3d picture control

[DonRoth]

I have developed a reasonable facility for CT type visualization using the 3d surface graph.  If you close the FP just prior to rendering and then re-open when processing is finished, you can get some reasonable rendering times with this strategy.  The data below is a fairly large data cube in which each 2d image frame has been bundled into a cluster. The latter is a memory trick that Damien Gray told me about that I believe removes the contiguous memoryrequirement for the data cube.   I have 100 frames, each 446 x 446.  My computer is a 2 - 3 year old Dell Xeon, ~ 3 GHz, 2 gbyte RAM, winXP.  The rendering takes about 20 seconds.  I was thinking an in-place element structure might help further improve things with regards to unbundling each of the frames but it did not speed things up.

I will at some point try to do something similar with the 3D picture control.  Still only having looked at the 3D picture control examples preliminarily, It would seem I would need the ability to create a set of 2d shapes with each to be textured by one of the frames.  Then each of the shapes would be stacked similarly to how it is done when adding plots to the 3d surface graph.  I have to believe that this control will provide improved rendering performance over the 3d surface graph, but at least we know that the use of the latter can provide a workable solution.

 

Message Edited by DonRoth on 03-26-2009 11:06 AM

[Ben]

Thanks for the update Don!

So the new 3d picture version is faster, cool!

Do you have a revised example that we can point at when asked?

Ben

[DonRoth]

What you see above (the 3d volume) is using the 3d surface graph. (By the way, the 2d sections are just intensity graph containers.)  What I was saying is that I do hope to try this same thing if it is possible with the 3d picture control. I am hoping to look into this soon.  I will keep you statused....Don

[DonRoth]

Although I have not yet had time to look into Volume Rendering applications using the 3d Picture Control, I wanted to know what your experience has been using the surface graph approach with a Windows Vista machine.  Rendering using this approach on a WinXP machine has been great.  However, I have two Windows Vista machines and am getting consistent crashes on both using our present rendering strategy.  I am still on v8.5 and after I release my next version of the NDE Wave & Image Processor (this week), I will go to 8.6 with the intent to go to the next version that is released at NIWeek in September or so.

Sincerely,

Don

[Justin_P]

Hi Don,

I'm interested in the crash you are seeing on your Vista machines.  I'm not sure if it has been brought to R&D's attention yet.  Do you have the crash log for this?

[DonRoth]

Hello Justin -

Just get the message from Vista that the program has stopped working and Windows is checking for a solution.  Then after a few minutes, it says problem has caused the program to stop working.  There is no way to get further info.

Sincerely,

Don

[Justin_P]

Hi Don,

It's too bad that we can't get anymore information on the crash.  If you could tell me what I need to do to reproduce this, I would appreciate it.

Otherwise I'll let you have your thread back

[DonRoth]

Look at Ben's first post on:

http://forums.ni.com/ni/board/message?board.id=170&thread.id=210821&view=by_date_ascending&page=2

I made some further modifications of it related to axis scaling but it is close to this.  So I would reproduce that code and try it on both XP and Vista machines.

Else I can post a simplified version of LLB.

Sincerely,

Don

[Justin_P]

Hi Don,

I think that's enough for me to reproduce this

Thanks

[DonRoth]

I have had some intermittent success on VISTA this weekend with a small number of images (4).  But I did the same thing last weekend and had many crashes.  With 20 - 50 slice images, I seem to get consistent crashes.  Each VISTA machine has 3 gbyte RAM.

Sincerely,

Don