[1.2] Real Solar System v12.0 Dec 8

[asmi]

According to experiments in the UR thread, KSP/Unity will reject textures over 8x4. I'm thinking we should try out an 8x4 Earth heightmap for kicks, see how that goes. If we're replacing, might as well go whole hog. But you're right, it's a drop in the bucket. We'd get 0.043 degree (~5km) per pixel resolution at the equator. I wonder if there's any way to use the scatter system for Munar craters to generate procedural detail on Kerbin?

Well then we'll need to mod PQSMod (lol) so that it would work with texture patches instead of whole thing. But then there is a problem of memory - these textures aren't gonna be small. To get our KEarth to Kerbin resolution, well need to increase resolution by factor of 11, which will increase total texture size by factor of 11^2 (area is proportional to radius squared)...

Which basically means that we will have to implement paging mechanism for heightmaps and textures since they simply won't fit into process memory.

By editing KSP's config file I've made engine to render planet beyond max level 10 (I've made it to lvl 11, but theoretically we can go even further).

Another problem of PQS system is that it generates geometry in RAM using CPU, which could become a bottleneck (although as far as I understand Unity allows initializing its' objects in other threads, so moving init code out into separate thread could possibly alleviate the issue). In my engine for Orbiter (and another space sim engine I'm working on now in a background) I was generating all that stuff on GPU right within rendering pass - via vertex shader for initial vertices setup, then tessellation(for DX11) or Geometry shaders(for DX10) for subdivision. PQSMod components even do bilinear filtering on CPU (while in GPU there is hardware dedicated for that, so there you get it essentially for free).

Bottom line - this is not easy nor simple, but it could be done. And whoever manages to do that will probably end up hired by Squad since this mechanics would be useful for stock game as well

ZRM...heh, life. :]

We have complete access to Unity runtime, which basically means we can do just about anything that Unity itself can do (and as such, Squad devs can do).

-------------

Another thing - you've probably noticed that speed indication doesn't automatically switch to orbit mode as you do ascent. My experiments have shown that it does so once your altitude is more than 6% of celestial body's radius, which probably won't fly in our case as 6% of Kearth is 384 km

[NathanKell]

Yeah, we'd definitely need to page. And re: geometry, the quadtree system I wrote (bouncing off Sean O'Neil's old demo) for my senior project used that method (I just passed IIRC 9 base meshes to the GPU, then vertex transformed based on shader-calculated noise) but the cards the CompSci department had didn't even support tesselation! (and I didn't have time to do a full geometry shader version--bah, semester projects with illnesses!)

Anyway. Do you know how much modern shader support Unity actually grants? Might be worth thinking about a port of your engine....

(DX->OGL is bad enough, let alone wrappers, but just thought I'd throw it out there)

The life remark was a rueful one re: ZRM's semester load interfering with writing this, or rather the other way around, and the universality and annoyingness of that problem. That was all. Apologies all around if it was unclear

I had noticed. I guess I could add an on-fixed-update check to change navball mode, but I'd hate to force-switch someone who had already switched. And since I'm usually annoyed at the autoswitching, it never bothered me much. But 384km _is_ a bit much!

Edited November 2, 2013 by NathanKell

[Lucius]

Nathan:

Hi, first off, absolutely LOVE this mod. It reinvigorated my passion with KSP, as an old orbiter diehard. So first and foremost, as i see your name plastered on just about every "must have" mod for me, from DR to Modular Fuel to this and seemingly everywhere else at once, THANK YOU. What you're working towards is just amazing.

Second, to contribute some ideas here:

1) What is going on with kerbin/earth/whatever we're calling it now's atmosphere. it seems to taper out around 80km, i was under the impression that the thermosphere started around 160km. Are there eventual plans to relayer the atmosphere, or is this deliberate?

2) Now, i KNOW this isn't priority, but i'd like to say, this mod actually integrates QUITE well with career mode, when one makes some adjustments to the science limits from experiments. With a bit of a ... wider pool of science points to be gained without having to leave earth's SOI, it has been sensationally immersive for me! I'm not sure if you have any plans to tie in this world to career mode or just leave it open, but if you do intend to, it does not need much!

**Edit: I should guess you're going for the Karman line, but, i am just curious as to whether you intend to increase atmospheric drag out as far as it really is, or if it's modelled and i just keep crashing so much that i never notice atmospheric drag being modelled until about 140km

Edited November 2, 2013 by Lucius

[Tobyz28]

Nathan:

Hi, first off, absolutely LOVE this mod. It reinvigorated my passion with KSP, as an old orbiter diehard. So first and foremost, as i see your name plastered on just about every "must have" mod for me, from DR to Modular Fuel to this and seemingly everywhere else at once, THANK YOU. What you're working towards is just amazing.

