PDA

View Full Version : Rendering images from 3D cube with different morphological



saded
09-19-2013, 03:05 PM
Hi

I Looking for help in obtaining images,
So I have a cube contains geometric shapes
The videos sample also describes

https://www.youtube.com/watch?v=IradEi1ERpA
https://www.youtube.com/watch?v=-r_Ndy5zqoE

is it possible to do this with Lightwave 3D :lwicon:

nickdigital
09-19-2013, 03:56 PM
Maybe something like this?

Danner
09-19-2013, 05:47 PM
or this..
http://www.volumedic.com/

saded
09-20-2013, 08:45 AM
117208117209

in the file two-case.rar, I use two case for modeling
Case 1: I draw a cube and Balls;
Case 2: I draw a set of parallèes planes (volume of a cube) in a layer and the second layer i created a number of balls.
My idea is to generate images of the dimensions plan and containing circles various diameters in each section, as shown in the example file images.rar

For the first case, I think we need a boolean to these images.
and for the two cases, I try to make the intersection between planes and balls, for plans that contients circles.

Note: there are plans that have no contain circle , ex: the first plane.

Can you help me get these images with both.

for nickdigital; how do you do this?
for Dannerm; Volumic is plugin for LW ?

nickdigital
09-20-2013, 10:28 AM
I believe what you want is an animated boolean. See if these links help.

http://www.youtube.com/watch?v=azM5_U4FzLY

Clip Map Cut Away Tutorial:
http://www.md-arts.com/tutorials/lw_tutorials/tutorials.html

http://forums.newtek.com/showthread.php?119003-DP-Boolean-Question

Volumedic is a third party plugin for LW.

saded
09-20-2013, 11:55 AM
I only use the modeler, I have not mastered the Layer for this moment
Can you help me for the two cases.

nickdigital
09-20-2013, 01:05 PM
You're not going to be able to complete an animation without going through Layout.

What do you want? To just slice the spheres where your planes are? You could do that with the boolean tool.

saded
09-20-2013, 01:29 PM
When you turn the face of the cube to take pictures
For example, if I want to get 100 images as file: images.rar
Keep the camera in front of the cube
for example, if used in the cutawaye Mètode sauvgardement for the work of these 100 photos
Each image contains circles of different sizes

I hope to be may benefit , i am not good at English well

nickdigital
09-20-2013, 01:37 PM
What are the 100 images? That plane animating through the cube?

saded
09-20-2013, 01:43 PM
it's like the tomograph, it scan the 3D element (cube) and saving images in jpg or bmp

nickdigital
09-20-2013, 01:48 PM
I don't think there's a native way to do what you want in LW. What you're trying to recreate is something you see in a medical scanner. Volumedic will probably be the closest thing to what you want if you're using actual data.

If you're just trying to fake the effect then the links that I posted earlier should point you in the right direction.

saded
09-20-2013, 02:04 PM
I think with lightwave is simple, use a boolean that each time moving the camera, takes a picture

saded
09-20-2013, 02:13 PM
117220 this sample is generate with 3D cube contenains balls

nickdigital
09-20-2013, 02:15 PM
117220 this sample is generate with 3D cube contenains balls

So you figured it out then?

saded
09-20-2013, 02:24 PM
yes

dwburman
09-22-2013, 04:25 PM
Here's the volumedic link: http://www.volumedic.com/
For most people here the Volumedic CE version is all that is necessary.

I don't know what tools it has for taking a 3D object and slicing it up.

You might also want to check out 123D Make. http://www.123dapp.com/make


Here's a rendering test I did. In order for this to work with a set of cross section images (image slices) like we're talking about in this thread would require a separate image per polygon in the polygon stack. That would be tedious to set up by hand. You CAN set it up pretty easily with a couple of DP Kit nodes: Apply your image sequence in the DP Kit>Shaders>Image Map node and plug the index out from the DP Kit>displacement>Part Move node into the Frame input on the Image Map node. As long as your polygon order (and therefore part number) matches the frame number, you should get the results you're looking for except that it takes a long time to render. I rendered one frame with no Antialiasing at 320x180 and it took around 7 minutes. I don't want to think about how long it'd take for 1080p with adequate antialiasing and soft shadows.


http://youtu.be/QZJ3L5lZFWM

saded
09-23-2013, 02:43 PM
This is a good idea
but for ;em i not use DP Kit :cool::cool: and sequence image with LW

if i have the dimensions of the cube: a*a*a, and i want to save image sequences from the front dimensional a*a
Taking an example of a record of 10 min = 10 * 60 = 600 sec
if you want to take 100 images seccessives, so you have 600/100 = 6 sec for saving one image with size a*a

are you see this with my exemple : cube.rar

I will try as well as I'm experimenting with this work :lwicon:

dwburman
09-23-2013, 05:01 PM
For example, if your cube is 1m * 1m * 1m and you want 100 slices.

Camera Setup

I would suggest using the Orthographic camera type.

