Dear DAVID Community,

As you are probably aware, the entire DAVID company, team and product range has been integrated into HP Inc.
This forum is now read-only. Relevant content has been migrated to HP and merged into HP's Support Forums on November 1, 2016.

To start new discussions for 3D Scanning, please register and post your new topic at the HP Support Forums

Disappointing Results

The place for all topics related to the 'Shapefusion' tool of DAVID.

Disappointing Results

Postby Christian_Mendez » Wed Jan 11, 2012 12:18 am

Well, as the thread title says - I´m disappointed with the results i´m getting with Shape Fusion.
I´ve just read through some threads and it seems like I´m not the only one around having this.

To make a long story short;

The time it took me to get some decent quality in my scans that I´m actually happy with can be counted
in months by now. All the researching, experimenting and so forth (you know what I´m talking about)
is now beeing questioned by unsatisfying results with Shape Fusion. I really invest time to get good detailed scans
and please don´t get me wrong - but as soon as I fuse all the scans together I lose a lot of detail. I even maxed out the
resolution up to 5000 but still details are lost. (btw. 5000 what?...pixels? vertices? does anyone know the unit?)
I´ve expected a lot more to be honest.

All the invested time and work seem to be useless. In my opinion it´s pointless if details are lost when it matters most...namely
when all scans are fused together.

So where should I start troubleshooting? Is this even something I can solve? Is this a "fact" that I must accept until the enterprise edition
comes out?
www.greyscale3di.com "if it ain´t movin´, we can scan it!"
Christian_Mendez
 
Posts: 188
Joined: Thu Nov 17, 2011 12:45 am
Location: Vienna, Austria

Re: Disappointing Results

Postby vchmiele » Wed Jan 11, 2012 1:56 am

Unfortunately this is caused when meshes overlap and the Poisson filter needs average the polygons.

A few things to consider:

-Try eliminating redundant scans. Ideally the meshes only intersect slightly.
-Once aligned (but not merged). I will export all of my meshes and will remove any extra information (I use Zbrush for this, since it's easy to edit meshes).
-Also, meshlab has some remeshing tools (similar to David). However, there are some features that can help clean your mesh up (removing long faces, isolated faces, etc.). Plus the Poisson filter has more variables (a little buggy though).

Also, from what I can tell there is some very nice non-rigid registration software that will be available relatively soon.

Hope this helps,

Vince

PS- The SL scan that you did of your hand looked very clean. I am still experiencing some distortion in my scans (just started with SL). Any tips for achieving clean scans?
vchmiele
 
Posts: 61
Joined: Mon Feb 21, 2011 11:09 am

Postby hal » Wed Jan 11, 2012 3:28 pm

Hello Christian_Mendez,

May I ask you to post here a comparison between the single scans and the fusion result? Maybe just a zoom over a choosen (the same for both) area will be enough to understand how much detail you lost.
I strange the this happen. I've just scanned, last night, something with small details and fused 17 scans at 1200. All the details are there. Obviously take in mind that the fusion operation does an optimization, so if you have very sharped small and fine details (like small pinched points) in the scans, probably they became more rounded in the fusion.

If you are satisfied only with the scans and don't like the fusion, you can:
- align the scans;
- erase manually the overlapped areas;
- combine all the scans togheter;
- save it as single file.

If you have holes, you can import the "combined" mesh into Meshlab (or other software) and fix them with "Close holes" tool.

Regards,
Mattia
PC: DELL Alienware 17, Intel I7-4710MQ, 32 Gb Ram, AMD Radeon R9 M290X 4GB.
SO: Win 8.1 64Bit.
CAMs: USB 3.0 CMOS B/N (1280x960, 60 fps).
LENS: 12 mm.
PROJECTOR: DAVID SLS1 ACER K11.
User avatar
hal
Moderator
 
Posts: 1527
Joined: Thu Mar 01, 2007 4:08 pm
Location: Cavaria con Premezzo (VA), Italy.

Re: Disappointing Results

Postby Christian_Mendez » Wed Jan 11, 2012 8:42 pm

Thank you both for responding guys...

I gone through the process rescanning the object again. The level of detail can be better but this should be enough for trouble shooting.
Here are some step by step screenshots as I progressed scanning. You can see that the level of detail of both scans are good. Both scans
are also aligned very well...so far so good. Details are significantly smoothed out as soon as the scans are fused. I raised the resolution to
3000 and yet the details are smoothed out to a level that I can´t be happy with...

Yes, I left a big overlap at the mid-section of the alosaurus body on purpose because I wanted to see if there was a difference in areas where
2 meshes are fused together and in areas where there are NO overlaps. Notice the loss of detail on the head of the alosaurus.
I can imagine that Shape Fusion calculates an average that eventually leads to "over-smoothing", however that does not explain
the loss of detail on the alosaurus head.

Anyway, back to your suggestions. To preserve detail, one should align with Shape Fusion and save the aligned mesh without fusing.
Next, edit the mesh in Meshlab or similar programs...

As always, your help is much appreciated - thanks again guys...


The object to be scanned - a pretty neat alosaurus...
Image

Scan result...
Image

Cleaned up everything...
Image

Scanned the next segment...
Image

Aligned both scans...
Image

Close-Up of the green and blue scans at the overlapping area
Image

Image


Before and after fusion results...
Image

Image


P.S.;

Vince, check your Calibration Corner if it has a perfect 90° angle and if the boards of your Calibration Corner are warped.
You can recognize a warped board by looking along all 4 edges of the board. You should be seeing a perfect straight edge.
If you have a "perfect" Cal-Corner then start worrying about your camera lens. What camera are you using?
www.greyscale3di.com "if it ain´t movin´, we can scan it!"
Christian_Mendez
 
Posts: 188
Joined: Thu Nov 17, 2011 12:45 am
Location: Vienna, Austria

Re: Disappointing Results

Postby Christian_Mendez » Wed Jan 11, 2012 10:49 pm

Here´s an update...

