Jump to content

nli2work

Members
  • Posts

    2,836
  • Joined

  • Last visited

Everything posted by nli2work

  1. it's possible, just not a single collider. you have to approximate the inside with multiple colliders.
  2. no collider on GrappleNode. it should be an empty transform with Z pointing at the grab target. the part does need collider. I was thinking of this one.
  3. The Klaw can't be the Root part. there's minor bug in the way force is applied when you detach the claw such that if the Klaw is the root part of the vessel, you get NullRef and IndexOutofBounds error (you should see it in the debug log in game) and it'll screw up the game and you get symptoms like you described and the Klaw will not attach to anything else. Padishar logged the bug in the Tracker, but no telling when it'll be fixed.
  4. Yeah. once the base coats are dry, slather watered down browns and yellows and let it collect in the crevices and joints. Qtips are handy to sop up the wash that runs where you don't want. start light and layer more to get it darker. then apply fairly thick base color paint over the wash in light passes to let the paint build up some texture. superglue + fine sand are good to build up heavy textures. once dry you can paint/wash over that like anything else.
  5. Done! About 7 weeks of night work, I'm happy with it!
  6. it's based on offset from the root position. has nothing to do with collider. node_attach = Offset from Root X, Y, Z, Direction X, Y, Z, node Size (not used in surface attach). e.g. node_attach = 0,0,0,1,0,0,0 will attach from the part's root; with the root's +X pointed at whatever it is attaching to.
  7. might help to have some mesh overlap where you are getting leaks. Edges that meet up exactly tend to leak light because shadow map resolution isn't high enough.
  8. I'm not so sure about this since 1.1.x to be honest.
  9. You mean the piloted pod right? just tested it in 1.1.3 it works as expected. Flight computer icon is red as there's no signal processor unit, but it shouldn't require connection to KSC. helper drone has fairly wide capture angle, ~ +/- 30 degrees. Testing in 1.1.3... I don't expect any issues as the updates in 1.1.3 doesn't seem to affect any part modules relevant to the parts. If you encounter any issues please let me know.
  10. I see, so the source mesh is messed up... in that case it'd be better to fix it in Max before exporting. I don't know which version you are working with, but with 2015 there's occasional issues with FBX export, I add an STL check modifier before export, that usually keeps things in order. Importing tangents is important if you are baking Normals from Max as the pixel values are calculated based on the vertex tangents. When you calculate tangent in Unity, you are likely to get different Tangents on some vertices and you end up with display artifacts after applying the normal map. If you are just creating Normal maps from bump maps, using Unity to calculate tangets not much of an issue.
  11. something with unity set up. remove mesh filter and mesh renderer components from collider object. The collider object should only have Collider component in most intances: Box/Sphere/Capsule/Mesh. Mesh collider should be marked Convex. For this particular part, it's best to create a low res mesh for collider object.
  12. it should be the grayscale if it's converted to DDS. R is put copied to G and B; G is put into Alpha. this minimizes DTX compression artifacts in Normal maps. Mirroring texture can work for some texture layouts, but limits your option if you need tiled areas like tire tracks. This is set in the Texture asset's wrap mode.
  13. wasn't really because UV patch being inverted... but probably because triagulation isn't mirrored. so try mirroring the triagulation along the seams and see if things improve. Here's better explaination http://wiki.polycount.com/wiki/Normal_Map_Modeling#Mirroring
  14. not that I know of. you can flip the UV patch, but it would mess up the round attachment area.
  15. It's due to the mirrored half being inverted in UV space... their W is pointed in the opposite direction as the half you initially unwrapped before mirroring. I would rearrange the UV such that the stacked patches cover a whole repeated section instead of half of one as you have it now. each section would include the whole round attachment cut-out plus half of the separation panels on each side. the black/white markers will be tiled as well if you include them in the main mesh/texture. if they have to asymmetrical I'd split them into separate mesh object that sample main texture. something like this, assuming the upper area in the lower right isn't taken up by anything. You would have 4 UV sections stacked together, that covers the cylindrical shape. then the endcaps and other bits
  16. first make sure the Normal Map is marked as Normal Map Texture Type in Unity; then try different Tangent setting in the mesh import. https://docs.unity3d.com/Manual/FBXImporter-Model.html inverting mesh normals has to be done in 3d app before export. your screencap doesn't look like flipped normals, those usually look "transparent"
  17. TsubSys range from 0 to 1; 1 being the end of deploy anim, 0 the start. it controls when the wheelCollider gets turned on. with 1, wheelCollider pops in under the full weight of craft, that may momentarily overstress the leg. Stock LT-2 turns the wheelColliders on very early in the deploy animation, so much so you can see the wheelColliders levitating the craft before the feet even touches the ground. https://www.youtube.com/watch?v=CXU5pnMFoeA I'd try a TsubSys value of ~0.2, or even lower, I suspect this is the setting that's most relevant... and/or increase the suspensionDistance to something high to make sure it clears the piston_collider. then gradually lower it till the leg touches ground. All that said, it maybe easier just to wait and see what the next update does to wheel modules... it might make all this work for naught.
  18. I don't know the specifics unfortunately. :/
  19. if you are able to get cubemap frame sequence from a camera in KSP, the rest is just a matter of converting cubemap to lat/long map. then convert the frame sequence to a video, inject the youtube metadata for 360 video and upload. Youtube only takes lat/long probably because that format is more common with cameras. http://alexcpeterson.com/2015/08/25/converting-a-cube-map-to-a-sphericalequirectangular-map/
  20. I figure you'd just need to save out the Unity final frame buffer as a cubemap, or convert to equirectangular before saving. conversion to video could be done as a unity/ksp plugin; or in some standard video editor. https://docs.unity3d.com/ScriptReference/Camera.RenderToCubemap.html; I'd check TRReplacer or Reflection Shader plugin for code references too.
  21. child a non-collision geo to the wheel collider so you can see what it's doing during deployment. I usually do a disc with a sphere or something so I can tell the orientation as well. it should move with the deployTgt transform. if it isn't. try setting deployTgt to default layer instead of wheelColliders also try setting TsubSys to something lower than 1; it maybe over stressing from the wheelCollider popping at the very end of the animation
  22. in flight? or in VAB/SPH? it's normal in VAB/SPH. if you want normal playback speed in VAB/SPH, set instantAnimInEditor = false in ModuleAnimateGeneric
  23. A big sigh of relief when I opened the box, very happy with the result and glad I didn't waste several hundred dollars. 6x8" Intuos3 and microSD card for scale. http://i.imgur.com/JngUQZU.jpg http://i.imgur.com/aDIcESs.jpg http://i.imgur.com/3rOpJY6.jpg http://i.imgur.com/FP8Py5a.jpg
  24. what does F3 log show when the legs explode? I'd try removing clipObject = piston_collider in ModuleWheelBase; clipObject is destroyed after vessel unpack. I think it's more of a debugging tool or something to provide visual guide while in editor mode to prevent wheel block. Since it's also referenced in suspensionModule, that may screw things up when it's gone. Edit: looking at the stock LT-1 config... that's in there too. so I don't know. Some of of your values are different. maybe try using the same exact values? if things work out, adjust from there.
×
×
  • Create New...