3D Scanners
On Fri, Nov 22, 2024 at 01:16:25AM -0500, gene heskett via Discuss wrote:
If running klipper, install kiauh-master, cd to it and do ./kiaoh.sh
and follow the menu prompts, bearing in mind it spends a lot of time
in the backgound downloading the latest repo lists. I've been using
it even for system upgrades for around 3 years. The only thing you
need to do when it is done is a reboot. octoprint amd its kin have
been history here for years.
The one that is not on raspberry pi is the 2D foam cutter. I use it
for one job that is about 60 gcode commands long and it runs for 15
minutes.
There is nothing that klipper would help with. It runs stock marlin on
the controller board and an older version of octoprint, recent at the
time of install. I just hit "print" on it every time I need a new
sheet sliced, and then it goes on its merry way. The GCode viewer
doesn't work with my manually written GCode. I don't know why. I don't
care. I know what it is going to do. And the buffer of commands is
large enough that it jumps to "25%done" in the first second after
starting a print. That's sortof useless. I'tll say "100%" for 5
minutes before it is done.
Roger.
--
** R.E.Wolff@BitWizard.nl ** https://www.BitWizard.nl/ ** +31-15-2049110 **
** Delftechpark 11 2628 XJ Delft, The Netherlands. KVK: 27239233 **
f equals m times a. When your f is steady, and your m is going down
your a** is going up. -- Chris Hadfield about flying up the space shuttle.
** 'a' for accelleration.
The RevoPoint software converts point clouds into some intermediate data
and then into mesh. You can control each of the conversions/transitions
between each data form, and can use external software to process each,
in addition to the RevoPoint software.
I use Blender (dragged there kicking and screaming) to smooth out scans
once they are in mesh form. Anything that automatically smooths
everything will also reduce detail
Jon
On 11/21/2024 11:48 PM, pca006132 via Discuss wrote:
There is no right answer to this sort of postprocessing, only
heuristics. If you assume the ground truth are mostly flat surfaces,
it is possible to fill in small holes, voxels make it easier to do
this kind of operations because you won't get invalid meshes from it.
It can be far from the ground truth, but it can never be invalid geometry.
And I'm not saying it is easy to recognize in software that the input
is a flat surface or smooth curve, this is difficult. What I am saying
is if you assume the target is like this, you can write heuristics to
target it. State-of-the-art mesh reconstruction from scanned data
often requires both point-cloud as well as visual data. With visual
data it is easier to figure out continuous surfaces, given the
lighting condition is good. But even with all these, it can still be
hard to obtain a full scan when your input have valleys, and it may
require stiching multiple scans together. In that case, the visual
information can help as well by locating feature points using computer
vision methods. There are PhDs and companies working on this, but they
target more general cases and have very different requirements.
Re. PhD level research, a lot of issues important to openscad users
are like that. Good mesh simplification, mesh repair to recover
manifoldness and remove overlaps, faster mesh offset, generic ways to
implement fillet. There are literatures out there, but a lot of them
are missing details, or are problematic in some ways. It is not like
PhD level research is something much harder than our regular
development...
On Fri, Nov 22, 2024 at 12:09 PM Leonard Martin Struttmann via Discuss
discuss@lists.openscad.org wrote:
Roger, you are probably right, but, <sarcasm> isn't that the
unspoken promise of AI? To relieve us humans of the burden of
doing these things ourselves? </sarcasm> Who knows?
In any event, I ordered a Miraco Pro today. I will report back
here after I've had some time to play with it.
Thanks! Len
On Thu, Nov 21, 2024 at 9:26 PM Rogier Wolff via Discuss
<discuss@lists.openscad.org> wrote:
On Thu, Nov 21, 2024 at 08:04:53PM +0800, Chun Kit LAM via
Discuss wrote:
It might be possible to convert the point cloud into voxels,
convert
the voxel into a mesh, and do some post-processing on the
mesh. If
the target features are cube like (flat surfaces, right angles,
etc.), this may work well.
I suspect this (the post-processing) is still a PHD level research
project to "do it right". For a human it is trivially easy to
recognize: Oh, that's supposed to be a flat surface, and that
other
thing is a smooth curve.
Roger.
