[WIP][1.8.x] SSTULabs - Low Part Count Solutions (Orbiters, Landers, Lifters) - Dev Thread [11-18-18]

[CobaltWolf]

2 minutes ago, Shadowmage said:

Thanks for the offer; might just take you up on that some day.   I'm sure your model will be much more usable than the raw NASA one

Sadly it looks like it may be a few months at least before I get back to working on lander-oriented stuff; have tons to finish up + a few other projects I've been putting off for awhile.

Sure thing! In any case, working from a 3D reference - NASA's or mine - is very helpful for understanding a complex shape like the LEM.

[Shadowmage]

2 minutes ago, tater said:

Yeah, I know. I was using stock only rockets, as well as stock plus SSTU for testing (though mostly SSTU). I tend to ignore the stock parts in real play, aside from the mk1 pod which as the stock pods go is OK, it's decently Mercury-like.

Remember that tier 4 is only 45 science, however, which is trivial.

What about the SSTU probe core happening very early but only 0.625 diameter and then ALSO the 0.625m tank. Min tank with nosecone+probe core+tank... all we need is a tiny engine, and the first craft is an SSTU sounding rocket!

Hmm....  I will definitely have to give that a bit of thought;  sounds like it might be a workable alternative to the stock career start.  I have no problem with a sounding-rocket start to the space program, and would actually prefer to get away from the stock 'everything must be manned!' mentality at the start of the game.

However that shifts the problem to -- what existing engine(s) would be appropriate for a 0.625m rocket?  (keep in mind that I don't currently have time to design/model/texture new engines; I could make them 'next' on the todo list, but that still places them out towards the end of Dec/Jan; though perhaps a temporary placeholder-rescaling might be workable)

[tater]

Or a 0.625 solid rocket unlock. The tank+nosecone would just be for aero/looks.

[tater]

Large super draco, but with half the fuel unloaded so it can actually launch. Goes ~40km. Nose is min tank with nosecone. 8 parts, 4 of which are stock fins.

 

Solid version. 9 parts, 4 fins, plus I added reaction wheel. Forgot I could gimbal solid, so it should be 8. Made it (just) to space as an IRBM.

 

Edited November 8, 2016 by tater

[riocrokite]

21 hours ago, Shadowmage said:

assymetrical centrifuge concept stuff

dunno if that helps but here's a simple concept I made some time ago, 2x 1.25m modules on a 150m tether - very long radius allows for low rotation speed. I guess the diameter would be too much for ksp and you would need to animate lifts on tethers to accomodate for transport of kerbals between modules and the center.

animation with speed increased by 2.5x :

 

[tater]

35 minutes ago, Shadowmage said:

Hmm....  I will definitely have to give that a bit of thought;  sounds like it might be a workable alternative to the stock career start.  I have no problem with a sounding-rocket start to the space program, and would actually prefer to get away from the stock 'everything must be manned!' mentality at the start of the game.

However that shifts the problem to -- what existing engine(s) would be appropriate for a 0.625m rocket?  (keep in mind that I don't currently have time to design/model/texture new engines; I could make them 'next' on the todo list, but that still places them out towards the end of Dec/Jan; though perhaps a temporary placeholder-rescaling might be workable)

So the solid rocket version is pretty effective, and better yet, it's already done.

I'd be inclined to unlock the least capable SRB at 0.625 from the start, along with perhaps probe core at the same diameter, and the MFT-A also at 0.625 (people can configure it as EC for a sounding rocket).

[Shadowmage]

Have opened a new issue ticket to keep track of the tech-tree / career suggestions:  https://github.com/shadowmage45/SSTULabs/issues/384

 

Back on the subjects of inflatables... tried out various different forms of bracing and structure for them yesterday/last night, but only a few will work out due to the animations (well, lots of other options exist if I wanted to spend a few days/weeks animating and rigging things, which I don't; would also end up with an overly complex model not really suited to game use).

The one other bracing method that I found that might look okay is basic bracing lines / tension lines:

It animates decently, I'm just not sure that I really like how it -looks-.  Its not terrible, but still seems to be lacking something.  Suggestions/ideas?

Going to work on unwrapping the rest of the torus models today and probably work on the initial texturing for the torus rings; going to take a bit to figure out how to best lay them all out / re-use the textures / minimize distortion on the texturing, and most of it can be done independently of the rest of the models (the torus will need to be unwrapped regardless of what the rest of the geometry is).

