Logo by MarkJayBee - Contribute your own Logo!

END OF AN ERA, FRACTALFORUMS.COM IS CONTINUED ON FRACTALFORUMS.ORG

it was a great time but no longer maintainable by c.Kleinhuis contact him for any data retrieval,
thanks and see you perhaps in 10 years again

this forum will stay online for reference
News: Visit us on facebook
 
*
Welcome, Guest. Please login or register. March 22, 2019, 08:06:04 AM


Login with username, password and session length


The All New FractalForums is now in Public Beta Testing! Visit FractalForums.org and check it out!


Pages: [1] 2   Go Down
  Print  
Share this topic on DiggShare this topic on FacebookShare this topic on GoogleShare this topic on RedditShare this topic on StumbleUponShare this topic on Twitter
Author Topic: Voxel stack from x y z axis and combine result  (Read 2712 times)
0 Members and 1 Guest are viewing this topic.
barcud
Navigator
*****
Posts: 69


« on: January 25, 2016, 10:59:45 AM »

A question to the Voxel stackers in this community.
I have been playing with voxel stacks for quite a while now BUT I never really got the results I wanted. Part of the problem is - I guess - the kind of shapes I would like to convert into a model.
Those are mainly my spaceships and objects which have sharp edges and corners e.g.:
http://kronpano.deviantart.com/art/Induction-ring-584032572
http://kronpano.deviantart.com/art/Torus-fleet-in-space-586031155

After voxel stacking I loose all the edges and everything gets very rounded - no matter how fine I stack it.
Now I had an idea but I don't know if that actually helps:
I could create voxel stacks of the a fractal from 3 different angles (like an x y and z voxel stack)
but does anybody know if there is a way to actually increase the accuracy of a resulting object by then using those 3 stacks??
Using those 3 stacks would - in my mind - result in an object that consists of small blocks - not slices any more which might be more accurate?
 
So far I use each stack on its own I end up with 3 models of the same object sliced in 3 different ways but again - does that really help me in any way shape or form??
Is there a way to "combine" the 3 views of an object to increase accuracy?

Had a bit of a look around but could not find a way in Fiji, jWildfire or Monolith. Thought perhaps some medical imaging SW but did not really find anything.
Anybody any idea huh?

Cheers
Logged
thargor6
Fractal Molossus
**
Posts: 789



WWW
« Reply #1 on: January 25, 2016, 02:37:27 PM »

Hi, I'm also very unsatisfied by the results of voxel stack approach, and I'm currently working on a different solution to generate meshes (not by using voxelstacks) from within MB3D, but it is at very early stage, but I could use your models as test-subjects (no idea how it would actually turn out) and post some status here later.

Cheers
Logged
Sockratease
Global Moderator
Fractal Senior
******
Posts: 3181



« Reply #2 on: January 25, 2016, 10:39:46 PM »

What may work best is a low poly base shape with high resolution Normal and Displacement Maps.

That works well for highly detailed 3D Models in other, non-fractal, software.  But the principle is sound and worth investigating.

I think...
Logged

Life is complex - It has real and imaginary components.

The All New Fractal Forums is now in Public Beta Testing! Visit FractalForums.org and check it out!
thargor6
Fractal Molossus
**
Posts: 789



WWW
« Reply #3 on: January 25, 2016, 10:51:45 PM »

What may work best is a low poly base shape with high resolution Normal and Displacement Maps.
This is indeed part of the thing I'm working on, but what puzzles me, is how to get the projection of those maps right, or vice versa, how to create them that they can lead to the right result. But, I think, this can be a good way, otherwise the number of polygons would be too much.
I recently played with pure meshes a lot (having a few billions of polygons, which is pure overkill), and the lack of detail has been always disappointing compared to the original surface. It is a rather cheap technique to create some weird stuff, though cheesy
Logged
slon_ru
Iterator
*
Posts: 167



WWW
« Reply #4 on: January 26, 2016, 09:33:59 PM »

With normal & ao map you can make nice low poy objects.
I use xnormal(www.xnormal.net),
voxel object - ~10 mil poly, render object - 30k poly + 2 textures 2048*2048.

p.s. voxel object - http://www.fractalforums.com/index.php?action=gallery;sa=view;id=17185


* 2normal_ok_00.jpg (185.11 KB, 1280x720 - viewed 230 times.)
« Last Edit: January 26, 2016, 11:01:11 PM by slon_ru » Logged

barcud
Navigator
*****
Posts: 69


« Reply #5 on: January 27, 2016, 05:04:03 PM »

slon_ru - those objects you posted the links to look very detailed to me.
At least I have never achieved such detail.

How many Z slices did you use?
And possibly  - how did you assemble/ process the slices?

They don't seem to have any steps and still a lot of detail and sharp edges.

Care to share a bit more how you get those results??
Logged
barcud
Navigator
*****
Posts: 69


« Reply #6 on: January 28, 2016, 09:34:35 PM »

Here is my latest try in voxeling
Used 1000 slices of 1500x1500 the into Fiji - reduced size on import and used the smooth filter there, loaded it into Meshlab , ran the de-duplicate and Taubin smooth.
Ended up with 10 million vertices and 20 million faces and the file is 2GB big.

