CompuServe Thread

#Boolean in Editor

10 messages in this thread
#137105From: Scott WomackNov 20, 1994 8:16 AM
I often get an error message when I am tryingto create some boolean operations. It usually occurs when I am trying to cut a 3D object in half. It tells me to "adjust the geometry". In just what way am I supposed to adjust/change the objects to correctly make the operation work? -Scott
#137110From: Jonas Ruikis [ADESK]Nov 20, 1994 9:52 AM
Hi Scott, << . In just what way am I supposed to adjust/change the objects to correctly make the operation work? -Scott >> object/tesselate is a good one to try first.
#137126From: Yost GroupNov 20, 1994 10:50 AM
Check out page 7-55 of the reference manual. – G
#137190From: EDWARD HARVEYNov 20, 1994 11:00 PM
Ive had hit and miss experiences with boolians. Today I had some luck with intersect in order to perform a subtraction that wouldnt work with subtract. Must be an art rather than a science. Ed
#137331From: Scott WomackNov 21, 1994 5:52 PM
Thanks Gary, I'll do that. Just as the attached reply says, I have had to use intersection to get subtraction, etc. at times. -Scott
#137553From: John K. JordanNov 22, 1994 10:15 PM
Scott, >>boolean, "adjust the geometry" I am sorry to be so late in answering this, but… When unable to complete a boolean there is one other VERY IMPORTANT (in my opinion, anyway <g>) thing to do that I've haven't seen written down – or if I have, my poor brain has forgotten where: *** Don't try to boolean an object which is composed of two or more elements which overlap in the area of the boolean. Well, go ahead and try, but it may not work. <g> If you manually overlap several objects then attach them, you will may well get the dreaded "adjust the geometry" message. For example, consider an object made of three overlapping sphere elements, thru the middle of which you wish to bore a square hole by subtracting a simple box. For me, this fails every time. The solution: separate the object back into three spheres using Create/Element/Detach, make two copies of the box object, then subtract one copy from each sphere. Finally, reattach the three resultant objects. For me, this works every time. JKJ
#137853From: John RedfernNov 25, 1994 2:31 AM
JKJ, >> *** Don't try to boolean an object which is composed of two or more elements which overlap in the area of the boolean. Well, go ahead and try, but it may not work. <g> >> Absolutely right. I think it also applies to multiple booleans as well as attached elements. I just did a bool with one central object and two mirrored objects left and right. Left and right needed to smooth together below the centre so I made them overlap slightly. Left bool'd with centre ok but right would not bool with the new object. Move left and right apart – just more than the weld threshold – no problem but you have to manual weld. I notice that if you subtract a hole through a multiple bool'd object, even if it works you can get a lot of stray vertices left around. Another thing is that bool'd objects can have faces that are d****d difficult to smooth. Ed asks, "Art or science?" Both, and I too have noticed you sometimes have to invert the commands to get what you want! Keeps us awake. John
#138109From: Scott WomackNov 27, 1994 4:41 PM
Thanks guys! The oject I am having problems with was made by attaching 3 different objects! When I separated them again, the boolean worked just fine. Thanks for the info! -Scott
#138166From: John K. JordanNov 27, 1994 11:10 PM
Glad it helped, Scott! In the case of a boolean failure on multiple overlapping elements, there is one other thing you can do that I forgot to mention: Instead of building a composite object by _attaching_, combine the components using the boolean Union. This will leave the object with a single surface instead of overlapping elements surfaces. A subsequent subtraction will then have no problem. (This method is actually less work than what I described earlier – sorry I forgot to mention it) I love the boolean improvements the Yosters put into R3 – MUCH improved from R2. And that's no bool. <g> JKJ
#138378From: Scott WomackNov 28, 1994 5:28 PM
Thanks for the additional info. I'll give it a try once the current project is done rendering in a day or so. -Scott