--
** R.E.Wolff@BitWizard.nl ** https://www.BitWizard.nl/
<https://www.BitWizard.nl/> ** +31-15-2049110 **
** Delftechpark 11 2628 XJ Delft, The Netherlands. KVK:
27239233 **
f equals m times a. When your f is steady, and your m is going
down
your a** is going up. -- Chris Hadfield about flying up the
space shuttle.
** 'a' for accelleration.
_______________________________________________
OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.org
_______________________________________________
OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.org
OpenSCAD mailing list
To unsubscribe send an email todiscuss-leave@lists.openscad.org
--
This email has been checked for viruses by AVG antivirus software.
www.avg.com
Jon,
<feeble attempt at humor>"BLENDER!", he cried. "Oh, NO!!!" Not
that!"</humor>
Yeah, I tried Blender several years ago for a tensegrity tower project and
found it... unwieldy. I guess that I may need to revisit it. Or, perhaps,
MeshLab.
Thanks!
Len
On Fri, Nov 22, 2024 at 6:08 AM Jon Bondy via Discuss <
discuss@lists.openscad.org> wrote:
The RevoPoint software converts point clouds into some intermediate data
and then into mesh. You can control each of the conversions/transitions
between each data form, and can use external software to process each, in
addition to the RevoPoint software.
I use Blender (dragged there kicking and screaming) to smooth out scans
once they are in mesh form. Anything that automatically smooths everything
will also reduce detail
Jon
On 11/21/2024 11:48 PM, pca006132 via Discuss wrote:
There is no right answer to this sort of postprocessing, only heuristics.
If you assume the ground truth are mostly flat surfaces, it is possible to
fill in small holes, voxels make it easier to do this kind of operations
because you won't get invalid meshes from it. It can be far from the ground
truth, but it can never be invalid geometry.
And I'm not saying it is easy to recognize in software that the input is a
flat surface or smooth curve, this is difficult. What I am saying is if you
assume the target is like this, you can write heuristics to target it.
State-of-the-art mesh reconstruction from scanned data often requires both
point-cloud as well as visual data. With visual data it is easier to figure
out continuous surfaces, given the lighting condition is good. But even
with all these, it can still be hard to obtain a full scan when your input
have valleys, and it may require stiching multiple scans together. In that
case, the visual information can help as well by locating feature points
using computer vision methods. There are PhDs and companies working on
this, but they target more general cases and have very different
requirements.
Re. PhD level research, a lot of issues important to openscad users are
like that. Good mesh simplification, mesh repair to recover manifoldness
and remove overlaps, faster mesh offset, generic ways to implement fillet.
There are literatures out there, but a lot of them are missing details, or
are problematic in some ways. It is not like PhD level research is
something much harder than our regular development...
On Fri, Nov 22, 2024 at 12:09 PM Leonard Martin Struttmann via Discuss <
discuss@lists.openscad.org> wrote:
Roger, you are probably right, but, <sarcasm> isn't that the unspoken
promise of AI? To relieve us humans of the burden of doing these things
ourselves? </sarcasm> Who knows?
In any event, I ordered a Miraco Pro today. I will report back here
after I've had some time to play with it.
Thanks! Len
On Thu, Nov 21, 2024 at 9:26 PM Rogier Wolff via Discuss <
discuss@lists.openscad.org> wrote:
On Thu, Nov 21, 2024 at 08:04:53PM +0800, Chun Kit LAM via Discuss wrote:
It might be possible to convert the point cloud into voxels, convert
the voxel into a mesh, and do some post-processing on the mesh. If
the target features are cube like (flat surfaces, right angles,
etc.), this may work well.
I suspect this (the post-processing) is still a PHD level research
project to "do it right". For a human it is trivially easy to
recognize: Oh, that's supposed to be a flat surface, and that other
thing is a smooth curve.
Roger.
--
** R.E.Wolff@BitWizard.nl ** https://www.BitWizard.nl/ **
+31-15-2049110 **
** Delftechpark 11 2628 XJ Delft, The Netherlands. KVK: 27239233
**
f equals m times a. When your f is steady, and your m is going down
your a** is going up. -- Chris Hadfield about flying up the space
shuttle.
** 'a' for accelleration.
OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.org
OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.org
OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.org
http://www.avg.com/email-signature?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=emailclient
Virus-free.www.avg.com
http://www.avg.com/email-signature?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=emailclient
<#m_-7348258836865286824_DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2>
OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.org
If you limit your Blender use to smoothing, it works fine. The rest of the UI just irritates me beyond description. There are YouTube videos about how to use Blender just for smoothing.
