search

booya-at / OpenGlider

python library to build paragliders
GNU General Public License v3.0
65 stars 25 forks source link

Total line length #66

Closed Mx74 closed 2 years ago

Mx74 commented 2 years ago

Hello,

For kite design, that would be convenient to be able to put the pilot point further down than 10m. Is there a technical reason for limiting this distance to 10m? If not, removing the limit would be a plus. Mx

looooo commented 2 years ago

thanks for reporting: https://github.com/booya-at/OpenGlider/commit/1ea23e47afde56b63dd8ba4769644a43a915db2c

Mx74 commented 2 years ago

Yep, the 10m limitation is gone. It somehow works; but adding a fifth level of line (high, mid, low, mixer + long lines) mess up the line calculation.

looooo commented 2 years ago

normally there shouldn't be a limitation. can you send an example?

Mx74 commented 2 years ago

Indeed, this is not only a number of level problem, but of of lenght. If I set my mixer too short, then some weird things happen. WIth 4 levels, setting the risers at 40cm isn't a problem.

But with 5 levels, this is what happens: Mixer (equivalent of risers) at 1m or 90cm, everything is fine Capture-20211216185514-480x810

Mixer at 30cm or 50cm: weird angle Capture-20211216185614-443x747

looooo commented 2 years ago

oh I see this is a problem of overcompensating the residual forces. You can try to set the factor at the end (0.5) to a smaller value to reduce the compensation per step. Some more explanation is inside the code. https://github.com/booya-at/OpenGlider/blob/develop/openglider/lines/lineset.py#L330L334

Mx74 commented 2 years ago

Perfect, thanx a lot. Made a simple trial with 0.3, and it looks better. I'll investigate more for finding the right parameter.

Mx74 commented 2 years ago

Just a thought, as it's hard coded, this change will disappear at every update. Not a big deal when you know, but for people not aware of it, would it be possible to integrate it as a parameter in the gui?

looooo commented 2 years ago

It's not the best solution anyway. I guess we can store a value for every line and look if the residual gets smaller for every run. If the residual gets bigger choose a smaller value. But I guess there is also the option to minimize the whole system at once.

Mx74 commented 2 years ago

Yep, and just changing the parameter isn't as easy as it looks. Indeed, I did a rough test without really taking care of the final result (you can see really long upper lines for example), but then trying to refine the model to get coherent lenght for the upper lines seems to be complex.

looooo commented 2 years ago

what do you mean with coherent?

Mx74 commented 2 years ago

realistic?

looooo commented 2 years ago

ah ok, yes you are right the process is quite time-intensive, I guess changing the line-length on the 3d model would be more elegant.

Mx74 commented 2 years ago

Yep, it would just have been convenient to ge tthe total line ength, in order to avoid guessing the space between the 2 sides. But that's easy to work around. Thanks