The problem is twofold. First, I want the orientation you see in the 3.5 Trident example and went to some trouble to achieve it. That's a standard sort of view in the CAD/CAM world.
Second, I fed the two versions of Away3D exactly the same API calls with the same parameters. They should produce the same orientation. One of them (3.6) is wrong. The Z-axis is reversed (0, 0, -1) instead of (0, 0, 1) because it needs to be inverted in the CAD/CAM world. Best, BW On Thu, Apr 7, 2011 at 6:29 PM, Stephen Hopkins <[email protected]>wrote: > Hm.. Well as someone who also works with Flex (Flash 4, though), I > will say that your code is very flexy :P. Maybe I just haven't need to > write an app that requires overwriting UIComponent or canvas methods > and if I ever did, it was only so that I could attach a sprite > (view3D) to a flex app. In flash 4, it seems sufficient to just create > a SpriteVisualElement and add your main sprite there. > > Anyways, even from the pictures... what is the problem? I see that the > trident or the camera is in a different orientation in both the > pictures. Are they supposed to look exactly the same? Also you have > the z-axis up, make sure you are getting the correct up axis. > > On Apr 7, 2:07 pm, Bob Warfield <[email protected]> wrote: > > Stephen, you can see what you get for a Trident once its compiled here: > > > > http://devluchadore.wordpress.com/2011/04/04/issues-converting-from-a... > > > > That shows a screenshot of both Tridents in my actual CAD/CAM application > so > > you can see what I'm trying to achieve (the 3.5 Trident) versus what I am > > getting (the 3.6 Trident). > > > > So, if I understand this dot product issue, I should not be passing 0's, > but > > rather 0.001's in the upAxis vector? Instead of {0,0,-1} I need > > {0.001,0.001,-1}? Since it was easy, I tried it. In fact I tried making > > any 0 in sight 0.001. Didn't make a lick of difference. > > > > I went back and looked at the eulers/rotation of the camera. It's quite > > different in 3.5 vs 3.6, as it would have to be: > > > > 3.5: > > > > 46.83172 -2.4E-06 -44.9983 > > > > 3.6 > > > > 41.3751 -24.258 -70.003 > > > > Again, this sure looks like a 3.6 bug as I am feeding it exactly the same > > parameters in both versions and getting a different answer. > > > > I don't see how a forum will tell you any more than the mailing list. > Just > > because somebody viewed the post doesn't mean it was somebody on the team > or > > that knew anything. The real issue is it takes a lot of pushing to get > > answers to questions. This is my second thread started on this issue > before > > anyone responded at all. > > > > John, RE pure AS3, I don't know non-Flex, LOL. In any event, I don't > think > > there is much "Flexy" going on in that code other than the transparent > layer > > "hitArea". I debated whether to include it in the sample but did just in > > case it might be the cause. It's there in the first place to deal with > > another Away3D bug I could never get to the bottom of. My mouse events > were > > getting eaten if the mouse point touched any segment in any mesh. > > > > Still need help figuring this out. Sure looks like a bug to me! > > > > Cheers, > > > > BW > > On Thu, Apr 7, 2011 at 12:59 PM, Stephen Hopkins <[email protected] > >wrote: > > > > > > > > > > > > > > > > > Hard to say what you are trying to achieve/encountering since there is > > > no compiled version, but the problem might be with the lookAt > > > function. I and maybe a few others had trouble using the lookAt > > > function. > > > > >http://groups.google.com/group/away3d-dev/browse_thread/thread/97f7f4. > .. > > > > > I just wanted the camera to look directly downwards, or along a > > > global axis, but the camera would would not rotate for some reason > > > unless I skewed the vector away from the the unit axis vectors. (i.e., > > > instead of looking at (0, 0, 0), I made it look at (0, 0, 0.001) ). > > > Never found a fix to it or why it was happening. Such a bug did not > > > occur in Away 3.5 as I had many demos/projects that did such a thing. > > > This is why we need a forum. I would post a new issue, but I just hate > > > the fact that there's no way of telling if they have been reviewed. I > > > already have two outstanding issues that I don't know whether they > > > will be solved or addressed, or just fixed in away3d 4.0 > > > > > On Apr 7, 8:36 am, Bob Warfield <[email protected]> wrote: > > > > I have attached source code for a real simple demo. It does nothing > > > other > > > > than show the Trident. If you compile it with 3.5 libraries, the > Trident > > > is > > > > correct. If you compile with 3.6, the Trident is cockeyed. > > > > > > There are 2 changes required to switch between the libraries: > > > > > > - Comment or uncomment the import of Number3D. > > > > - Comment or uncomment the routines at the top that say "3.5 > Version" or > > > > "3.6 Version". > > > > > > There's not an awful lot that changes between the two versions other > than > > > > whether I user Number3D or Vector3D yet the two Tridents are very > > > > different. That doesn't seem right. > > > > > > Can somebody please help? > > > > > > Best, > > > > > > BW > > > > > > GWAway3D.mxml > > > > 1KViewDownload > > > > > > GWizCamDemo.as > > > > 10KViewDownload >