Second, to contribute some ideas here:

1) What is going on with kerbin/earth/whatever we're calling it now's atmosphere. it seems to taper out around 80km, i was under the impression that the thermosphere started around 160km. Are there eventual plans to relayer the atmosphere, or is this deliberate?

2) Now, i KNOW this isn't priority, but i'd like to say, this mod actually integrates QUITE well with career mode, when one makes some adjustments to the science limits from experiments. With a bit of a ... wider pool of science points to be gained without having to leave earth's SOI, it has been sensationally immersive for me! I'm not sure if you have any plans to tie in this world to career mode or just leave it open, but if you do intend to, it does not need much!

I'd also like to expand on this, I'm trying to configure DR with Real Earth and FAR (normal KSP got too easy for me lol). I'm also playing in career mode and... am having difficulty advancing ... I've taken the the setting recommended from DR 2.3 thread and having issues entering the atmosphere with G-Forces now. I think this may be relate to what you were referring to Lucius, the atmosphere starts too low and increases in density too fast to allow for realistic G-Forces . On a VERY shallow entry into the atmosphere using Real Earth im travelling around 5-6k/sec, i get nailed by G forces every time with my pod exploding due to g forces (never from heat). I have yet to achieve orbit with the few parts unlocked yet in campaign mode.. perhaps my entry angle is still not shallow enough??

I really feel like the combination similar to mine (FAR/DR/RE(and modular fuel)) is a very very nice one for players wanting a more real, and MUCH harder KSP experience. Would it be possible for you to try the above combination with a small pod re-entering The problem should be come apparent pretty fast I think you might just have some fun while at it too!

[Lucius]

hahaa toby it's funny you should say that, i have about 7-8 nodes unlocked now, nothing major but enough to slap together some semblance of a rocket, and i FINALLY got my first manned capsule back to ground safely. Not even an hour ago!

Mind you i'm running this mod, DRC with nathan's suggestions for real earth (12 heat multiplier, .9 shockwave exponent), MFT, KJR, remotetech (which JUST came out hence i was playing an hour ago), FAR, but i also had KW Rocketry. Which is, i imagine, the only way i managed this at all. With a very careful climb and a close watch on your TWR, you can at very least even in low tier rockets, ensure that your trajectory runs fairly flat with the earth. Any steep entry, and i really want to emphasize ANY STEEP ENTRY, will produce too much g force, regardless of heat.

I literally had an angle of no more than 6/7 degrees retrograde during the largest part of my descent, and while the ablative shield held just fine during descent, even at this careful angle, i NEARLY wound up in the red of g forces for almost a minute solid. Scary little minute there! So yes, i know, the recovering pods for every extra inch of science is good, but yeah reentry angle is everything, watch your gravity turn on ascent it has to be so, so specific

[NathanKell]

Lucius: aw, thank you so much! Do note that I'm standing on the shoulders of giants, especially with DRE which are direct continuations of ialdabaoth's great work with them.

Regarding the atmosphere, there was a discussion some pages back, and some discussion in the Realism Overhaul thread too. Basically, using the legacy (exponential) atmosphere model, a planet's atmospheric density is clamped to 0 when the exponential calculation (as shown on the wiki) falls below 0.00001. That means, with a scale height of 7.5km, it hits zero at approximately 104km. I'm currently trying to switch the atmosphere model over to the new pressureCurve system, which will allow a quite-close copy of Earth's atmosphere. However, when density > 0, then time acceleration is disabled (and the game thinks of you as not in space). So it makes sense to formally cut it off at some point. Ferram suggest 135km. However, regarding the thermosphere, asmi (after finishing the other mods they're working on) was thinking of making a plugin to handle orbital drag and stationkeeping, so that would be taken care of too.

Regarding career. If you look in the Add-On Development forum, you'll see a thread by MedievalNerd on a realistic tech tree (I'm collaborating with him on it). Career mode is indeed coming part-and-parcel with all this; the TechLevel system in MFSC was designed for it. The idea is to model rocketry from soundnig rockets and WW2 SRBMs to the present day, with the potential of a career stretching from then to now.

Regarding Deadly Reentry. I recently posted on that thread suggested configurations, because there was some confusion. Let me restate the relevant one here. I recommend leaving ALL DRE stats as they are when unzipped, and instead increasing heat shield dissipation. Note in particular that the new shockwave exponent and multiplier are only for people who want a deadlier reentry on stock-size Kerbin; the 1.17 exponent I mentioned was to approximate heating on real earth when using 600km-radius-Kerbin. You, who are using a real 6371km-radius-Kerbin, don't need that, and indeed shouldn't use it.