On 11/22/2024 10:19 AM, Leonard Martin Struttmann wrote:
Jon,
<feeble attempt at humor>"BLENDER!", he cried. "Oh, NO!!!" Not that!"</humor>
Yeah, I tried Blender several years ago for a tensegrity tower project and found it... unwieldy. I guess that I may need to revisit it. Or, perhaps, MeshLab.
Thanks!
Len
On Fri, Nov 22, 2024 at 6:08 AM Jon Bondy via Discuss <discuss@lists.openscad.orgmailto:discuss@lists.openscad.org> wrote:
The RevoPoint software converts point clouds into some intermediate data and then into mesh. You can control each of the conversions/transitions between each data form, and can use external software to process each, in addition to the RevoPoint software.
I use Blender (dragged there kicking and screaming) to smooth out scans once they are in mesh form. Anything that automatically smooths everything will also reduce detail
Jon
On 11/21/2024 11:48 PM, pca006132 via Discuss wrote:
There is no right answer to this sort of postprocessing, only heuristics. If you assume the ground truth are mostly flat surfaces, it is possible to fill in small holes, voxels make it easier to do this kind of operations because you won't get invalid meshes from it. It can be far from the ground truth, but it can never be invalid geometry.
And I'm not saying it is easy to recognize in software that the input is a flat surface or smooth curve, this is difficult. What I am saying is if you assume the target is like this, you can write heuristics to target it. State-of-the-art mesh reconstruction from scanned data often requires both point-cloud as well as visual data. With visual data it is easier to figure out continuous surfaces, given the lighting condition is good. But even with all these, it can still be hard to obtain a full scan when your input have valleys, and it may require stiching multiple scans together. In that case, the visual information can help as well by locating feature points using computer vision methods. There are PhDs and companies working on this, but they target more general cases and have very different requirements.
Re. PhD level research, a lot of issues important to openscad users are like that. Good mesh simplification, mesh repair to recover manifoldness and remove overlaps, faster mesh offset, generic ways to implement fillet. There are literatures out there, but a lot of them are missing details, or are problematic in some ways. It is not like PhD level research is something much harder than our regular development...
On Fri, Nov 22, 2024 at 12:09 PM Leonard Martin Struttmann via Discuss <discuss@lists.openscad.orgmailto:discuss@lists.openscad.org> wrote:
Roger, you are probably right, but, <sarcasm> isn't that the unspoken promise of AI? To relieve us humans of the burden of doing these things ourselves? </sarcasm> Who knows?
In any event, I ordered a Miraco Pro today. I will report back here after I've had some time to play with it.
Thanks! Len
On Thu, Nov 21, 2024 at 9:26 PM Rogier Wolff via Discuss <discuss@lists.openscad.orgmailto:discuss@lists.openscad.org> wrote:
On Thu, Nov 21, 2024 at 08:04:53PM +0800, Chun Kit LAM via Discuss wrote:
It might be possible to convert the point cloud into voxels, convert
the voxel into a mesh, and do some post-processing on the mesh. If
the target features are cube like (flat surfaces, right angles,
etc.), this may work well.
I suspect this (the post-processing) is still a PHD level research
project to "do it right". For a human it is trivially easy to
recognize: Oh, that's supposed to be a flat surface, and that other
thing is a smooth curve.
Roger.
--
** R.E.Wolff@BitWizard.nlmailto:R.E.Wolff@BitWizard.nl ** https://www.BitWizard.nl/ ** +31-15-2049110 **
** Delftechpark 11 2628 XJ Delft, The Netherlands. KVK: 27239233 **
f equals m times a. When your f is steady, and your m is going down
your a** is going up. -- Chris Hadfield about flying up the space shuttle.
** 'a' for accelleration.
OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.orgmailto:discuss-leave@lists.openscad.org
OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.orgmailto:discuss-leave@lists.openscad.org
OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.orgmailto:discuss-leave@lists.openscad.org
[https://s-install.avcdn.net/ipm/preview/icons/icon-envelope-tick-green-avg-v1.png]http://www.avg.com/email-signature?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=emailclient Virus-free.www.avg.comhttp://www.avg.com/email-signature?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=emailclient
OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.orgmailto:discuss-leave@lists.openscad.org
So, my RevoPoint Miraco Pro arrived yesterday and I played with it most of
today. It seems to work... OK. I guess that I had higher expectations
than I should have.
