2018 Booleans - Lots of problems.

2018 Booleans - Lots of problems.

revel68
Advocate Advocate
6,743 Views
30 Replies
Message 1 of 31

2018 Booleans - Lots of problems.

revel68
Advocate
Advocate

Is anybody out there using the new booleans?

They seem highly problematic across all kinds of stuff i'm trying to create today.   I have extensive max experience so I know how to try different things to try to get it to work. Everything i'm doing is just brutal and broken.  I even sometimes get Invalid UV errors (that are unfixable even with all the little uvmap/edit poly tricks) on some things I create. This error comes up during rendering.

 

I would like to submit bugs but it just seems like the entire thing is bugged. 

Right now I'm trying to take 2 very basic legit models and Union them and it refused to.  All of the operations are not acting properly but I don't know how to make it any simpler of a case.  2 Oil tanks edited and collapsed and reset transform and they won't union properly. And i'm only doing this because when i do both of them as Subtract in layers on my base object one of them refuses to actually subtract.  So i'm trying to merge and then subtract.  

 

This seems unworkable. Am I the only one?

 

0 Likes
Accepted solutions (2)
6,744 Views
30 Replies
Replies (30)
Message 2 of 31

revel68
Advocate
Advocate

See attached gif.

 

Both objects Editable Poly. Both Reset X-Form and face normals going the proper direction.

Both are not "shells". They are water tight simple quaded objects.

Both within reasonable distance and scale from origin.

 

I even tried at far lower triangle count.  Where these  objects are only like 400 triangles each with very very simple basic geo with no verts close to each other. 

 

I can't get these 2 to boolean together. Not just Union but all of the other ops have issues and basically make the entire model "disappear". It's still selected but you see nothing.  

 

This object was started with a very common primitive "OilTank"  Could that be the issue!?

 

 

0 Likes
Message 3 of 31

Anonymous
Not applicable

I tried to replicate your issue with similar geometry and the highres-version failed like in your example.

Then i slightly changed the position of one operand and it started to give correct results. The Lowpoly-version always worked.ProBoolean worked on booth highres and lowres. Seems that you hit a boundry condition that causes the Boolean Operation to fail. This is to be expected with polygon boolean operations.

As for ProBoolean vs. Boolean: I had cases where Boolean worked and ProBoolen failed and vice versa.So you should try booth.

Also if possible, i would perform Boolean Operations on low poly meshes and smooth after.

You can also report your findings to Autodesk by providing a file that replicates the issue.

 

0 Likes
Message 4 of 31

revel68
Advocate
Advocate

Cool. Low poly version failed.  I wonder if its possible to give somebody a file with the objects in it and see.  Contact to Autodesk seems slim. May be bug reports with an attachment or something. I have a few here now.

 

For now i'm doing the booleans in zbrush.

 

 

 

0 Likes
Message 5 of 31

Anonymous
Not applicable

@revel68 wrote:

Cool. Low poly version failed.  I wonder if its possible to give somebody a file with the objects in it and see.  Contact to Autodesk seems slim. May be bug reports with an attachment or something. I have a few here now.

 

For now i'm doing the booleans in zbrush.

 

 

 


Post the file and i will take a look into it.

0 Likes
Message 6 of 31

revel68
Advocate
Advocate

Thanks!

So this is interesting. I took the file home to set it up for you.  I tried  the 2 booleans i wanted to do and was having issues with and they are now working! 

The only real difference I would know of is Windows 7 vs 10 and 3dsmax 2018.3 vs 2018.4

 

The X shaped one worked at the very low poly level and with 2 turbosmooth divisions. But did not with 4.   At work none of those worked, even the base low poly ones.

 

Although I did notice during this that my entire Utilities panel here at home is just a grey box -_-. But that is for a different post/bug.......

 

I attached the file anyways incase you or anybody wants to look at it.  I tried this exact thing a few times at work and it looked bad. But now looks fine at home.  

 

 

 

 

0 Likes
Message 7 of 31

Anonymous
Not applicable
Accepted solution

Yes, it indeed Boolean fails with more than 4 subdivisions. ProBoolean worked fine however.Also, if you are below the 2019 release you can probably forget about Boolean anyway, because it produces tri-faces on non-planar polygons.

Only 2019 and up there`s an option to 'Retain Non-Planar Faces'

Boolean.JPG

Message 8 of 31

revel68
Advocate
Advocate

Thanks. Yes I noticed the triangles showing up on a lot of the faces :/.

 

Wait did you say "above 4"? Meaning it worked at 4?  Mine didn't work at 4 but it did a 2 and the base.  EDIT: Actually looking at your picture it looks like it didn't at 4 either. 

 

That is good about proboolean. I'll remember that. But i'm really trying to use the new system and i'm trying to give feedback to Autodesk.  I wanted to do the editable dynamic system with stacks on the booleans (as instances) and all that fancy stuff. So I can turn up and down turbosmooths  or move things dynamically.    I'm excited to have a good boolean system for 3dsMax.  But I do run in to issues. When it first came out I had videos of problems that i never really sent to Autodesk. Old news now since it was probably like Max 2017.