Will probably know later today what the final lineup of inflatable torus parts will be.  Going to be a second set of rigid/deployable/non-inflatable centrifuges as well, but these will be done as a second set of parts (will require new textures and wholly new geometry).

 

 

[JoseEduardo]

now that it has struts it feels more complete to me

but if you want you could add extendable trusses under the struts

EDIT: or add more struts and a small centrifuge in the opposite direction, like Porkjet's: https://imgur.com/a/fZpyU

Edited November 8, 2016 by JoseEduardo

[Jimbodiah]

Not using the tanks because they can be longer makes no sense, as you could just stack as many as you want. Mage pointed this out as well a while ago. If I use BDB and/or tweakscale, or some other parts mod with tanks, there is always a larger tank available at early nodes. Let's not be holier than the pope here    The Merlin 1A is out of place maybe, but the rest of the tanks and engines are quite nicely balanced.

42 minutes ago, tater said:

I'd be inclined to unlock the least capable SRB at 0.625 from the start, along with perhaps probe core at the same diameter, and the MFT-A also at 0.625 (people can configure it as EC for a sounding rocket).

This is why there is a miniaturization node, for the 0.625m scaled parts. 0.625 as only option in the first nodes will make it obsolete and completely useless. If you want to go that route, use Unmanned-before-manned and make an SSTU patch for that.

[tater]

Part of the reason to use SSTU, aside from the obvious, is that stacks of little tanks are absurd, even with KJR. Rockets with X tank capacity have a tank built to the right size, be it a V2, Restone, or SLS.

Once the first VAB and pad upgrades happen, things start to even out quickly between stock and SSTU, frankly.

[CobaltWolf]

1 hour ago, Shadowmage said:

Hmm....  I will definitely have to give that a bit of thought;  sounds like it might be a workable alternative to the stock career start.  I have no problem with a sounding-rocket start to the space program, and would actually prefer to get away from the stock 'everything must be manned!' mentality at the start of the game.

However that shifts the problem to -- what existing engine(s) would be appropriate for a 0.625m rocket?  (keep in mind that I don't currently have time to design/model/texture new engines; I could make them 'next' on the todo list, but that still places them out towards the end of Dec/Jan; though perhaps a temporary placeholder-rescaling might be workable)

1 hour ago, tater said:

Or a 0.625 solid rocket unlock. The tank+nosecone would just be for aero/looks.

You could do Scout or Vanguard if you don't mind the upper stages being even smaller... same for Diamant, for that matter. They're all over-scaled in BDB.

[tater]

15 minutes ago, Jimbodiah said:

This is why there is a miniaturization node, for the 0.625m scaled parts. 0.625 as only option in the first nodes will make it obsolete and completely useless. If you want to go that route, use Unmanned-before-manned and make an SSTU patch for that.

I can go on for days about the stock career, lol, but that node is under electronics, anyway (from top to bottom the tree is rocket stuff, aero stuff, crew stuff, then electrical/science stuff). I might put the probe stuff there, instead.

6 minutes ago, CobaltWolf said:

You could do Scout or Vanguard if you don't mind the upper stages being even smaller... same for Diamant, for that matter. They're all over-scaled in BDB.

I made a "diamant" version of that solid posted above. The 1st stage was a 1.25m solid (min length) with a 2:1 taper nose to the decoupler, then that upper stage shown in space. Easily made orbit, by the use of careful thrust curves and staging near apoapsis from first stage.

Edited November 8, 2016 by tater

[riocrokite]

41 minutes ago, Shadowmage said:

Have opened a new issue ticket to keep track of the tech-tree / career suggestions:  https://github.com/shadowmage45/SSTULabs/issues/384

 

Back on the subjects of inflatables... tried out various different forms of bracing and structure for them yesterday/last night, but only a few will work out due to the animations (well, lots of other options exist if I wanted to spend a few days/weeks animating and rigging things, which I don't; would also end up with an overly complex model not really suited to game use).

The one other bracing method that I found that might look okay is basic bracing lines / tension lines:

It animates decently, I'm just not sure that I really like how it -looks-.  Its not terrible, but still seems to be lacking something.  Suggestions/ideas?

Going to work on unwrapping the rest of the torus models today and probably work on the initial texturing for the torus rings; going to take a bit to figure out how to best lay them all out / re-use the textures / minimize distortion on the texturing, and most of it can be done independently of the rest of the models (the torus will need to be unwrapped regardless of what the rest of the geometry is).