If you want a Mk1 heatshield config that works with this mod, and DRE multipliers unchanged, check the DRE thread. You should, on a correct reentry, suffer no more than 9 drag Gs in a Mk1 Pod. I just tested a suborbital launch with it and got Mercury's 11, so that's survivable too.

I have also made a custom heatshield for the RealScale Gemini (the set of CFGs for frizzank's Gemini to give it real mass/size/performance, see the FASA thread) and that also reenters just fine. The readme offers suggested reentry perigee (indeed, suggestions for the entire flight).

[Tobyz28]

Lucius: aw, thank you so much! Do note that I'm standing on the shoulders of giants, especially with DRE which are direct continuations of ialdabaoth's great work with them.

Regarding the atmosphere, there was a discussion some pages back, and some discussion in the Realism Overhaul thread too. Basically, using the legacy (exponential) atmosphere model, a planet's atmospheric density is clamped to 0 when the exponential calculation (as shown on the wiki) falls below 0.00001. That means, with a scale height of 7.5km, it hits zero at approximately 104km. I'm currently trying to switch the atmosphere model over to the new pressureCurve system, which will allow a quite-close copy of Earth's atmosphere. However, when density > 0, then time acceleration is disabled (and the game thinks of you as not in space). So it makes sense to formally cut it off at some point. Ferram suggest 135km. However, regarding the thermosphere, asmi (after finishing the other mods they're working on) was thinking of making a plugin to handle orbital drag and stationkeeping, so that would be taken care of too.

Regarding career. If you look in the Add-On Development forum, you'll see a thread by MedievalNerd on a realistic tech tree (I'm collaborating with him on it). Career mode is indeed coming part-and-parcel with all this; the TechLevel system in MFSC was designed for it. The idea is to model rocketry from soundnig rockets and WW2 SRBMs to the present day, with the potential of a career stretching from then to now.

Regarding Deadly Reentry. I recently posted on that thread suggested configurations, because there was some confusion. Let me restate the relevant one here. I recommend leaving ALL DRE stats as they are when unzipped, and instead increasing heat shield dissipation. Note in particular that the new shockwave exponent and multiplier are only for people who want a deadlier reentry on stock-size Kerbin; the 1.17 exponent I mentioned was to approximate heating on real earth when using 600km-radius-Kerbin. You, who are using a real 6371km-radius-Kerbin, don't need that, and indeed shouldn't use it.

If you want a Mk1 heatshield config that works with this mod, and DRE multipliers unchanged, check the DRE thread. You should, on a correct reentry, suffer no more than 9 drag Gs in a Mk1 Pod. I just tested a suborbital launch with it and got Mercury's 11, so that's survivable too.

I have also made a custom heatshield for the RealScale Gemini (the set of CFGs for frizzank's Gemini to give it real mass/size/performance, see the FASA thread) and that also reenters just fine. The readme offers suggested reentry perigee (indeed, suggestions for the entire flight).

Thanks Nathan, I think I've isolated what my problem was . I was using the settings listed in the other thread, Deadly re-entry was working as designed, I think i was a "tech-point" in game where following the tech tree, in order to reach space i was still entering at too sharp an entry on return - for real sized kerbin that is... I needed to enter from a nearly circular orbit

i've unlocked a few more nodes and seem to be able to achieve orbit, with a proper angle i can not "get squished" I do need to tweak the head sheids as mentioned - off to do that! Thanks for the help!

[Lucius]

I'm currently trying to switch the atmosphere model over to the new pressureCurve system, which will allow a quite-close copy of Earth's atmosphere.

Oh god yes... Also thank you, i've been watching Realistic Tech Tree Overhaul and cannot wait for that as well! Didnt even notice you were involved in that too (why am i not surprised)

[Tobyz28]

hahaa toby it's funny you should say that, i have about 7-8 nodes unlocked now, nothing major but enough to slap together some semblance of a rocket, and i FINALLY got my first manned capsule back to ground safely. Not even an hour ago!

Mind you i'm running this mod, DRC with nathan's suggestions for real earth (12 heat multiplier, .9 shockwave exponent), MFT, KJR, remotetech (which JUST came out hence i was playing an hour ago), FAR, but i also had KW Rocketry. Which is, i imagine, the only way i managed this at all. With a very careful climb and a close watch on your TWR, you can at very least even in low tier rockets, ensure that your trajectory runs fairly flat with the earth. Any steep entry, and i really want to emphasize ANY STEEP ENTRY, will produce too much g force, regardless of heat.

