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

Show messages:  1-17  …  78-97  98-117  118-137  138-157  158-177  178-197  198-217  …  258-264

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

The reason I wanted all quads was to make the mesh work better with Blenders bevel modifier. For some surfaces the Ngon export works fine but curved surfaces tend to be hit or miss, I can clean the mesh up to have it work better but having all quads would be a nice workflow improvement. I know I can bevel in Moi but the flexibility having the bevels as a modifier in Blender is very useful and I find that the edge selection in Moi to be a little tedious (when beveling the whole object isn't wanted) where in Blender it can be almost automatic. This way I could export a triangulated mesh out of Moi as well and use the beveled version from Blender as a high res for baking, at least that's what I was thinking for my workflow.


Would it be possible to implement an all quad export for Moi? It would be a great time saver and the models would be more flexible with other software outside of CAD.

---

Hi Foxfire, yes typically in CAD modeling the beveling is done on the CAD model itself and so the polygons that you export are all in the end shape already.

In the future I would like to work on an all quads export for MoI but it will likely be a quite difficult task and take a considerable amount of effort to develop. It needs a much different approach than the current mesh export.

Currently the way you would get all quads would be using retopo tools.

- Michael

---

I can heartily recommend ZBrush ZRemesher or 3D-Coat Autopo for automatic quad-retopology of meshes exported from MoI:

https://www.metinseven.nl/blog/Ag4W/3d-coat-vs-zbrush-auto-retopology-comparison

---

Tried it out with the corvette tutorial model from cgmasters. Imported both the underlying clean reference surfaces as well as the detailed parts to review at which point it makes sense to convert to nurb. Both come in beautifully!
https://cgmasters.net/training-courses/master-car-creation-in-blender/

It is a little on the heavy side but this should be huge for my team- We have been increasingly using SubD techniques for concept modeling. This will be extremely useful for downstream communication with Engineering teams- even though the surfaces are heavy it is still something that a Mech Engg team member can open and evaluate component clearances general design intent etc:

Image Attachments:
Screenshot 2018-11-18 22.21.23.png  Screenshot 2018-11-18 22.33.12.png  Screenshot 2018-11-18 22.33.21.png  Screenshot 2018-11-18 22.34.04.png  Screenshot 2018-11-18 22.34.28.png  Screenshot 2018-11-18 22.36.54.png  Screenshot 2018-11-19 11.01.52.png  Screenshot 2018-11-19 11.01.59.png 

---

Hi parel,

> Tried it out with the corvette tutorial model from cgmasters.

It's looking good! One thing to check - make sure you're exporting the roughest base control cage. You generally don't want to export polygons that have already been subdivided some already because that will make things a lot heavier.

- Michael

---

Does this point count look appropriate for the control cage imported?

Image Attachments:
Screenshot 2018-11-19 18.24.21.png  Screenshot 2018-11-19 18.24.50.png 

---

Hi parel yeah I guess so, each polygon in the original control cage that is at an extraordinary vertex will end up as a degree 6 Bezier patch meaning it has a grid of 7x7 control points, these are then combined with neighbors.

It could be possible to have an option to do a different style of patch combining where only the patches right around an extraordinary vertex get degree 6 and other areas get degree 3 patches with 4x4 control points. It would be fewer surface control points but also split up into more surfaces with smaller ones around extraordinary vertices.

It's kind of good to have a little lighter control cage, that one is kind of on the dense side.

However, despite the high control point count you should see very good shape quality with the surfaces all matching up very well.

Is it possible for you to send me the model at moi@moi3d.com so I can do a couple tests with it?

Thanks, - Michael

---

HI Michael,

Putting together a file for visual 3d pdf presentation - the file is 560meg - 2600 parts.
Moiv3 File takes 1.34 minutes to open
Moiv4 File takes 5.06 minutes to open
Rhino5 takes 0.50 minutes to open

I can organise for a shared link to file to run a load test - what do you think causes the wide variation in file handling time.
thanks
Alex.

---

Hi Alex, I have not yet come across any files that load slower in MoI v4 than in v3 so I don't really have any idea why that would be. Yes if you could please send me a shared link to it at moi@moi3d.com I can do some tests on it to see what is going on.

Also what are your system specs, do you have plenty of system RAM?

Thanks,
- Michael

---

Just playing around with some boolean union cubes, save to .obj, and perform SubD.
Different results are obtained with SubD upon the .obj files, depending upon the original location with respect to the origin,
of the original 5 cube objects .3dm.
- Brian
(Out of upload room, will clear space and post .obj later.

---

Hi Alex, re: large file loading - thanks for sending the file. Over here I get:

MoI V4 beta Oct-17 : 15 seconds
MoI v3: 70 seconds
Rhino v5: 27 seconds to load + 43 seconds "Creating meshes..." delay on first view navigation.

Additionally over here the V4 behavior is better all around than both MoI v3 and Rhino v5. On Both MoI v3 and Rhino v5 the main window is just frozen during the load while in MoI v4 the window doesn't freeze and there is a progress bar and cancel button.

