Full Version: V4 beta Oct-17-2018 available now - SubD to NURBS conversion

Show messages:  1-7  8-27  28-47  48-67  68-87  88-107  108-127  128-147  …  248-264

Thread Split: Some posts in this thread have been moved here

Hi Michael,

New build is great - just one with the STEP import or Open still getting recursive naming is there a way to fix this?

---

Hi Stefan,

re:
> It says now in "Size" 200x199,99x199,99 mm and in the "Edit Size" 200x200x200 mm.
>
> I am wondering now what readout is the correct value? As i figured out this happens also in MoI v3.1.

The one in the "Edit Size" pop out menu is the more accurate one. The one that is shown on top level UI is an approximation based off of the display mesh. Calculating the accurate tight bounding box of a solid can be a potentially time consuming operation and the top level UI changes with every selection.

When you pop out the menu it will do the more detailed and accurate (and slower) calculation at that time.

- Michael

---

Hi Stefan,

> I wonder what causes this in MoI, with my ViaCAD model???

The bounding box size isn't something that's fundamentally built in to the NURBS definition, it's something that has to be calculated using analysis techniques and a tolerance value.

So various factors are at play, it's just by luck that you see different behavior between those 2 cases I think.

- Michael

---

Hi Alex,

> New build is great - just one with the STEP import or Open still getting recursive
> naming is there a way to fix this?

Is this something that's different behavior between MoI v3 and v4?

Also I wanted to ask you if the missing object names when exporting to SolidWorks is fixed up?

Thanks,
- Michael

---

Hi Przemas,

re:
> Even if I won't be able to make MoI work under Linux with Nvidia GPU without glitches
> (https://vimeo.com/296282117) this feature alone will make it a no brainer purchase.

I just got a computer with an Nvidia card on it over here but I am not able to reproduce that selection behavior at all, it's working normally for me.

This is on:

Linux Mint 19 Cinnamon
NVIDIA GeForce GT 750M
wine-3.0 (Ubuntu 3.0-1ubuntu1) - installed with "sudo apt-get install wine-stable" and "sudo apt-get install wine32", (wine32 is needed for the installer)

And I tried with both the proprietary and the nouveau drivers and both were working for me. Those are
nvidia-driver-390 (Version 390.77-0ubuntu0.18.04.1)
xserve-xorg-video-nouveau (Version 1:1.0.15-2)

Are your OS, Wine, or driver versions substantially different from those?

I've also tested previously with my main system that has AMD graphics and I did not see that delayed frame behavior there either.

- Michael

---

Hi Michael,

thanks for the replies! I can understand the thing about tolerances and calculations
which has to been made for the display values, but i don't understand once a model
is imported that it can display values which are changing, while the native one does
not change its display values, when repeating commands.

Best regards
Stefan

---

Hi Stefan, can you please send me the the native and non-native files so I can take a look?

There are various kinds of geometry processing that can happen with importing non-native files, things like UV trim curves being created from 3D curves. So there can be tiny differences from that type of stuff.

Thanks, - Michael

---

Hi Michael,

The objects names now port to SW correctly - the recursive name has been a problem in MOIv3 - it would be great fix that one.

Thanks again
Alex.

---

Hi Michael,

thanks for the info! I already send you yesterday the .obj from which i created the MoI file
and the .sat file from ViaCAD. Hope it is not in your spam folder. In case i must re-send, i
can do it when i come from work.

Regards
Stefan

---

Hi Stefan, I got a big .obj file from you but I don't seem to have the .sat / .obj files for this case.

- Michael

---

Hi Michael,

i just forwarded the email again. Subject is/was "Scale discrepancies" and the attachment
is named BC.tar.gz. Hope that you received it now!

Regards
Stefan

---

Hi Michael,

there seems to be a mailer problem, which did not show up yesterday.

----- The following addresses had permanent fatal errors -----
<moi@moi3d.com>
(reason: 552 Message limit exceeded 23048261>20971520)

----- Transcript of session follows -----
... while talking to [xxx.xxx.xxx.xxx]:
>>> MAIL From:<xxxxxxxxx@xxxxx.xxx> SIZE=23048261
<<< 552 Message limit exceeded 23048261>20971520
554 5.0.0 Service unavailable

Regards
Stefan

P.S. when i come home i will upload the files to my keybase
account and then send you a download link via email.