 

The system is also desirable because at work it means I can change things in front of a lead/art director or turn stuff on/off easily.  

 

Thanks for the heads up on 2019.  I may look in to that eventually but I try to stick with a Max version for a while because each time i move its a big deal and have to vet the program for work. FBX/Normal map tests.  Scripts.  You name it.  I felt lucky I  got 2018 working for me and my workflow.  I always keep my 2014 setup in my back pocket.  I bet i'l be waiting for a 2020 or what have you before I even try.

 

 

Message 9 of 31

hagen.deloss
Alumni
Alumni

Hi @revel68

 

Thanks for posting this issue on the forums, it looks like you and @Anonymous were able to work through this issue and figure out how to merge your oil tankers together? From my understanding, the boolean system had a little trouble with anything turbosmooth iteration above 4? 

 

I can help troubleshoot if there are any further issues!

 

 



Hagen Deloss
Community Manager | Media & Entertainment
Installation & Licensing forums | Contact product support | Autodesk AREA


 

0 Likes
Message 10 of 31

Anonymous
Not applicable

@hagen.deloss@revel68

Well, apparently it seems to work with 5 iterations, but fails with 4 iterations?

 

So testing with the scene @revel68 provided:

1 iteration: OK

2 iterations: OK

3 iterations: Fails

4: iterations: meshes disappear

5: iterations.OK

 

boolNext.JPG

0 Likes
Message 11 of 31

hagen.deloss
Alumni
Alumni

@Anonymous

 

That is wild! Hmmm, and to be clear, is this with both the boolean and proboolean tools? Or is this just the boolean tool?

 

I tested this on my machine and while it took a while to process, a turbosmooth of 1-5 all worked with my boolean operation on 2 oil tanks. Were there any other modifiers I forgot to apply? Here is my mesh and modifier stack.

Screenshot (138).png

 

If we aren't able to figure out why this issue is happening, I will be taking all this information back to development so they can take a look.

 

 



Hagen Deloss
Community Manager | Media & Entertainment
Installation & Licensing forums | Contact product support | Autodesk AREA


 

0 Likes
Message 12 of 31

Anonymous
Not applicable

@hagen.deloss wrote:

@aon.914858227

 

That is wild! Hmmm, and to be clear, is this with both the boolean and proboolean tools? Or is this just the boolean tool?

 

I tested this on my machine and while it took a while to process, a turbosmooth of 1-5 all worked with my boolean operation on 2 oil tanks. Were there any other modifiers I forgot to apply? Here is my mesh and modifier stack.

Screenshot (138).png

 

If we aren't able to figure out why this issue is happening, I will be taking all this information back to development so they can take a look.

 


It works perfectly  as long as the operands aren`t properly aligned to each other. Can`t get it to work with 5 iterations anymore.I recorded a video. Depending on the mood of the board software it might show up.

 

 

 

Message 13 of 31

revel68
Advocate
Advocate

Could it be that i'm on 2018.4 (and .3) .  

One thing you could try is doing 2 turbo smooths. Place one on top and set it to 2 divisions and then check the Smoothing Groups box. Then put another turbo smooth on top and set that to 2.  Does that break it or is it still fine for you?

 

Thanks!

 

Also I'm only using regular Booleans. I want to try to have an editable situation where I can change things on the fly later or under criticism.  Every once in a while I check back in to the new booleans. This time I was 3 for 3 on failing to do what I wanted.  Though I think one of the problems was caused by an IToo tool called "Clone" that made it impossible to render a scene (scanline) after boolean.  I am not going near Clone anymore. That thing seems to cause big issues with meshes.    

Message 14 of 31

hagen.deloss
Alumni
Alumni

Hi @revel68 and @Anonymous

 

I haven't had any issue with my booleans at 5 iterations, besides just a little slow down while it processes the command, not to be dismissive, I really want to help out hahah! 

 

It could be a combination of processing power and 2 dense meshes combining, have either of you checked out your processors allocation while the boolean is being calculated?  My CPU spikes a little while the process is being run (from 1% to 4-6%).

 

How vital to your workflow is working with an object with 5 iterations of turbosmooth on it? I would suggest working with 3 iterations if you can. Testing out proboolean might be good, though it doesn't create as much extra geometry.

 

What do you think @Anonymous I can log this into our system if you think this is an issue? 

 

@revel68 ahhh I have been getting more railclone and forestpack issues come up with itoo stuff, have you interacted with their support/forums at all?

 

Looking forward to hearing what you both think about this 😄

 

 



Hagen Deloss
Community Manager | Media & Entertainment
Installation & Licensing forums | Contact product support | Autodesk AREA


 

0 Likes
Message 15 of 31

Anonymous
Not applicable

@hagen.deloss wrote:

Hi @revel68 and @Anonymous

 

I haven't had any issue with my booleans at 5 iterations, besides just a little slow down while it processes the command, not to be dismissive, I really want to help out hahah! 