My machine is fairly old, Intel Core i5-2500K Quad core CPU, 16 GB RAM, with an old style spinning disk.

I'd guess that your machine is probably a laptop? My theory is that with MoI v4 making heavy use of all CPU cores your machine doesn't like that sudden spike in energy use and it throttles the processor speed way down.

You can probably avoid this by making sure it is plugged in rather than running on battery power and then go to "Power Options" in the Windows Control Panel and make sure that the "High performance" power use plan is the one that is active. It's probably currently set to a plan that sacrifices performance for less battery use.

Does that make any difference?

Also I will put in a setting for the next v4 beta so you can turn off the multi-core processing for file loading.

- Michael

---

Yah, I'll stick to my retopo tools for now but hopefully think about adding an all quad exporter! It would probably broaden Moi's use and appeal.

---

Hi Michael,

Did a speed check on my desktop workstation and same time as yours with my laptop I always have it plugged in and with best perfomance setting. Will look into other settings to see if can improve load times.

thanks again for your time.

regards
Alex.

---

Hi Alex, maybe check the advanced properties of the "High performance" power plan. Maybe it still has some CPU throttling set on it. If not I guess it is just getting too hot with all CPU cores running and throttling anyway. I'll make a setting in the next v4 beta where you can limit the number of CPU cores used for file loading maybe using just 2 or 3 cores on that particular machine will work better if it just can't thermally handle all at once.

- Michael

---

Maybe he has a lower mesh angle setting on the one computer.???

Just a thought......

On another note, something i just noticed, but maybe it is normal, as i get the same thing in v3....

I think you called it "z fighting"...

So like in the 3d viewport, at some distance you can see edges if surfaces which are UNDER a surface, if you zoom in closer, the underlying edge goes away and you see the surface only.....

However, in the 2d views, the edge stays visible, even as close as you can get.... this didnt seem right, or at least i hadnt noticed it before.....

Comment?

---

Hi Burr, yeah unfortunately that's a pretty hard type of display artifact to avoid completely especially in a 3D perspective view. In a perspective view the distribution of depth values isn't linear like it is in a 2D viewport. It makes it difficult to deal with things drawn in separate passes and with somewhat different draw mechanisms like shaded surfaces vs curve drawing.

- Michael

---

Hi Michael,
Maybe you misunderstood?

My 3d viewport works ok... it is the 2d viewports that wont elimimate the underlying edge when zooming in....

Did you get that part?

Thanks for the responses...

---

Hi Burr, sorry no I didn't understand correctly. So it sounds like you're talking about this kind of display artifact discussed here:
http://moi3d.com/forum/index.php?webtag=MOI&msg=4987.2

A 3D perspective viewport has a little different kind of depth behavior than a 2D one, so it's not too surprising that you would see slightly different behavior with depth sensitive operations between them.

I do have some ideas that I'd like to try to experiment with for reducing that type of display artifact in the future.

- Michael

---

Thanks Michael!

---

Some results from saving cubes in MoI as .obj, and applying create subD:

Attached are the solid models 4Box01.3dm and 4Box02.3dm, created in MoI, derived from the Box tool, which appear to be identical, but are not. Saving as .obj files and then using create SubD results in different SubD models. There are at least two additional permutation models, which result in different .obj and subD results. Saving some .3dm and reloading them, then saving to .obj and applying SubD may result in different results than the pre-saved .3dm to .obj save.

Creating 4Box02.3dm:
1. Create a box.
2. Use circular array to create 4 boxes touching by their edges.
3. Change color and Separate the inner 4 faces, leaving the outer 20 faces joined into 4 open cubes.
4. Select the 4 open cubes, excluding selection of the inner 4 faces, and join.
5. Select the 20 face outer joined surface, then select the inner 4 faces one by one. (they can be selected in different orders), and join to form the 24 face solid.
6. Save as .obj.
7. Use create SubD on the .obj file to yield 2 or 3 subD objects.

Creating 4Box01.3dm:
1. Create a box.
2. Use circular array, or copy to create 4 boxes touching by their edges.
3. (May change color of the 4 inner faces.) Select all 4 cubes and Separate.
4. Select the 20 outer faces and join to a joined surface.
5. Select the outer 20 face joined surface, then select each of the 4 inner faces, one by one, and join to one solid. (Note that selecting the 4 inner faces by their color, or by their name, does not give the same result.)
6. Save as .obj.
7. Use create SubD on the .obj file to yield 1 subD object.

Using meshlab, it turns out that the 4Box solids are not manifolds, and that they have different non-manifold edges. The meshlab command:
Filters/ Cleaning&Repairing/ Remove Faces from non manifold edges,
shows the difference between the two .obj solids.

(Three boxes may be arranged around a central triangle with comparable results.)
- Brian
Deleted this old attachment, for space.

---

Show messages:  1-17  …  78-97  98-117  118-137  138-157  158-177  178-197  198-217  …  258-264