Will probably know later today what the final lineup of inflatable torus parts will be.  Going to be a second set of rigid/deployable/non-inflatable centrifuges as well, but these will be done as a second set of parts (will require new textures and wholly new geometry).

 

 

@Shadowmage you might consider adding darkened stripes - something like in the picture below: one practical explanation would be to see rate of rotation on a ship from a distance (similar to black stripes on rocket tanks):

Edited November 8, 2016 by riocrokite

[Jimbodiah]

14 minutes ago, tater said:

I can go on for days about the stock career, lol, but that node is under electronics, anyway (from top to bottom the tree is rocket stuff, aero stuff, crew stuff, then electrical/science stuff). I might put the probe stuff there, instead.

I get more frustrated with the stock tree and science system every day. They make no sense and are not balanced out with each other at all.

[RedParadize]

I like the wire, can we have more of them?

Try some "X" or "V" wire. Straight wire would not be effective for spin acceleration/deceleration. Some on the "V" wire on the vertical axis would be good to minimise vertical acceleration deformation.

As for the texture. I would suggest to have reinforced material at the base of the wire, maybe even angled wire that go around the torus.

Edited November 8, 2016 by RedParadize

[Temeter]

1 hour ago, Shadowmage said:

It animates decently, I'm just not sure that I really like how it -looks-.  Its not terrible, but still seems to be lacking something.  Suggestions/ideas?

 

Looking how 'real' concepts look, I found this quite interesting::

Spoiler

Mainly because of two elements:

Not only the thicker, dark grey rings, but especially the 'folding girders' ankered into them and the torus (which I imagine stabilize the unfolding and structure). The less smooth, mechanical look feels more realistic and less gamey.

I imagine it takes a lot of effort to model and animate the whole thing correctly, tho.

Edited November 8, 2016 by Temeter

[Shadowmage]

4 minutes ago, RedParadize said:

Try some "X" or "V" wire. Straight wire would not be effective for spin acceleration/deceleration. Some on the "V" wire on the vertical axis would be good to minimise vertical acceleration deformation.

 

Just now, Temeter said:

 

Looking how 'real' concepts look, I found this quite interesting::

  Reveal hidden contents

Mainly because of two elements:

Not only the thicker, dark grey rings, but especially the 'folding girders' ankered into them and the torus. The less smooth, mechanical look feels much more realistic and less gamey.

I imagine it takes a lot of effort to model and animate the whole thing correctly, tho.

Modeling details such as those are no problem.  I can get some fairly detailed (static) mechanics modeled if I wanted to.

The problem with that is the rigging and animation.  The torus is animated through simple scaling and there is no (easy) way to keep the orientations/rotations/positions of animated elements in synch with the torus inflation.  Tried a few of those yesterday, and it looked terrible due to things getting out of synch/misaligned during the animation; it looked fine when deployed, and not terrible while stowed, but during the animation it made me cringe....  (Someone with experience at skinned meshes / bone based rigging could probably pull it off; but I've not been able to get bone rigging to do -anything- properly, it always mangles the model in undesirable ways).  Even if I could keep things in sych, anything that is part of the torus mesh itself ends up beings skewed by the scaling (such as the attach points for the braces, end up being odd shapes in the stowed animation state).

Take for example the external scissors-jack type structure from the image; If I parent the structure to the torus itself it will stay mostly in synch, but then the meshes get skewed during the animation (e.g. a round pipe will end up looking flat when in the stowed state; things that should be square end up diamond shaped).  If I rig the external structure as independent meshes the animation becomes much more straightforward to rig, and looks much better (no skewing), but then I have ~128 separate animated objects in the model, which is terrible from a 'use in a game' optimization standpoint (and still may take hours to animate).  Even if I were to add bones I'm pretty sure I would need >100 bones in order to animate it properly.  If I use BlendShapes I could keep the mesh as a single object, but then I need to manually animate >1000 vertices (as you can't use standard duplication/array techniques in BlendShapes, at least not that I've been able to figure out).

(For reference I'm pretty sure that PorkJets updated habs/inflatables, pics posted above, use BlendShapes for their rigging).

Still going to play around with a few more ideas, who knows, maybe I'll figure out some new techniques that make some of these concepts viable for an in-game model.

 

40 minutes ago, riocrokite said:

@Shadowmage you might consider adding darkened stripes - something like in the picture below: one practical explanation would be to see rate of rotation on a ship from a distance (similar to black stripes on rocket tanks):

Yea, the textures as-shown are simply placeholders  -- I re-used the textures from the HAB models merely to see how the distortion would look for a squared UV map on a torus.