The main purpose for getting it was to easily/quickly model complicated
electronics parts and assemblies so that I did not need to model them
myself in native OpenSCAD. My first target was a 1602 LCD display with
attached controller board. After learning the Miraco, it still took several
iterations in PrusaSlicer to simplify and correct errors until I got an OBJ
with enough detail AND could be imported into OpenSCAD in a reasonable
amount of time:
[image: LCD1602bottom.png]
[image: de947e2e-edc1-4f69-a699-ea045760e08f.png]
This is a 589KB OBJ file with 17,654 faces. It imports into OpenSCAD on my
Windows machine in 2.9 seconds.
On Fri, Nov 22, 2024 at 10:06 AM jon jonbondy.com jon@jonbondy.com wrote:
If you limit your Blender use to smoothing, it works fine. The rest of
the UI just irritates me beyond description. There are YouTube videos
about how to use Blender just for smoothing.
On 11/22/2024 10:19 AM, Leonard Martin Struttmann wrote:
Jon,
<feeble attempt at humor>"BLENDER!", he cried. "Oh, NO!!!" Not
that!"</humor>
Yeah, I tried Blender several years ago for a tensegrity tower project and
found it... unwieldy. I guess that I may need to revisit it. Or, perhaps,
MeshLab.
Thanks!
Len
On Fri, Nov 22, 2024 at 6:08 AM Jon Bondy via Discuss <
discuss@lists.openscad.org> wrote:
The RevoPoint software converts point clouds into some intermediate data
and then into mesh. You can control each of the conversions/transitions
between each data form, and can use external software to process each, in
addition to the RevoPoint software.
I use Blender (dragged there kicking and screaming) to smooth out scans
once they are in mesh form. Anything that automatically smooths everything
will also reduce detail
Jon
On 11/21/2024 11:48 PM, pca006132 via Discuss wrote:
There is no right answer to this sort of postprocessing, only heuristics.
If you assume the ground truth are mostly flat surfaces, it is possible to
fill in small holes, voxels make it easier to do this kind of operations
because you won't get invalid meshes from it. It can be far from the ground
truth, but it can never be invalid geometry.
And I'm not saying it is easy to recognize in software that the input is
a flat surface or smooth curve, this is difficult. What I am saying is if
you assume the target is like this, you can write heuristics to target it.
State-of-the-art mesh reconstruction from scanned data often requires both
point-cloud as well as visual data. With visual data it is easier to figure
out continuous surfaces, given the lighting condition is good. But even
with all these, it can still be hard to obtain a full scan when your input
have valleys, and it may require stiching multiple scans together. In that
case, the visual information can help as well by locating feature points
using computer vision methods. There are PhDs and companies working on
this, but they target more general cases and have very different
requirements.
Re. PhD level research, a lot of issues important to openscad users are
like that. Good mesh simplification, mesh repair to recover manifoldness
and remove overlaps, faster mesh offset, generic ways to implement fillet.
There are literatures out there, but a lot of them are missing details, or
are problematic in some ways. It is not like PhD level research is
something much harder than our regular development...
On Fri, Nov 22, 2024 at 12:09 PM Leonard Martin Struttmann via Discuss <
discuss@lists.openscad.org> wrote:
Roger, you are probably right, but, <sarcasm> isn't that the unspoken
promise of AI? To relieve us humans of the burden of doing these things
ourselves? </sarcasm> Who knows?
In any event, I ordered a Miraco Pro today. I will report back here
after I've had some time to play with it.
Thanks! Len
On Thu, Nov 21, 2024 at 9:26 PM Rogier Wolff via Discuss <
discuss@lists.openscad.org> wrote:
On Thu, Nov 21, 2024 at 08:04:53PM +0800, Chun Kit LAM via Discuss
wrote:
It might be possible to convert the point cloud into voxels, convert
the voxel into a mesh, and do some post-processing on the mesh. If
the target features are cube like (flat surfaces, right angles,
etc.), this may work well.
I suspect this (the post-processing) is still a PHD level research
project to "do it right". For a human it is trivially easy to
recognize: Oh, that's supposed to be a flat surface, and that other
thing is a smooth curve.
Roger.
