Pancho

Logged as a bug.

The only difference I can achieve in the solve is to change the rest angles mode. No rest angle = Screws instantly up into the bridge/curve like shape. The other two don't seem to make a difference. : (

Thanks so much for your scene file. Indeed in your example it works. In mine it did not, until I rotated my chain to be aligned with the z-axis. Then it worked too. Not sure how to solve this. Probably related to the root or local transforms?

I wanbt to add that I'm constantly getting python viewer state event callback errors as soon as I touch a kinefx node. Let's say in 50% of my interactions. I click them away ("ignore future errors"), but they keep on coming back. Not sure how to deal with that and if is part of the problem.

I'll immediately update and see if it makes a difference. I'm on 19.0.498 here. Didn't think about it to be a possible bug. Actually it's annoying that they didn't make the sop solver fit for the task to solve multiple joints. Will look at your file asap. Thank you very much so far!

Hi, I try to bend my head around the kinefx ik solvers. I'd be glad if the "ik chains" would solve more than a regular leg or arm, since I have more than just an elbow joint. But it doesn't. So there's either the "solveik" in the rig vop left and the "fbik". Unfortunately the solveik doesn't seem to respect my initial transforms and comes up with something which looks like a curve solver, but a twist vector is present in the solver. While the fbik seemed to be promising in the beginning, it easily ends up totally garbled. Initial joint transforms don't seem to matter for the solve and the perfectly aligned joints move into totally wired positions. Twist vector is missing. Bit unpleasant situation. Put the three possibilities into the attached file. Is there anybody who found an solution for this problem? How can I tweak the "solveik" that it respects the neutral joint positions/rotations and/or is there a way to have an up vector for "fbik"? A multiple joint chain should be so difficult to solve in 2022. At least not in Houdini... Cheers Tom ik samples.hiplc

Write a python script to collect all the parms of the selected nodes you like to display. Display the formated parameters in the console and copy them into the font node. p.label() is the name you want to display, p.name() the name of the parameter which you need to connect to the node path, like "/obj/geo/" + "xrows".

Very nice!

Sounds like a good project for getting into lidar hardware. For this purpose a rotating one might be nice, but one which could be carried around is probably even better.

Ah, don'T build one. THe ones I've seen are more a gadget, than a lidar scanner. The two from DIJ are between 800-1500€ I think. ONe seems to be a single scanner, the other a triple one. Both are not rotating, but scanning in one direction. Wonder whether the speed of a car at 30-50km/h would still deliver nice results. Scanning distance is something like 40m max as far as I can remember.

Cool!!!

Wow! I'll check the wip! Did you capture the lidar yourself? Need to capture something like a long road. Thought to mount a lidar like the one from DIJ on a bike, but not sure whether this will work. Tried photogrammetry. While Nuke is able to retrieve a pass the point clouds are a mistery. Agisoft on the other hand comes up with great points clouds, but looses the spatial sequence, so instead of one long passage, you get a wired mix of small chunks. Still looking for a solution. Something which works as Nuke could/should work.

I just used UE4 as a render engine for some animations. All this VAT to export animations is a pain. Houdini engine should take care of this. The videos about the LABS tool pretend that it is simple, but it's like "Pitfall", the C64 game. I guess I don't need to get into details. The nurse was pretty busy whiping away the foam from my mouth for hours, days, weeks? : ) The pain with every new package is not doing some stuff, but to get the philosophy of how things are done. In H I don't care about a few 1.000k, but in Unreal? So just the get info whether it's bessed to export spline, geo or whatever is difficult to get. That's almost as terrible as converting a receiving high res DEM models! : ) Cheers!

Thanks for the reply. Funnily I just started to look at culling after it was menitonend in the "From ant to space" (or something) video from Unreal. THe guy presented it in combination with world space displacement for flowers. The further away from the camera, the lower the instances would be placed. If they are beneath ground some form of culling lets them disappear. I guess cutting the street into pieces would be the way to go. Probably the heightfield could still carry the color information of the street even if the geo isn't there anymore. In combination with LODs this could work nicely. Will give it a try asap! Are you familiar with UE4?

Ha,ha! THanks for the tip!!! : ) Watched it already. But I was a bit disappointed when he just instanced the rails tracks with visible gaps. : / Hoped that I would find the secret in this video. Unfortunately I can't find anything on the web so far covering this topic in regards to what splits my brain. I just can't tell which way I should bring in the geo so that it suits the needs of a game engine. Probably I could just generate the road mesh in Houdini which would be a no brainer. But since the street is long, how would I prepare the geo in order to just load parts visible to the camera or reduce it's LOD in the distance or do something else to it that it is UE4 "safe"? Bringing in a heightfield is one thing, bringing in the instances another, but the roads... ???