NathanKell

Correct on both counts. Good catch.

Sorry, I really have no idea about unity <-> video card interaction :\ All I can suggest is try replacing the 8192 textures *one at a time*, starting with the planet you visit the least, with the 4096s.

KSP has that setting in its own settings menu, but the choices are hard-coded to to be either the Kerbin default, or 24h day, 365 day year. If you want a custom display it has to be coded itself (as a separate window), or try to overwrite on LateUpdate every time a time/date is written anywhere.

Oh, indeed! That I don't take issue with in the slightest. I was taking issue only with your statement that it was doing nothing that the game itself didn't, but since I didn't quote it, how were you to intuit my intentions? :] Actually, it would be best if it detected if a file was a normal map based first on checking ATM's list of normal maps, and then (if that fails) perhaps if the file was mostly blue-purple to catch the leakers, or already in DXT5nm-ish format just in png (to catch RSS normal maps). Then compress all remaining 24bit images to DXT1, and the 32s to DXT5.

slumpie: except it was done without the memory leaks and loader bugs that the game's PNG and TGA loaders suffer.

I may have my differences of opinion regarding priorities and bugs, but boy howdy I'd be the first to say that Squad *does* work hard.

Yes, a couple months back Calisker announced that KSP.Edu would be having female as well as male names for kerbals, even before the female kerbal models were added.

The forum post is called "How to get support" for a reason. It's because we need that info in order to get you your support, not because it itself will fix the issue. Please follow the example issue reports.

Welcome to the forums! Please follow the guidelines in the READ FIRST sticky. Please model your issue report on the samples included. We do need all that information.

Rockets are already inefficient compared to real ones. Don't think of Isp in isolation; the problem is that tanks are about 4x as heavy as they should be, and engines about 3-4x as heavy. Due to how the rocket equation works, that's worth a few dozen % of your Isp (simplifying; MR is under a logarithm). So sure, you may be getting 390s out of that LV-909 instead of the 340 you should be (~tops~), or 350 instead of 290 from a lifter (assuming KSP engines run off standard hypergolic storables, which would explain the infinite restart capability), but since your mass ratio is a quarter what it could be absent payload, you're actually less efficient.

Engines don't run stoich. Russians tend to run ox-rich, and US engines fuel-rich.

Alt+G 10char

The contrails are just FX. Assuming you have both density exponents set to the same thing, they will be *partly* connected, but never trust their presence or absence.

The max diameter and max length are set to float.Infinity That I'm doing PP doesn't have relation to boiloff per se, because as I answered there, a single part may have many tanks inside it and thus you shouldn't base (RF "sub" tank) surface area on the part's surface area. Consider a proc part service module with some LOX and LH2 for a fuel cell, some NTO/AZ50 in highly pressurized tanks, some NTO/MMH in membrane tanks, some Helium for tank pressurization, and some batteries (ElectricCharge). No way the LH2 boiloff rate should be based on the entire SM's surface area.

Yeahwell I majored in philosophy I based the boiloff for non-cryo (i.e. "lightly insulated") tanks on the stated 700lb/hr boiloff on S-IVB, but that doc looks worth a good read, thanks.

Note that since all the KSP loader does *is* convert the mbms to DXT (dds) it's not surprising memory usage will remain the same. What you did is just doing that step once and saving the result. It is however good to know that the mbm loader doesn't leak, if your memory usage didn't decrease.

And for further context, swamp_ig is basically done with KSP, so despite the OP being his I'm, with OtherBarry's help, the maintainer...

NathanKell whacks himself for not reading the OP better

Northstar1989: That's a good point. However, since a RealFuels "tank" part is, unless it is a balloon tank, just a structural thing inside which actual tanks are placed, I can't assume exterior geometry (and therefore surface area) is the area of the tank. I think a decent approximation would be to calculate loss rate as something like the (2/3) power of volume, rather than just volume, multiplied by some constant (because we assume that tanks are capsule-shaped, not spherical). Note that boiloff is always based on the maxAmount of the resource, never the current amount, so that doesn't have to be changed. Rabada: Felbourn is correct. The other change is that because xenon is much less dense, but the ratio of liters of xenon : units of Electric Charge is unchanged, you will be using much, much more power than before.

TGA has two loader bugs: 1. It's left uncompressed rather than compressed to DXT 2. As of .25, it sometimes just won't load (random). Neat idea! BTW have you verified that multiple-nozzle RCS with PROPELLANT nodes works properly? That was broken in .24, and I thought still broken.

FuelTank is deprecated. Parts are just Part, now. They have a list of PartResouirces which are the resources on the part. You can check .amount and .maxAmount

Ah, ok. Cool.

I get an NRE from ATM on load. With context, so you can see where: AgentList: 30 agents parsed and loaded. (Filename: C:/BuildAgent/work/d63dfc6385190b60/artifacts/StandalonePlayerGenerated/UnityEngineDebug.cpp Line: 49) NullReferenceException: Object reference not set to an instance of an object at ActiveTextureManagement.ActiveTextureManagement.SetupLoaders () [0x00000] in <filename unknown>:0 at ActiveTextureManagement.ActiveTextureManagement.Start () [0x00000] in <filename unknown>:0 (Filename: Line: -1) [CelestialBody]: Kerbin's solar day length is 6h, 0m, 0s long. sidereal day length is 5h, 59m, 59s long (Filename: C:/BuildAgent/work/d63dfc6385190b60/artifacts/StandalonePlayerGenerated/UnityEngineDebug.cpp Line: 49) [HighLogic]: =========================== Scene Change : From LOADING to MAINMENU ===================== This is with 3-8, and things do appear to be working properly ingame. But I figured I should point it out. I will try on a stock install, etc, to see what's causing it.

1. Someone posted a list of the remapping in the SXT dev thread, I think. Have that open, open *all* your CFGs in Notepad++ or whatever, then do "find/replace in all open documents" to fix everything at once for each changed name. 2. No, you need absolute paths in MODEL nodes. 3. All Mk2 parts were replaced with new SP+ parts. Note that the Mk3->Mk2 adapter doesn't even work anymore, since it's still set for the old Mk2 fuselage shape.

Assuming you use MODEL and just specify the model, you don't need to use texture=, you can just have the models reference the same texture themselves. See how FASA does it for an example.