Regards
Stefan

---

Game changer. Thanks for this. I was hoping for something like this in V4!!

---

re:
>I just got a computer with an Nvidia card on it over here but I am not able to reproduce that selection behavior at all, it's working normally for me.

thx for looking into it Michael! Appreciate it.
Looking at your specs - are you sure you've been running it on an Nvidia GPU?
I've noticed that you have GTX750M which suggests a discrete GPU on an Optimus enabled laptop - which would default on Linux to intel card without a bit of trickery. And even if it was running proprietary driver it would require primus / bumblebee which sadly adds a bit of variables to the mix.

I've did a test on a friends laptop which has a discrete GPU as well (GTX 660M). With nouveau driver it worked. Haven't had a chance to try proprietary driver on this machine though - I was a tad afraid to mess it up (my friend wouldn't be happy about it :D ). Also I forgot to check whether it was X11 or Wayland session (as it was Fedora 28 it is quite huge chance it was Wayland) - which would add another variable :/ .

I did a test on both of my workstations though and with proprietary driver I'm seeing the same problem. There's one more thing I'll check - both are dual Xeon machines. While unlikely it adds to the mix I'd like to test it too. I think I've tested it in the past and ruled it out, but as it has been 2 or 3 years I might be wrong.

I've launched MoI in Wine with Xephyr (thx to Lutris) and in such setting all those highlights / selection windows / etc update correctly. But it is not usable - under Xephyr it is super slow.

Part of the reason I look forward to V4 with the API selection option in cfg - I have a bit of hopes that maybe with proper forced API and DXVK the issue will be gone without extra hassle.

---

Hi Przemas,

> Looking at your specs - are you sure you've been running it on an Nvidia GPU?

Pretty sure - the information I posted came directly from the "System Information" OS dialog and there were options to switch between Nouveau or the proprietary driver.


> I've noticed that you have GTX750M which suggests a discrete GPU on an Optimus
> enabled laptop - which would default on Linux to intel card without a bit of trickery.

It's a slightly older iMac. I was hoping it might reproduce your problem as well as a Mac specific problem.

- Michael

---

ok, so it is not a laptop with Optimus. No clue how it has been implemented on iMac.
if I manage to do some more thorough tests during the weekend I'll get back with my findings. I've also resurrected a thread over at WINE forums.

---

Wow!! Amazing feature!

Marc

---

Message 9100.85 was deleted

---

This is a much needed tool in the industry when it comes to helping the VFX world communicate with the CAD world. There are a few tools on the market but most are at the higher end of the $$ so it's nice to see that MoI is bringing this out.

Two questions

First - The option to "predetermine" edges from the quad mesh to the NURBS geometry. So this wouldn't be edges weighting as much as it is selecting a quad patch and designating it for the boundaries edge. This would make for the downstream CAD application to have a "better" edge to start from (HA see what I did there). The general quilted patch, while very much a huge step over working with a mesh, is not giving the topology flow that say if you're looking to create a parting line from. Having the ability to basically set that up with the quad mesh just gives that next level of push towards the down stream CAD application. See image below. Shown on the left is the CAD model converted without the quads in the SubD model being selected, the middle shows the selected quads that then become hard edges in the NURBS model shown in the CAD software on the right.




Second - Along the possible workflow of the above being able to have the ability to paint the quads in such a way in areas that during the conversion process it will use smaller patches and say larger patches where it's not painted. There are times when an very evenly spaced NURBS quilt is needed but the possible option to select where a more dense level of detail is needed would be great to have.

Image Attachments:
SubD to NURBS.JPG 

---

Hi Arthur, for v4 the focus is just on getting the overall shape converted over reliably. It could be possible in the future to add what you are describing though, maybe having the surface combining part pay attention to material assignments could be a practical way to do that. Like large patches would only be combined with faces belonging to the same material assignment.

Would that cover the type of thing you're asking about or is it more detailed than that?

- Michael

---

Show messages:  1-7  8-27  28-47  48-67  68-87  88-107  108-127  128-147  …  248-264