--
** R.E.Wolff@BitWizard.nl ** https://www.BitWizard.nl/ **
+31-15-2049110 **
** Delftechpark 11 2628 XJ Delft, The Netherlands. KVK: 27239233
**
f equals m times a. When your f is steady, and your m is going down
your a** is going up. -- Chris Hadfield about flying up the space
shuttle.
** 'a' for accelleration.
OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.org
OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.org
OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.org
http://www.avg.com/email-signature?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=emailclient
Virus-free.www.avg.com
http://www.avg.com/email-signature?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=emailclient
<#m_-1138463494526674595_m_-7348258836865286824_DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2>
OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.org
Nice job. I am seeing a number of artifacts, but with some tweaking of the
settings and more experience with the unit, it will get much better.
On Sat, Nov 30, 2024 at 8:16 PM Leonard Martin Struttmann via Discuss <
discuss@lists.openscad.org> wrote:
So, my RevoPoint Miraco Pro arrived yesterday and I played with it most of
today. It seems to work... OK. I guess that I had higher expectations
than I should have.
The main purpose for getting it was to easily/quickly model complicated
electronics parts and assemblies so that I did not need to model them
myself in native OpenSCAD. My first target was a 1602 LCD display with
attached controller board. After learning the Miraco, it still took several
iterations in PrusaSlicer to simplify and correct errors until I got an OBJ
with enough detail AND could be imported into OpenSCAD in a reasonable
amount of time:
[image: LCD1602bottom.png]
[image: de947e2e-edc1-4f69-a699-ea045760e08f.png]
This is a 589KB OBJ file with 17,654 faces. It imports into OpenSCAD on
my Windows machine in 2.9 seconds.
On Fri, Nov 22, 2024 at 10:06 AM jon jonbondy.com jon@jonbondy.com
wrote:
If you limit your Blender use to smoothing, it works fine. The rest of
the UI just irritates me beyond description. There are YouTube videos
about how to use Blender just for smoothing.
On 11/22/2024 10:19 AM, Leonard Martin Struttmann wrote:
Jon,
<feeble attempt at humor>"BLENDER!", he cried. "Oh, NO!!!" Not
that!"</humor>
Yeah, I tried Blender several years ago for a tensegrity tower project
and found it... unwieldy. I guess that I may need to revisit it. Or,
perhaps, MeshLab.
Thanks!
Len
On Fri, Nov 22, 2024 at 6:08 AM Jon Bondy via Discuss <
discuss@lists.openscad.org> wrote:
The RevoPoint software converts point clouds into some intermediate data
and then into mesh. You can control each of the conversions/transitions
between each data form, and can use external software to process each, in
addition to the RevoPoint software.
I use Blender (dragged there kicking and screaming) to smooth out scans
once they are in mesh form. Anything that automatically smooths everything
will also reduce detail
Jon
On 11/21/2024 11:48 PM, pca006132 via Discuss wrote:
There is no right answer to this sort of postprocessing, only
heuristics. If you assume the ground truth are mostly flat surfaces, it is
possible to fill in small holes, voxels make it easier to do this kind of
operations because you won't get invalid meshes from it. It can be far from
the ground truth, but it can never be invalid geometry.
And I'm not saying it is easy to recognize in software that the input is
a flat surface or smooth curve, this is difficult. What I am saying is if
you assume the target is like this, you can write heuristics to target it.
State-of-the-art mesh reconstruction from scanned data often requires both
point-cloud as well as visual data. With visual data it is easier to figure
out continuous surfaces, given the lighting condition is good. But even
with all these, it can still be hard to obtain a full scan when your input
have valleys, and it may require stiching multiple scans together. In that
case, the visual information can help as well by locating feature points
using computer vision methods. There are PhDs and companies working on
this, but they target more general cases and have very different
requirements.
Re. PhD level research, a lot of issues important to openscad users are
like that. Good mesh simplification, mesh repair to recover manifoldness
and remove overlaps, faster mesh offset, generic ways to implement fillet.
There are literatures out there, but a lot of them are missing details, or
are problematic in some ways. It is not like PhD level research is
something much harder than our regular development...
On Fri, Nov 22, 2024 at 12:09 PM Leonard Martin Struttmann via Discuss <
discuss@lists.openscad.org> wrote:
Roger, you are probably right, but, <sarcasm> isn't that the unspoken
promise of AI? To relieve us humans of the burden of doing these things
ourselves? </sarcasm> Who knows?
In any event, I ordered a Miraco Pro today. I will report back here
after I've had some time to play with it.
