Draw Largest Rectangle Possible Within Space

[Anonymous]

Hey Guys,

I don't know if this can be done but I figured I would ask. I am looking for a lisp that will ask a user to select a closed polyline and then have it draw the largest rectangle possible within said closed shape. I need this new rectangle to also account for any existing line/layer/shape within said shape. Essentially I am looking to click the red layer in image 1 and then have the new rectangle pop up (layer black) in image 2 after it is done calculating everything and taking all other layers into account. I guess it's just a big if statement on setting the proposed rectangle to a set layer and then while drawing within the p-line it's set to avoid any intersection with any layer that is not the originally set one? I found some base code here https://forums.autodesk.com/t5/visual-lisp-autolisp-and-general/draw-largest-rectangle-within-a-limi... but what it doesn't work on my end and even if it did it doesn't account for other layers/shapes/lines that could be within said p-line. Any assistance is much appreciated!

[LyleHardin]

Sorry this reply will not be helpful, but I find this to be a fascinating exercise. I'm sure someone will come up with something clever. Also, does the rectangle have to be 'horizontal' ? A better fit may be a rotated rectangle. I wonder if Civil 3D's Generative Design feature and Dynamo may be useful for this situation. Won't do much good for plain AutoCAD, though.

[Anonymous]

Any rotation would work!

[Shneuph]

Interesting... and likely very difficult.  W/o having the dwg.. I wonder if this rectangle may actually have a larger area than the one you drew in?  Definitely longer but narrower.  I also won't be much help in the task of automating this task.  But I'm interested to see what people come up with.

[Anonymous]

No DWG needed you could literally just draw any p-line shape and drop some shapes inside, as long as the layers are all different than the rectangle being drawn. Also yes, my point exactly that rectangle probably is bigger (total sqr footage) but that's what I want the command to figure out and do. It will see ideal spots for the largest rectangle that are harder to spot.

[LyleHardin]

The more I look at this, the more I think Generative Design is the best solution. (Not the only one, of course.)
It will draw thousands of rectangles based on certain constraints and find the one with the maximum area for you.
Alas, I've not done any Generative Design programming, so I'm of no help there. Sorry.

[hak_vz]

@Anonymous 

In post that you mention, you may find lisp codes from @CodeDing @marko_ribar and my own try on the subject. From the length of that post you can guess I took us a lot of time and effort to make something that really works. Code was made in accordance to @hs800150 request. At the end solution from @marko_ribar was rightfully selected as final, since he has spent a lot of time bulletproofing  his code. All posted solutions produced similar results regarding largest inscribed rectangle and OP request, and all deserved to be selected as a solution.

Adding test for collisions with internal object is really complex task. To use above mentioned lisps create a single  boundary that joins internal obstacles and you will for sure get largest inscribing rectangle.

I remember that this was one of really interesting requests and I enjoyed working on it, but I probably won't participate in writing more code to fulfill  your request. For a start try my proposal with boundary.

[Anonymous]

Very true. I figured it would be very complex. How can I get the referenced code to work though? Even just that part I try it and it just gives me a bundle of scribbles within the pline.

[hak_vz]

Based on your sample images your request is really hard to solve. Method that I described in my previous post would work for aligned rectangle and probably give correct answer. Adding rotated rectangle into it really rises level of complexity.

[Anonymous]

I'm trying to get your old code to work. I want to work off it but I can't get the base command to work properly. Below is a random closed p-line I drew and then I used your "ripl" command and got the following result. Same kind of thing as last time just a jumble.

[hak_vz]

---

@Anonymous wrote:

I'm trying to get your old code to work. I want to work off it but I can't get the base command to work properly

---

Download code at this link  (post no 37) and use command maxalignedrect. It should work correctly for closed lwpolyline that may contain arc sements (not as aligning edge)

[jadranko]

And criterion for "the largest rectangle possible" is the area, right?