(In short, what is shown is -not- what the torus textures will look like; about the only thing they will have in common is the base cloth texture, and they will both be segmented)

[RedParadize]

@ShadowmageFor the V or X shaped wire. Have a simple cube or cylinder mesh with both end skinned at 100% on each end. It will get a tad slimmer on deployment but who care right?

How your torus is skinned? 2 joints with  scale X Z animated?

Edited November 8, 2016 by RedParadize

[Shadowmage]

39 minutes ago, Jimbodiah said:

I get more frustrated with the stock tree and science system every day. They make no sense and are not balanced out with each other at all.

Indeed, me too.  So much so that if I weren't already knee-deep in a Part mod I would probably be undertaking the creation of an overhauled Career mode (tech tree, science system, funds, contracts, mission planning).

Perhaps if/when I ever get SSTU to a more complete state, and if KSP is still a viable game, I'll consider it as my next undertaking.

 

9 minutes ago, RedParadize said:

@ShadowmageFor the V or X shaped wire. Have a simple cube or cylinder mesh with both end skinned at 100% on each end. It will get a tad slimmer on deployment but who care right?

How your torus is skinned? 2 joints with  scale X Z animated?

It is not skinned at all.  Simply a solid torus mesh that is scaled on X/Z for the deflation animation (using transform based scaling).

[Temeter]

20 minutes ago, Shadowmage said:

Still going to play around with a few more ideas, who knows, maybe I'll figure out some new techniques that make some of these concepts viable for an in-game model.

If you go for scaling, maybe at least the girder structure could work if you replace the scissors with telescopic sticks (like the tension/bracing lines)?

Although they (and the 'hinges') are still gonna get squeezed, guess.

Edited November 8, 2016 by Temeter

[RedParadize]

@Shadowmage Humm, I think its still doable using similar procedure. If you want I could send you a mock-up that does this, but I am not sure if it would work unity. I am fairly convinced it would trough. A alternate method would be to scale joint instead of transform. If I recall correctly scaling joint might be a problem for unity. But if it work it would be doable with only two joints. With 3 you would have much more control on the retracted shape trough...

Edited November 8, 2016 by RedParadize

[riocrokite]

38 minutes ago, Shadowmage said:

if I could keep things in sych, anything that is part of the torus mesh itself ends up beings skewed by the scaling (such as the attach points for the braces, end up being odd shapes in the stowed animation state).

Shadowmage, dunno if this helps however I might have had similar problem when making concepts of a big antenna lately, anyway what might help is:

- attach empty objects (I do it in blender) to vertices on the tube then parent them to the tube. Now you can use 'look at' those empty objects constraint for your beams where appropriate. It helped me with deployable dish animation - beams have constant crossection and I only change their length to account for deployment animation, example below red circles are the empty objects parented to the dish (which is heavily scaled in x/y and a bit in z scale as well):

36 minutes ago, Shadowmage said:

Indeed, me too.  So much so that if I weren't already knee-deep in a Part mod I would probably be undertaking the creation of an overhauled Career mode (tech tree, science system, funds, contracts, mission planning).

Perhaps if/when I ever get SSTU to a more complete state, and if KSP is still a viable game, I'll consider it as my next undertaking.

 

It is not skinned at all.  Simply a solid torus mesh that is scaled on X/Z for the deflation animation (using transform based scaling).

yah I guess the end caps would be fixed and be a part of the center / outer tube and all 8 beams could be grouped into 2 objects, each with 4 X and 4 Y beams scaling outwards from the center.

last thing that might help with animation:

https://wiki.blender.org/index.php/Extensions:2.6/Py/Scripts/Animation/Dynamic_Parent

this tool was handy for me some time ago if you're using blender, you can parent / unparent objects dynamically for only a part of animation so you're much more flexible with compound animations etc

Edited November 8, 2016 by riocrokite

[Shadowmage]

Blue = standard rigging, using hooks, look-at, follow, and transform-based animation.  Would require a substantial number of separate meshes and game-objects.

Gray = BlendShaped based animation.  Probably much better from an optimization/performance standpoint, but a bit tricker to rig up.

Either would be workable, but neither is optimal for one reason or another (its either easy to rig, but bad for performance; or hard to rig but better for performance).