After installing Meshlab just a few minutes ago I fused the two scans just as suggested by Mattia and Vince. The difference
is simply stunning ! The loss of detail in Shape Fusion is so dramatic...see for yourself. The green mesh was fused in Meshlab
and the blue mesh was fused in Shape Fusion (resolution 3000)...

Though I´m relatively glad to know that Meshlab can handle detail preservation quite well (as far as I can judge after playing with it for about 20 min.) I still remain unhappy with Shape Fusion´s results in terms of detail preservation. Meshlab is another step in my workflow that I´d rather not take if not necessary as it takes up time to learn...Shape Fusion is sufficient if the surface of the object is relatively simple or smooth. At the moment Meshlab seems the way to go if you are striving to keep high detail level in your scans after fusions.

Image

SL Scan with Logitech C601 @ 1280x1024 / 5fps
Acer K11 @ 800x600 / 60Hz

see you around...
Christian
www.greyscale3di.com "if it ain´t movin´, we can scan it!"
Christian_Mendez
 
Posts: 188
Joined: Thu Nov 17, 2011 12:45 am
Location: Vienna, Austria

Postby hal » Thu Jan 12, 2012 12:09 am

Hello,

just a test with Shapefusion: try do duplicate 2 or 3 times both scans, and than fuse all the 4-6 scans all togheter. any differences?
Please, see here what I mean: the first image are 2 single scans, aligned. Second img is the fusion at 1200. Third is an other fusion at 1200, but each scans were duplicated 2 times (so every scan are x3 = 6 scans overlapped).
As you can easily seen, if you multiply the 3D data, you can increase the sharpness of the fine details. I kindly ask to Simon to wrote here more about the math behind the Fusion.

MatM_Fusion duplicated.jpg

I know that this trick literally eat the Ram amount of our computer, but for the moment, I hope that this can help a little.
Mattia
User avatar
hal
Moderator
 
Posts: 1527
Joined: Thu Mar 01, 2007 4:08 pm
Location: Cavaria con Premezzo (VA), Italy.

Re: Disappointing Results

Postby WalterMo » Thu Jan 12, 2012 8:50 pm

Hi,
The new DAVID Enterprise Edition will come with the Sharpness and the Refinement tools. I had the chance to do a comparison fusion of Sharpness 0 and 8 of an older SL face scan. See the difference. Fusion was set to 1200.

Walter
Attachments
Sharpness 0 and 8.jpg
User avatar
WalterMo
Moderator
 
Posts: 2363
Joined: Mon Apr 02, 2007 6:52 pm
Location: Braunschweig, Germany

Re: Disappointing Results

Postby Christian_Mendez » Thu Jan 12, 2012 9:52 pm

thanks Mattia and Walter,

I will try your suggestion Mattia, though I doubt that if its going to suck up my memory ( I have 8GB) that I will consider this as a part of my current workflow. I´d probably try and avoid it if I can ( especialy when having multiple scans @ 1600x1200 - i´m sure it´ll kick me out hehehe :lol: ) on the other hand, if I dont have too much to fuse then it´s well worth considering. Either way, I won´t be getting what I´m looking for without extra work in DAVID or post processing in Meshlab. Wish me luck.

I figured that the enterprise edition will surely improve this issue, so have you been doing more tests Walter ? I belong to the unpatient kind of people when it comes to these kind of things. I cant wait to test it out myself !
But anyway...any infos on the release date and cost for the enterprise edition Walter ?

Christian
www.greyscale3di.com "if it ain´t movin´, we can scan it!"
Christian_Mendez
 
Posts: 188
Joined: Thu Nov 17, 2011 12:45 am
Location: Vienna, Austria

Re: Disappointing Results

Postby WalterMo » Fri Jan 13, 2012 10:08 am

Christian,
Sorry, but I don't know much about the Enterprise Edition. It will come as a 64 bit software. But when and its costs....?

Walter
User avatar
WalterMo
Moderator
 
Posts: 2363
Joined: Mon Apr 02, 2007 6:52 pm
Location: Braunschweig, Germany

Re: Disappointing Results

Postby Christian_Mendez » Fri Jan 13, 2012 11:34 am

Thanks Walter...

I guess I´ll have to wait some more then...
www.greyscale3di.com "if it ain´t movin´, we can scan it!"
Christian_Mendez
 
Posts: 188
Joined: Thu Nov 17, 2011 12:45 am
Location: Vienna, Austria

Re: Disappointing Results

Postby vchmiele » Fri Jan 13, 2012 11:42 am

Currently I am having issues with the speed of capturing a scan (takes a couple second, even with only 12 patterns).

Is this slow speed a symptom of David or could it be my system? Will more memory allow the capture rate to increase?
vchmiele
 
Posts: 61
Joined: Mon Feb 21, 2011 11:09 am

Re: Disappointing Results

Postby Simon » Tue Jan 17, 2012 9:54 am

Hi Chrisitan
I'm sorry that you are not satisfied with the fusion results.
Just a short note to the Enterprise Edition and the Sharpness and the Refinement tools:
If start DAVID without a valid license (e.g. by temporarily renaming the license file "DAVID3_license.lic" to "DAVID3_license.lic_bak") you can test the Sharpness and Refinement feature yourself but without the ability to save the results.
User avatar
Simon
DAVID Developer
 
Posts: 423
Joined: Sun Jan 21, 2007 3:40 pm
Location: Braunschweig - Germany


Return to Shape Fusion

Who is online

Users browsing this forum: No registered users and 1 guest

cron