Thanks! Len
On Thu, Nov 21, 2024 at 9:26 PM Rogier Wolff via Discuss <
discuss@lists.openscad.org> wrote:
On Thu, Nov 21, 2024 at 08:04:53PM +0800, Chun Kit LAM via Discuss
wrote:
It might be possible to convert the point cloud into voxels, convert
the voxel into a mesh, and do some post-processing on the mesh. If
the target features are cube like (flat surfaces, right angles,
etc.), this may work well.
I suspect this (the post-processing) is still a PHD level research
project to "do it right". For a human it is trivially easy to
recognize: Oh, that's supposed to be a flat surface, and that other
thing is a smooth curve.
Roger.
--
** R.E.Wolff@BitWizard.nl ** https://www.BitWizard.nl/ **
+31-15-2049110 **
** Delftechpark 11 2628 XJ Delft, The Netherlands. KVK: 27239233
**
f equals m times a. When your f is steady, and your m is going down
your a** is going up. -- Chris Hadfield about flying up the space
shuttle.
** 'a' for accelleration.
OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.org
OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.org
OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.org
http://www.avg.com/email-signature?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=emailclient
Virus-free.www.avg.com
http://www.avg.com/email-signature?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=emailclient
<#m_3045179401418616933_m_-1138463494526674595_m_-7348258836865286824_DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2>
OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.org
OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.org
Thanks! My biggest disappointment was that when I used the Revo Scan 5
software to simplify the mesh, it then became non-manifold and OpenScad
didn't like it. I expected Revo Scan to be better than that.
On Sun, Dec 1, 2024 at 9:39 AM John David ebo.2112@gmail.com wrote:
Nice job. I am seeing a number of artifacts, but with some tweaking of
the settings and more experience with the unit, it will get much better.
On Sat, Nov 30, 2024 at 8:16 PM Leonard Martin Struttmann via Discuss <
discuss@lists.openscad.org> wrote:
So, my RevoPoint Miraco Pro arrived yesterday and I played with it most
of today. It seems to work... OK. I guess that I had higher expectations
than I should have.
The main purpose for getting it was to easily/quickly model complicated
electronics parts and assemblies so that I did not need to model them
myself in native OpenSCAD. My first target was a 1602 LCD display with
attached controller board. After learning the Miraco, it still took several
iterations in PrusaSlicer to simplify and correct errors until I got an OBJ
with enough detail AND could be imported into OpenSCAD in a reasonable
amount of time:
[image: LCD1602bottom.png]
[image: de947e2e-edc1-4f69-a699-ea045760e08f.png]
This is a 589KB OBJ file with 17,654 faces. It imports into OpenSCAD on
my Windows machine in 2.9 seconds.
On Fri, Nov 22, 2024 at 10:06 AM jon jonbondy.com jon@jonbondy.com
wrote:
If you limit your Blender use to smoothing, it works fine. The rest of
the UI just irritates me beyond description. There are YouTube videos
about how to use Blender just for smoothing.
On 11/22/2024 10:19 AM, Leonard Martin Struttmann wrote:
Jon,
<feeble attempt at humor>"BLENDER!", he cried. "Oh, NO!!!" Not
that!"</humor>
Yeah, I tried Blender several years ago for a tensegrity tower project
and found it... unwieldy. I guess that I may need to revisit it. Or,
perhaps, MeshLab.
Thanks!
Len
On Fri, Nov 22, 2024 at 6:08 AM Jon Bondy via Discuss <
discuss@lists.openscad.org> wrote:
The RevoPoint software converts point clouds into some intermediate
data and then into mesh. You can control each of the
conversions/transitions between each data form, and can use external
software to process each, in addition to the RevoPoint software.
I use Blender (dragged there kicking and screaming) to smooth out scans
once they are in mesh form. Anything that automatically smooths everything
will also reduce detail
Jon
On 11/21/2024 11:48 PM, pca006132 via Discuss wrote:
There is no right answer to this sort of postprocessing, only
heuristics. If you assume the ground truth are mostly flat surfaces, it is
possible to fill in small holes, voxels make it easier to do this kind of
operations because you won't get invalid meshes from it. It can be far from
the ground truth, but it can never be invalid geometry.