Set the camera type to Orthgraphic.
Set the Size to 1 m.
Set the camera resolution so the Width and the Height are the same (for example 1000 pixels)

The orthographic camera does not show perspective.

Cutter setup
Since you want 100 slices, we will move the cutter and add a key at frame 100.

How much do we want to move the cutter?
That can be expressed as box depth / number of slices. In our example our box is 1m deep, so 1m/100 = 10mm. Our cutter needs to move 10mm per frame.
Of course, all you really need to do is add 1m (or -1m) to the position of the cutter on the 100th key frame.

You may need to set the last key frame to 99 and set the render range from 0 to 99 to get your 100 slices.

Time

The render time will be however long it takes to render each slice * the number of slices (frames)

It looks like you figured that out.


What I was showing in my example was not the cutting of the slices, but rather the displaying of the slices.

saded
09-24-2013, 12:58 PM
I put the camera against the cube

As the picture
http://im36.gulfup.com/10seW.jpg (http://www.gulfup.com/?X0ErAR)

How can I add the process Cutter

dwburman
09-24-2013, 10:57 PM
What I call the cutter is just a clip map on the objects that is connected to a null as the texture's reference object (or selected as the object in the case of a "distance to object" gradient)

Look for the "Clip Map Cut Away Tutorial" at the following link:
http://www.md-arts.com/tutorials/lw_tutorials/tutorials.html

It is the original technique that I based my video tutorial on. The old tutorial might have an easier to understand method for the clip map, plus it is written so you may have an easier time translating it.

saded
09-26-2013, 01:26 PM
I will try this workout

there is a method that I think it's easier
You can try this method :
fixe the distance betwwen camera and the cube
for example, the cube dimensions: 1 * 1 * 1
If we are saving 100 sequences images
If we divided 1m/100 , we find 10 mm (milimetre)
The way to move the cube
you move the cub to 10 mm we cut off the part that we get it
Any cube that its dimensions are 1 m * 1 m * 0.01 m ( this part we hide)
And save this image (image1)
Thus, whenever we move to 10 mm we cut part and save second image

and so on until the last picture (image100)

http://im31.gulfup.com/5PLkw.jpg (http://www.gulfup.com/?UAO9xR)

XswampyX
09-26-2013, 04:41 PM
You could try it with a load of instanced planes, then animate it with a null?

http://i465.photobucket.com/albums/rr16/xXswampyXx/Mri_Test_Ins_zpsc93f05b0.jpg (http://s465.photobucket.com/user/xXswampyXx/media/Mri_Test_Ins_zpsc93f05b0.jpg.html)

Each instance has the transparency set so you run into the 64 raytrace limit.

tbagger
09-26-2013, 05:40 PM
Here's the volumedic link: http://www.volumedic.com/
For most people here the Volumedic CE version is all that is necessary.

I don't know what tools it has for taking a 3D object and slicing it up.

You might also want to check out 123D Make. http://www.123dapp.com/make


Here's a rendering test I did. In order for this to work with a set of cross section images (image slices) like we're talking about in this thread would require a separate image per polygon in the polygon stack. That would be tedious to set up by hand. You CAN set it up pretty easily with a couple of DP Kit nodes: Apply your image sequence in the DP Kit>Shaders>Image Map node and plug the index out from the DP Kit>displacement>Part Move node into the Frame input on the Image Map node. As long as your polygon order (and therefore part number) matches the frame number, you should get the results you're looking for except that it takes a long time to render. I rendered one frame with no Antialiasing at 320x180 and it took around 7 minutes. I don't want to think about how long it'd take for 1080p with adequate antialiasing and soft shadows.


http://youtu.be/QZJ3L5lZFWM

Turn your threads up to 64 it should render much much faster... Even if you dont have 64 threads it still helps the dp image node to have many virtual threads.

dwburman
09-26-2013, 11:58 PM
Thanks for the tip. :)

saded
09-27-2013, 10:09 AM
I'm no expert, but your idea is great
are you explain this with video in my example
the picture Mri_Test_Ins_zpsc93f05b0.jpg is small

XswampyX
09-27-2013, 01:00 PM
I'm no expert, but your idea is great
are you explain this with video in my example
the picture Mri_Test_Ins_zpsc93f05b0.jpg is small

It's not quite working..... :(


http://www.youtube.com/watch?v=nziVPLZcKCc

As I said, it's hitting the 64 raytrace limit.

I will try using clip maps, but the UV support is a little flakey. Leave it with me.

Cheers!

Danner
09-27-2013, 05:01 PM
Instead of making them progressively transparent.. there is surely a way of moving them out of the way quickly, or killing the instance, to avoid having so many layers visible. For something like this, it would work, but not for every type of scan I suppose.

saded
09-28-2013, 08:11 AM
This is the best idea of ​​the video XswampyX

saded
09-29-2013, 06:13 AM
The best soft 3D but I can not find an expert in the design:):help:

Elmar Moelzer
12-06-2013, 09:01 PM
Sorry for the late reply. VoluMedic CE would be ideal for this sort of task. It also comes with tools for converting geometry to volumes and back again.