I literally had an angle of no more than 6/7 degrees retrograde during the largest part of my descent, and while the ablative shield held just fine during descent, even at this careful angle, i NEARLY wound up in the red of g forces for almost a minute solid. Scary little minute there! So yes, i know, the recovering pods for every extra inch of science is good, but yeah reentry angle is everything, watch your gravity turn on ascent it has to be so, so specific

Haha yes i'm having a blast with this, the challenge is quite amazing, i managed to land just recently with a pod at a very very gentle angle.. only difference was able to do it with a shock wave exponent of 1 (which i think was default?). Have you run into any "clicking" issues after burning/breaking up in atmosphere and going back to the space center (Cant click on any buildings - just exit? It's happening 100% of the time, i don't have many mods, but have yet to go through figuring out which one it is. Also curious if i'm the only one with it!)

[Lucius]

Yes, i'm FAIRLY certain it has to do with DREC, but with the cocktail of mods i am hesitant to point a finger. Seems to happen ONLY when a ship breaks up due to g Force, not even due to temperature. SEEMS to be a 50/50 chance with manned flights, but 100% chance of error with probes.

If your probe explodes from g force, you cannot click on anything, even alt tabbing in and out, and have to reset the game. If it blows up from heat... doesn't seem to be a problem. Odd isn't it!

[NathanKell]

You're both welcome!

It's a known issue in the stock game, it's just that DRE making your ship explode triggers it. To fix: you need to somehow launch again. The way I do it is revert to VAB, launch, and revert again.

I've had it from heat as well as G, btw; it's intermittent, because there's some underlying stock issue it's bringing to the fore.

Edited November 2, 2013 by NathanKell

[Sax Man Aeronautics]

Nathan, will you be rescaling the sun and putting the rest of the planets of Kerbol at their neccesary posistions?

[Beatlejuice]

can i go in the files to modify the mod and make kerbin twice the normal size?

i would like a bigger kerbin but i think earth size is a little too big:confused:

amazing mod btw:)

[ZRM]

Yeah, we'd definitely need to page. And re: geometry, the quadtree system I wrote (bouncing off Sean O'Neil's old demo) for my senior project used that method (I just passed IIRC 9 base meshes to the GPU, then vertex transformed based on shader-calculated noise) but the cards the CompSci department had didn't even support tesselation! (and I didn't have time to do a full geometry shader version--bah, semester projects with illnesses!)

A custom PQSMod should handle paging nicely for close range detail, and for long range either you can change the scaled space proximity settings to use PQS from further away or make the scaled space model have a higher effective resolution by splitting it into chunks and using a different texture per chunk. You could integrate that with a simple LOD scheme on the scaled space object.

By the way, you may be surprised to learn that several of the bodies in the Kerbol system (especially some of the moons) do in fact have their surfaces defined by a combination of low-res height maps and procedural methods like Perlin noise, simplex noise and multifractals. This also applies to the surface colouration. Some only use heightmaps for specific features like a few mountains. So if you increase their maximum tesselation resolutions you should get better detail on most terrain for free.

The reason Kearth looks so bad from orbit right now is because you haven't changed the ranges at which scaled space fades in (via the ScaledSpaceFader component and the appropriate settings in PQS), so the simple textured model is used when really the PQS terrain should be used.

The problem with KSP really is how old all of its tech is, especially Unity, which if it weren't for some things looks like it came straight from 2004. On a relevant note I'm actually implementing a voxel-based scene representation for my final project for my degree right now. With a bit of work it should extend nicely to planetary scales, though that's outside the scope of the current project.

Anyway. Do you know how much modern shader support Unity actually grants? Might be worth thinking about a port of your engine....

(DX->OGL is bad enough, let alone wrappers, but just thought I'd throw it out there)

For the Kerbin Mini Shuttle I wrote a plugin that lets you load custom shaders and redefine materials for parts and internals. The shaders are all written in Cg, but the build of Unity used by KSP uses an oldish Cg version, and it only renders under DX9 on Windows and OpenGL 3.x on other platforms, so don't expect anything like tesselation or compute shaders any time soon. The plugin also works as a manager for shader resources. I will release it once the whole KMS project is ready. BTW I also started work on a mod designed to allow parts to attach via multiple points to different objects based on proximity to their attachment nodes, so that you can make things in the VAB like 3 point connections on the Space Shuttle, or space stations with loops. This may be something Ferram would be interested in implementing.