And I'm not saying it is easy to recognize in software that the input
is a flat surface or smooth curve, this is difficult. What I am saying is
if you assume the target is like this, you can write heuristics to target
it. State-of-the-art mesh reconstruction from scanned data often requires
both point-cloud as well as visual data. With visual data it is easier to
figure out continuous surfaces, given the lighting condition is good. But
even with all these, it can still be hard to obtain a full scan when your
input have valleys, and it may require stiching multiple scans together. In
that case, the visual information can help as well by locating feature
points using computer vision methods. There are PhDs and companies working
on this, but they target more general cases and have very different
requirements.
Re. PhD level research, a lot of issues important to openscad users are
like that. Good mesh simplification, mesh repair to recover manifoldness
and remove overlaps, faster mesh offset, generic ways to implement fillet.
There are literatures out there, but a lot of them are missing details, or
are problematic in some ways. It is not like PhD level research is
something much harder than our regular development...
On Fri, Nov 22, 2024 at 12:09 PM Leonard Martin Struttmann via Discuss <
discuss@lists.openscad.org> wrote:
Roger, you are probably right, but, <sarcasm> isn't that the unspoken
promise of AI? To relieve us humans of the burden of doing these things
ourselves? </sarcasm> Who knows?
In any event, I ordered a Miraco Pro today. I will report back here
after I've had some time to play with it.
Thanks! Len
On Thu, Nov 21, 2024 at 9:26 PM Rogier Wolff via Discuss <
discuss@lists.openscad.org> wrote:
On Thu, Nov 21, 2024 at 08:04:53PM +0800, Chun Kit LAM via Discuss
wrote:
It might be possible to convert the point cloud into voxels, convert
the voxel into a mesh, and do some post-processing on the mesh. If
the target features are cube like (flat surfaces, right angles,
etc.), this may work well.
I suspect this (the post-processing) is still a PHD level research
project to "do it right". For a human it is trivially easy to
recognize: Oh, that's supposed to be a flat surface, and that other
thing is a smooth curve.
Roger.
--
** R.E.Wolff@BitWizard.nl ** https://www.BitWizard.nl/ **
+31-15-2049110 **
** Delftechpark 11 2628 XJ Delft, The Netherlands. KVK:
27239233 **
f equals m times a. When your f is steady, and your m is going down
your a** is going up. -- Chris Hadfield about flying up the space
shuttle.
** 'a' for accelleration.
OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.org
OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.org
OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.org
http://www.avg.com/email-signature?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=emailclient
Virus-free.www.avg.com
http://www.avg.com/email-signature?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=emailclient
<#m_-6381106329078966213_m_3045179401418616933_m_-1138463494526674595_m_-7348258836865286824_DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2>
OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.org
OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.org
The Revo Scan software can close holes, if you can find the feature.
On 12/1/2024 11:07 AM, Leonard Martin Struttmann wrote:
Thanks! My biggest disappointment was that when I used the Revo Scan 5 software to simplify the mesh, it then became non-manifold and OpenScad didn't like it. I expected Revo Scan to be better than that.
On Sun, Dec 1, 2024 at 9:39 AM John David <ebo.2112@gmail.commailto:ebo.2112@gmail.com> wrote:
Nice job. I am seeing a number of artifacts, but with some tweaking of the settings and more experience with the unit, it will get much better.
On Sat, Nov 30, 2024 at 8:16 PM Leonard Martin Struttmann via Discuss <discuss@lists.openscad.orgmailto:discuss@lists.openscad.org> wrote:
So, my RevoPoint Miraco Pro arrived yesterday and I played with it most of today. It seems to work... OK. I guess that I had higher expectations than I should have.
The main purpose for getting it was to easily/quickly model complicated electronics parts and assemblies so that I did not need to model them myself in native OpenSCAD. My first target was a 1602 LCD display with attached controller board. After learning the Miraco, it still took several iterations in PrusaSlicer to simplify and correct errors until I got an OBJ with enough detail AND could be imported into OpenSCAD in a reasonable amount of time:
[LCD1602bottom.png]
[de947e2e-edc1-4f69-a699-ea045760e08f.png]
This is a 589KB OBJ file with 17,654 faces. It imports into OpenSCAD on my Windows machine in 2.9 seconds.
On Fri, Nov 22, 2024 at 10:06 AM jon jonbondy.comhttp://jonbondy.com <jon@jonbondy.commailto:jon@jonbondy.com> wrote:
If you limit your Blender use to smoothing, it works fine. The rest of the UI just irritates me beyond description. There are YouTube videos about how to use Blender just for smoothing.
