Start a new topic

"Jumping" 3D models on 2D trackable image

"Jumping" 3D models on 2D trackable image


Hi Sergio,

I had a look at the dae file you provided. Which version of the Wikitude3dEncoder do you have and do you run it on a MacOS or on a Windows? The application shouldn't crash but show you a messagebox with the errors.
When I load your dae model then I get a lot of error messages which tell me that the texture image files are missing.

I noticed that your 3d model was created with SketchUp. We recommend using Maya or Blender for creating the dae or fbx files. I removed the textures from your model, quickly added a light and exported it to fbx. You find the file attached. I made it really fast and didnt try to make it beautiful, so the material is just a gray color.

Hope that helps,
Christian

 

Hi Nicolas,

The problem of the image recognition is solved with a copy and paste of the 1_Image_Recognition_On_Target folder that is included in the Examples folder of the Wikitude SDK. Now, i try to play with the 3D_Image_Recognition, but i have problems with the Wikitude 3D Encoder because i can´t load my .dae files to get the .wt3. When i try to load my .dae file the Encoder closes. Here my .dae

https://www.dropbox.com/s/3lqlwt2ixg1ij4o/Peru%20Column%2001.dae

I want to know if you have .dae or .fbx files the that Encoder can work and take them as guide for test with differents 3D Models.

Regards!

Hi Sergio,

 

would you mind posting the target image in question together with the (relevant parts of the) sample code you are using?

That would help us to quickly spot the error.

 

Best regards,

Nicolas

I did that (change the id targets), but in my app doesn't print the X label (AR.Label('X', 1)) over the image that i set as target. Exist another way to save the targets in another format different to .wtc ?

Hi Boston,

depending on the target there will always be slight differences in the performance of image recognition algorithms but we consistently strive to enable the best user experience possible. Therefore I can assure you we are very committed to constantly improve Wikitude's image recognition algorithms. Even though there is no exact date I can share with you at this point in time, we are working on it right now.

In addition the new target management tool with more accurate rating and improved tracking quality is now publicly released and deployed to our servers. Creating an optimized target collection is as simple as opening up your existing project in Wikitude Studio or the Target Management Tool and recreating the wtc file.

 

Best regards,

Nicolas

Thanks Nicolas!

I have received some complaints about the target tracking performance even with these optimized collections. They were in comparison to the same image in Metaio -- with Wikitude they found it harder to recognize the target image especially when not at 90 degees, as well as the 3D model not being as stable on the image. 

Is it reasonable to expect Wikitude to be on par with Metaio? Are their any plans in the near future for further improvments aside from the new target manager?

Hi Boston,

 

no problem bugging me - the new target management tool will be released shortly :)

 

Both of your targets are tracking very well, the REVISED_RV_WALKER_TEST_1C is a bit better (3 stars) than REVISED_RV_WALKER_TEST_2C (2 stars).

I created a target collection containing both targets at http://goo.gl/R34g1t

 

Best regards,

Nicolas

 

Hey Nicolas,

Could you test two more images in the optimized target manager and let me know if they are 3 stars? We're continuing to tweak our target designs for the app.

Link: http://cl.ly/3a2e1o093Z0f

Sooner the new target manager is released the sooner I can start testing myself without bugging you ;)

Thanks,
Boston

The cause of my issue was I did not specify the particular image ID to the Trackable2DObject. I was folowing the example and missed the fact that var pageOne as a variable has to be linked to the image i intend to track.

The solution is to input the name of the image minus the format i.e var pageOne = new AR.Trackable2DOject (tracker "imageone,...") where imageone.jpg is your intended target image.

Hi Obari, ibrahimasank,

 

I am glad that I was able to help you out.

 

To help other developers remedy similar issues, please let me know

 - what was the concrete cause of the issue you experienced and

- which particular steps did resolve it?

 

Thanks for your cooperation.

 

Best regards,

Nicolas

OH i made it work..thanks...

Thanks for your help Nicolas

Hello,

 

To be able to help you out, please provide me the following details:

 

- which platform (Android, iOS) are you using and which version?

- which device are you testing on?

- which developer tools are you using (Eclipse, Xcode, PhoneGap, Titanium etc)?

- which version of SDK are you using?

- which target images are you trying to detect?

 

If possible, please also provide the java script code you are using.

 

Best regards,

Nicolas

 

 

Hello Nicolas , i tried the video sample provided by wikitude , it worked fine but am kinda havin the same problem as Obari, when i change the wtc file, the video isn't triggered, i followed the steps u gave to Obari and still stuck at the same place,, ..thanks for the help..

Hi Nicolas,

I am sorry for insisting so much on something that seems so simple but I have tried the solution you gave but the camera is still not recognizing the target image. In my case, I have try.jpg and best.wtc and the video file.

Is there some place where you can explain this on a more better note for me, I would really appreciate that.

tnx.
Login or Signup to post a comment