---

Regarding the atmosphere, there was a discussion some pages back, and some discussion in the Realism Overhaul thread too. Basically, using the legacy (exponential) atmosphere model, a planet's atmospheric density is clamped to 0 when the exponential calculation (as shown on the wiki) falls below 0.00001. That means, with a scale height of 7.5km, it hits zero at approximately 104km. I'm currently trying to switch the atmosphere model over to the new pressureCurve system, which will allow a quite-close copy of Earth's atmosphere. However, when density > 0, then time acceleration is disabled (and the game thinks of you as not in space). So it makes sense to formally cut it off at some point. Ferram suggest 135km. However, regarding the thermosphere, asmi (after finishing the other mods they're working on) was thinking of making a plugin to handle orbital drag and stationkeeping, so that would be taken care of too.

I'd just like to say that if you have any problems getting Kerbin to switch to the newer atmosphere system it's not that big of a deal because I should be able to get the Earth to use the new system when I manage to create planets from scratch (eventually).

---

Nathan, will you be rescaling the sun and putting the rest of the planets of Kerbol at their neccesary posistions?

Those are his current plans, as well as implementing my idea for axial tilt. He's currently working on getting the changes he makes to be defined via config files.

Edited November 2, 2013 by ZRM

[NathanKell]

By the way, you may be surprised to learn that several of the bodies in the Kerbol system (especially some of the moons) do in fact have their surfaces defined by a combination of low-res height maps and procedural methods like Perlin noise, simplex noise and multifractals. This also applies to the surface colouration. Some only use heightmaps for specific features like a few mountains. So if you increase their maximum tesselation resolutions you should get better detail on most terrain for free.

Saw that, and yup, can fix some of the rolling-hills issues on rescaled Kerbin that way too.

The reason Kearth looks so bad from orbit right now is because you haven't changed the ranges at which scaled space fades in (via the ScaledSpaceFader component and the appropriate settings in PQS), so the simple textured model is used when really the PQS terrain should be used.

Yeah, been putting that off because it works as is (just not prettily), but I should hit that soon.

The problem with KSP really is how old all of its tech is, especially Unity, which if it weren't for some things looks like it came straight from 2004. On a relevant note I'm actually implementing a voxel-based scene representation for my final project for my degree right now. With a bit of work it should extend nicely to planetary scales, though that's outside the scope of the current project.

Nice!

I'd just like to say that if you have any problems getting Kerbin to switch to the newer atmosphere system it's not that big of a deal because I should be able to get the Earth to use the new system when I manage to create planets from scratch (eventually).

Don't think it'll be an issue; looks pretty simple. The harder part is converting the Standard Model of pressure at altitude to a set of cubic bezier points and tangents. Not hard per se, just will take some twiddling. :]

Those are his current plans, as well as implementing my idea for axial tilt. He's currently working on getting the changes he makes to be defined via config files.

Yup. Got it working last night, testing a bit more today and adding tilt.

[AndreyATGB]

I've played a bit with inf fuel, noticed you changed the Sun's mass to be correct (radius seems the same).

-Kerbin and the Mun seem fine

-Duna is completely screwed. It's model ends at about 3 million KM, Ike orbits at -151k KM and Minmus is unreachable. Once you get close enough, it never registers its SOI and you can even go through it.

-Dres says it's moving at 62k m/s but it seems to be standing still.

-Jool isn't moving at the reported speed of 43k m/s. It seems to be moving at its old speed.

-Moho has a really weird model. I was very far from it (about 13 degrees difference) and it looked like I was at a few million KM over its surface.

[NathanKell]

Apologies. I left up a testing version. I'll replace with a real version ASAP.

[AndreyATGB]

Apologies. I left up a testing version. I'll replace with a real version ASAP.

Oh, well could you still briefly explain how it's possible for Jool to move at 43k but not really?

[NathanKell]

KSP calculates position in orbit based on period, but speed based on mass ratios and SMA. If one is updated but not the other you'll get that effect.

[NathanKell]

Reverted. Should behave like v3 again.

[asmi]

NathanKell:

Just watched this video and thought about this mod. Check it out (watch it all the way to the end):

[Lucius]

and

Not exactly telling in terms of "Scale" ... but two beautiful pictures of your massive world mixed with the cloud and city lights mod. Was absolutely gorgeous in action.

[sidfu]

would this effect mods like kethane, scansat, and the new remotetech or would they work fine?

[Starwaster]

would this effect mods like kethane, scansat, and the new remotetech or would they work fine?

kethane... maybe?

[SZDarkhack]

Awesome mod (combined with FAR + DREC of course) for some refreshing challenge.

Small (non-)issue, the velocity gauge remains in surface mode a long way after "space" begins. In stock Kerbin it auto-switches before the edge of the atmosphere. Is KSP setting the switching altitude on its own or can it be overridden in the mod?