 

It could be a combination of processing power and 2 dense meshes combining, have either of you checked out your processors allocation while the boolean is being calculated?  My CPU spikes a little while the process is being run (from 1% to 4-6%).

 

How vital to your workflow is working with an object with 5 iterations of turbosmooth on it? I would suggest working with 3 iterations if you can. Testing out proboolean might be good, though it doesn't create as much extra geometry.

 

What do you think @Anonymous I can log this into our system if you think this is an issue? 

 

@revel68 ahhh I have been getting more railclone and forestpack issues come up with itoo stuff, have you interacted with their support/forums at all?

 

Looking forward to hearing what you both think about this 😄

 


All i can say is that the new Boolean compound is definitely failing with the scene that @revel68 has provided. If you use Turbosmooth with more than 3 iterations on the operands. If i build a similar shape from scratch in a new scene it works.Here with 5 Iterations (Took 3sec to process on my machine with 12% CPU usage).

cont.JPG

 

So it might be something with this specific geometry that causes the operation to fail. However, the thing is that ProBoolean was working well on this geometry and i think we should expect the new Boolean to work as good as the old one? @hagen.deloss i would log this issue if you are able to reproduce this behaviour with the provided scene file.

Perhaps @revel68  can you provide additional scenes, since you said the new Boolean is constantly failing with your objects for Hagen to investigate?

 

 

 

 

 

Message 16 of 31

revel68
Advocate
Advocate

Hagen. To be clear it was failing at all levels at work when i first made it. All levels.

 

The desired work included a turbosmooth at 2 divisions set to Smoothing Groups and a turbosmooth set to 2. on top of that.  Then I would get a clean piec ewhere i'm not worrying about clean up or triangles or whatever.  Then that piece was to get cut out of a block on another model.

 

I was having trouble at any level. Even the 2 base models were clipping with each other.  When I brought the file home it started to work better for some reason even though the version of 2018 is very similar (.4 vs .3). I explain all this in my earlier messages.  

 

So yeah I've seen problems on low and mid poly levels.   The lower poly you go the more crappy results can be and require hand-work. Hand work means I should just model it by hand. Or use another program. 

 

I am back at work now so I can easily show it breaking .  I attached an image showing the base object with Union in the new Boolean system. BTW this is the 2018.4 one.   You can see that the object is clipping through itself and also "Cut".  

0 Likes
Message 17 of 31

revel68
Advocate
Advocate

Just showing this.  This was one of of the ones i ran in to before the one we are talking about above.  I did this boolean and thought i got it to work and then realized that one of the parts didn't "take". 

I attached an image to this post. you can see that there are all these cuts. They are great.  But there is one that just didn't work even though its an array of the same object working over a cylinder type object.  

 

I'm guessing people will say i'm working too high poly but much lower than this and the object will require a lot to be at the quality bar and what i need for our process. So i'd either model it out or use a different tool.

 

in this case I slapped a Symmetry modifier on the model to fix it 🙂

 

 

Message 18 of 31

hagen.deloss
Alumni
Alumni
Accepted solution

Hi @revel68 @Anonymous

 

I was able to reproduce a similar issue with the combination of turbosmooth and boolean, and my mesh does some pretty funky stuff, but this isn't surprising, Boolean has always had limits with how many poly's it can cleanly merge.

 

It sounds like your desired workflow is fighting the tools. While boolean is a powerful tool to get some cool cut outs in your geometry, i think you are finding the limit to the non-destructive workflow with high polygon meshes. I would recommend duplicating different versions of your models to show your art director. 

 

Thanks for bringing this to the forums, and I'm excited to see more of your work in the future! If you ever have anything to showcase, feel free to post in the 3DS Max showcase thread, and if you have any ideas that you think should be added into 3DS Max, you can post them on the ideas board.

 

 



Hagen Deloss
Community Manager | Media & Entertainment
Installation & Licensing forums | Contact product support | Autodesk AREA


 

0 Likes
Message 19 of 31

Anonymous
Not applicable

Far from the only one.  I had the misfortune of using boolean earlier today to see that the new user features include messing up material id's and causing max to crash when one attempts to detach the incorrectly id's objects and apply the correct material id's.  This was not like this in 2016.  The geometry merging of booleans has always been sketchy at best but with these new *user features* I won't be using it ever again.  Yet another thing that worked better in the past then it does today.  Can't wait to see what changes await for 2019!

0 Likes
Message 20 of 31

Anonymous
Not applicable

"It sounds like your desired workflow is fighting the tools." -

 

And comments like this I think reflect the overall attitude of the Autodesk team.

 

It looks like your desired workflow is to avoid any work.  Instead just blame users who take time to report issues with your poorly maintained tools rather the acknowledge and fix such issues.  Typical.