No idea on how to automatically smooth it without loosing the shapes
Need to look into xnormal.net - see if it helps.

Left a basic Mandelbulb3D render - nice and clean and on the right the not so clean object.
Well hey, it was worth another try.


* MeshLab_Mandelbulb3D.jpg (173.23 KB, 1224x600 - viewed 236 times.)
Logged
bib
Global Moderator
Fractal Senior
******
Posts: 2070


At the borders...


100008697663777 @bib993
WWW
« Reply #7 on: February 02, 2016, 05:09:27 PM »

Try to use Quadric Edge Collapse Decimation in Meshlab. You can do it step by step (reduce number of faces by 10 to 20% every time) and run a laplacian smoothing between every pass.
Experiment with all the parameters of the decimation filter, especially planar simplification if you have smooth surface. You will save a huge number of triangles.
Logged

Between order and disorder reigns a delicious moment. (Paul ValÚry)
barcud
Navigator
*****
Posts: 69


« Reply #8 on: February 03, 2016, 10:21:25 AM »

I did that with quite a few collapse and smooth filters but as those filters say - they collapse and smooth which means all my edges disappear and get round.
That is why I am so impressed by the models posted by
- they still have sharp edges and corners.
I always loose mine using those filters - perhaps there is a sort of equivalent to a high pass filter in photography which can find corners and enhance/preserve them?

No worries - I will keep on playing for a bit.
Logged
thargor6
Fractal Molossus
**
Posts: 789



WWW
« Reply #9 on: February 07, 2016, 01:07:57 AM »

First approach to use one of your models in the upcoming new mesh generator:

Mesh-generation time took about 20 minutes, volumetric resolution 768 x 768 x 768 -> 6 million faces

Loaded into MeshLab, used one pass of Taubin smooth, (did not change anything else due the lack of time, downsampling would probably work well on this model), looks not that bad:


Some test-render:
Logged
barcud
Navigator
*****
Posts: 69


« Reply #10 on: February 07, 2016, 03:34:16 PM »

Looks quite nice!!! Edges still nice and - well - edgy - not too rounded.
I would say if that is the first approach - that looks quite promising!!

Very nice in terms of development!! Cudos to the almighty developer!!
Logged
slon_ru
Iterator
*
Posts: 167



WWW
« Reply #11 on: February 08, 2016, 09:44:00 AM »

slon_ru - those objects you posted the links to look very detailed to me.
At least I have never achieved such detail.

How many Z slices did you use?
And possibly  - how did you assemble/ process the slices?

They don't seem to have any steps and still a lot of detail and sharp edges.

Care to share a bit more how you get those results??
Try to use zbrush.
I make 2048*2048*2048 m3d voxel stack -> ~1024*1024*1024 unified skin in zbrush -> zbrush decimation master -> xnormal.
« Last Edit: February 08, 2016, 09:51:16 AM by slon_ru » Logged

barcud
Navigator
*****
Posts: 69


« Reply #12 on: March 12, 2016, 12:23:24 PM »

Oh - just saw in an announcement over on the facebook group that the new version with 3D mesh generation is available.

I will certainly need to play with that and - a massive big THANK YOU for all the effort goes out to thargor6
Logged
slon_ru
Iterator
*
Posts: 167



WWW
« Reply #13 on: March 12, 2016, 08:16:28 PM »

"...announcement over on the facebook group that the new version with 3D mesh generation is available."

Can you please send me the link!?!
Logged

thargor6
Fractal Molossus
**
Posts: 789



WWW
« Reply #14 on: March 12, 2016, 09:17:41 PM »

Oh - just saw in an announcement over on the facebook group that the new version with 3D mesh generation is available.

I will certainly need to play with that and - a massive big THANK YOU for all the effort goes out to thargor6
Probably this version will not totally solve the problem with the sharp edges. But, it is a start, more improvements so come over time :-)
Logged
Pages: [1] 2   Go Down
  Print  
 
Jump to:  

Related Topics
Subject Started by Replies Views Last post
Voxel image stack and then what? Tutorials « 1 2 ... 9 10 » blob 143 40111 Last post January 03, 2017, 08:33:32 PM
by Mrz00m
More Voxel Stack mesh examples Movies Showcase (Rate My Movie) Don Whitaker 6 1054 Last post May 15, 2012, 10:58:32 AM
by arias
Slices 2 - Mandelbulb 3D voxel stack animation Movies Showcase (Rate My Movie) Don Whitaker 1 1504 Last post September 06, 2012, 08:43:50 AM
by art.cjbarnaby.com
mandelbulb 3d voxel stack question 3D Fractal Generation morbazan 1 1077 Last post September 07, 2016, 10:39:56 PM
by thargor6
voxel slice result completely different from navigator Mandelbulb 3d Mrz00m 0 365 Last post January 15, 2017, 02:22:07 AM
by Mrz00m

Powered by MySQL Powered by PHP Powered by SMF 1.1.21 | SMF © 2015, Simple Machines

Valid XHTML 1.0! Valid CSS! Dilber MC Theme by HarzeM
Page created in 0.198 seconds with 28 queries. (Pretty URLs adds 0.011s, 2q)