Going to test out a fully blend-shaped based setup that would allow for a single mesh to be used for the entirety of the torus, crew tubes, trusses, and rigging.  Should look much better, but will most certainly be a bit tricker to animate.   Hmm... perhaps blender will let me join multiple blendshape animated objects and not screw up their animation? (would make the duplication of details across segments oh-so-much-easier).
 

Bit of reference info on actual NASA torus/centrifuge concepts.  For their concepts the bracing is all on the inner structure; which was fully rigid.  No clue how they intended to launch the things...  Also shows how compact the actual inflatable portion of the torus can be (apparently they can pack down to ~2% of inflated volume; but who knows if those early concepts included adequate radiation or meteor/debris shielding).

http://history.nasa.gov/SP-4308/ch9.htm

Also shown is a novel folding semi-rigid centrifuge that I'm hesitant to try and replicate; neat concept, but I really don't think I could rig/animate it acceptably.

Other novel concepts are shown in image [278] (c) and (e)

 

26 minutes ago, riocrokite said:

Shadowmage, dunno if this helps however I might have had similar problem when making concepts of a big antenna lately, anyway what might help is:

- attach empty objects (I do it in blender) to vertices on the tube then parent them to the tube. Now you can use 'look at' those empty objects constraint for your beams where appropriate. It helped me with deployable dish animation - beams have constant crossection and I only change their length to account for deployment animation, example below red circles are the empty objects parented to the dish (which is heavily scaled in x/y and a bit in z scale as well):

Indeed it is a workable idea, already used for several of my other animations.  The problem with using it for this case would be the raw number of game objects/meshes that would need to be used (where each mesh usually requires a separate draw call).  Not entirely out of the question for some internal bracing / tension lines (maybe 24 meshes?), but no way I could use it for external scissors-bracing (would require at least 48 meshes + mover objects).

The best compromise would be if I can figure out Blenders vertex-hook system and combined that with some standard animations to basically 'bake' a BlendShape animation from the vertex deformation.

 

45 minutes ago, RedParadize said:

@Shadowmage Humm, I think its still doable using similar procedure. If you want I could send you a mock-up that does this, but I am not sure if it would work unity. I am fairly convinced it would trough. A alternate method would be to scale joint instead of transform. If I recall correctly scaling joint might be a problem for unity. But if it work it would be doable with only two joints. With 3 you would have much more control on the retracted shape trough...

I'm honestly a bit clueless when it comes to armatures / bones / skinning.  My only experience with it was attempting to rig up the engine fuel joints.... and it was a terrible experience that I would rather not repeat. 

Probably mostly due to lack of actual experience/training on using those features, but I could never get them to work anything close to what I had intended.  Didn't seem to matter how I set up my vertex weights or groups, things just got mangled terribly whenever the rigging was moved around or animated.

 

 

 

Anyhow, lots of work still to be done on these parts; hopefully I'll be able to come up with something decently 'complex' looking without requiring multiple days worth of animation work.

[RedParadize]

@ShadowmageThats nice! about joint and stuff, if you can do without it then you should. The setup @riocrokite have is a bit more complex but will give better result. (thanks allot by the way)

I believe the "V" should be upside down, the two attach should be on the middle. A bit like bike wheels. That's gonna be glorious!

 

 

[Shadowmage]

One more render/animation, using blend-shape based setup for animation of the bracing lines and external support ring.  This method would allow for the entire rotating torus section to be a single mesh, and wouldn't be too difficult to animate... but I'm not exactly sure if I like the looks of this setup (mostly the external bracing 'moving' compared to the torus during deployment).  Going to put together a few more variations of the rigging to see if there is a layout I like better.

If I didn't mind having a much less compact 'stowed' state I could do very similar to what PJ had done with his reworked hab parts -- collapse the major diameter of the torus but keep its minor diameter intact.  Sadly that would result in this part having a 6.25m 'stowed' diameter (2.5m + 2.5m + 1.25m), which is far larger than I would like or consider acceptable for a payload.

Also haven't figured out how to handle windows on an animated torus -- currently they would get very distorted on the collapsed shape; would require an almost completely new way of modeling and animating the part.

In the end, especially for the torus/centrifuge parts, I'm okay with the animation/deployment not being 100% logical/possible.  With the 'kit out' mechanics in place, it is supposed to be more of an animated stand in for construction rather than mechanical inflation.  This goes back to the 'needs to look good while deployed, decent while stowed, with an animation that doesn't make you cringe'.