On 11/22/2024 10:19 AM, Leonard Martin Struttmann wrote:
Jon,
<feeble attempt at humor>"BLENDER!", he cried. "Oh, NO!!!" Not that!"</humor>
Yeah, I tried Blender several years ago for a tensegrity tower project and found it... unwieldy. I guess that I may need to revisit it. Or, perhaps, MeshLab.
Thanks!
Len
On Fri, Nov 22, 2024 at 6:08 AM Jon Bondy via Discuss <discuss@lists.openscad.orgmailto:discuss@lists.openscad.org> wrote:
The RevoPoint software converts point clouds into some intermediate data and then into mesh. You can control each of the conversions/transitions between each data form, and can use external software to process each, in addition to the RevoPoint software.
I use Blender (dragged there kicking and screaming) to smooth out scans once they are in mesh form. Anything that automatically smooths everything will also reduce detail
Jon
On 11/21/2024 11:48 PM, pca006132 via Discuss wrote:
There is no right answer to this sort of postprocessing, only heuristics. If you assume the ground truth are mostly flat surfaces, it is possible to fill in small holes, voxels make it easier to do this kind of operations because you won't get invalid meshes from it. It can be far from the ground truth, but it can never be invalid geometry.
And I'm not saying it is easy to recognize in software that the input is a flat surface or smooth curve, this is difficult. What I am saying is if you assume the target is like this, you can write heuristics to target it. State-of-the-art mesh reconstruction from scanned data often requires both point-cloud as well as visual data. With visual data it is easier to figure out continuous surfaces, given the lighting condition is good. But even with all these, it can still be hard to obtain a full scan when your input have valleys, and it may require stiching multiple scans together. In that case, the visual information can help as well by locating feature points using computer vision methods. There are PhDs and companies working on this, but they target more general cases and have very different requirements.
Re. PhD level research, a lot of issues important to openscad users are like that. Good mesh simplification, mesh repair to recover manifoldness and remove overlaps, faster mesh offset, generic ways to implement fillet. There are literatures out there, but a lot of them are missing details, or are problematic in some ways. It is not like PhD level research is something much harder than our regular development...
On Fri, Nov 22, 2024 at 12:09 PM Leonard Martin Struttmann via Discuss <discuss@lists.openscad.orgmailto:discuss@lists.openscad.org> wrote:
Roger, you are probably right, but, <sarcasm> isn't that the unspoken promise of AI? To relieve us humans of the burden of doing these things ourselves? </sarcasm> Who knows?
In any event, I ordered a Miraco Pro today. I will report back here after I've had some time to play with it.
Thanks! Len
On Thu, Nov 21, 2024 at 9:26 PM Rogier Wolff via Discuss <discuss@lists.openscad.orgmailto:discuss@lists.openscad.org> wrote:
On Thu, Nov 21, 2024 at 08:04:53PM +0800, Chun Kit LAM via Discuss wrote:
It might be possible to convert the point cloud into voxels, convert
the voxel into a mesh, and do some post-processing on the mesh. If
the target features are cube like (flat surfaces, right angles,
etc.), this may work well.
I suspect this (the post-processing) is still a PHD level research
project to "do it right". For a human it is trivially easy to
recognize: Oh, that's supposed to be a flat surface, and that other
thing is a smooth curve.
Roger.
--
** R.E.Wolff@BitWizard.nlmailto:R.E.Wolff@BitWizard.nl ** https://www.BitWizard.nl/ ** +31-15-2049110 **
** Delftechpark 11 2628 XJ Delft, The Netherlands. KVK: 27239233 **
f equals m times a. When your f is steady, and your m is going down
your a** is going up. -- Chris Hadfield about flying up the space shuttle.
** 'a' for accelleration.
OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.orgmailto:discuss-leave@lists.openscad.org
OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.orgmailto:discuss-leave@lists.openscad.org
OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.orgmailto:discuss-leave@lists.openscad.org
[https://s-install.avcdn.net/ipm/preview/icons/icon-envelope-tick-green-avg-v1.png]http://www.avg.com/email-signature?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=emailclient Virus-free.www.avg.comhttp://www.avg.com/email-signature?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=emailclient
OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.orgmailto:discuss-leave@lists.openscad.org
OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.orgmailto:discuss-leave@lists.openscad.org
