceleration feature? Just curious because the frame rate is very jumpy. Your test application doesn't show an averaged frame rate so its hard to see... but watching it in a 10, 10, 10 (1000 shapes) test on the Performance3DTest1 and keeping a rotaiton active, I can see it ranging from sub-20 to a bit over 30 or so (and occasionally going over 40, and I think I saw something in the 50's a few times too when not rotating). Also... 10 x 10 x 10 is 1000 shapes (not sure if you're just using square shapes or 2 triangles like normal 3D rendering applications do) but 1000 shapes isn't even close to 'a ton of shapes' in a 3D rendering environment. I tried 100 x 100 x 100... which is 1,000,000 (a million) shapes... but that crashed my brower completely. An 'ok' first look at 3D in flash player 10... I say ok... because I've done all of this in flash player 7 by writing a software renderer all in AS2, including the perspective texture in your first test. Ya as in seven. I was expecting a lot more from this flash player 10 and I'm overall dissapointed. Just shows how far flash still needs to go to be a 3D renderer. Either way... nice tests, make some stress tests that won't crash the browser... I'd curious to see what the new flash player can handle. Sweet testing environment though. Where do I start…
With regards to tests with multiple faces (Performance3DTest1 and CubeTest):
First you need to add a sorting mechanic in order to remove the ‘optical illusion’ as you call it. Flash doesn’t sort the scene for you. The order that the objects are drawn in the scene is based on the initial order that you add them. Any time you change a 3D position or rotation of an object (by rotation, I mean with respect to the scene… not the individual objects) you have to resort the scene. That means you need to change the index (or depth as it used to be called) of each object since one may have moved behind another (in this case I mean the Z value). Here’s the catch… although it’s true that Z is positive into the screen and negative away from the screen… the index work opposite of that. Smaller index values are into the screen and higher index values are away from the screen. I’m not sure if and how you’re keeping track of the objects, but say you use an array, you can write a custom sort function and call yourArray.sort(customSortFunction) to do that for you.

With regards to the CubeTest:
Apply what the previous comment mentions and only half of the cube sides need to be shown on the scene. The other half that doesn’t need to be shown, (assuming you add the sorting mechanic) you won’t be able to see anyways since the normal will be facing away from the screen so there’s no point of showing them anyways.

Are you using the hardware acceleration feature? Just curious because the frame rate is very jumpy. Your test application doesn’t show an averaged frame rate so its hard to see… but watching it in a 10, 10, 10 (1000 shapes) test on the Performance3DTest1 and keeping a rotaiton active, I can see it ranging from sub-20 to a bit over 30 or so (and occasionally going over 40, and I think I saw something in the 50’s a few times too when not rotating).

Also… 10 x 10 x 10 is 1000 shapes (not sure if you’re just using square shapes or 2 triangles like normal 3D rendering applications do) but 1000 shapes isn’t even close to ‘a ton of shapes’ in a 3D rendering environment. I tried 100 x 100 x 100… which is 1,000,000 (a million) shapes… but that crashed my brower completely.

An ‘ok’ first look at 3D in flash player 10… I say ok… because I’ve done all of this in flash player 7 by writing a software renderer all in AS2, including the perspective texture in your first test. Ya as in seven. I was expecting a lot more from this flash player 10 and I’m overall dissapointed. Just shows how far flash still needs to go to be a 3D renderer.

Either way… nice tests, make some stress tests that won’t crash the browser… I’d curious to see what the new flash player can handle. Sweet testing environment though.

]]> By: David http://www.sakri.net/blog/2008/10/16/trying-out-flash-player-10-3d-features/#comment-590 David Thu, 16 Oct 2008 22:46:47 +0000 http://www.sakri.net/blog/?p=98#comment-590 Pretty cool insightful tests :) About that cube, or any solid convex geometry for that matter, it can be made to render correctly pretty easily with FP10 tho. Since you have all 3 rotation angles for any surface, you can calculate the normals and set the surface to invisible if it's pointing away from the eye. Pretty cool insightful tests :) About that cube, or any solid convex geometry for that matter, it can be made to render correctly pretty easily with FP10 tho. Since you have all 3 rotation angles for any surface, you can calculate the normals and set the surface to invisible if it’s pointing away from the eye.

]]>