For Science! - what was not announced or mentioned

[WelshSteW]

11 minutes ago, Stoup said:

Ahh, I hadn't seen either of those posts! Was there some kinda interview I missed or was this Nate just randomly popping in 

 

It looks like Discord, not the forum.

[Spicat]

10 minutes ago, Stoup said:

Ahh, I hadn't seen either of those posts! Was there some kinda interview I missed or was this Nate just randomly popping in 

He was just answering some random questions because he had time due to him being on the plane and unable to work:

[Periple]

8 minutes ago, Scarecrow71 said:

How does one actually optimize visual features?  I know you can go through the code and streamline that, but after a certain point you simply cannot optimize the actual code itself (you will eventually get to a point where you have minimized it all it can be).  So how beyond that do you optimize visual features?

I'm not asking to be a d-bag, but because I honestly would like to know how this is being done.  I'm a nerd.

That’s a really broad question and there’s no way to give a satisfactory answer that would fit into a forum post. It also boils down to “it depends.” But I’ll try.

Rendering something like KSP2’s terrain is pretty complicated. Your objective is to have exactly the right number of vertices of geometry and pixels of texture in every frame — too many and you’re wasting resources, not enough and it doesn’t look as good as it should. You also need to cull any geometry and textures that you’re not going to render — the back of a planet, anything not on camera, and so on. When you’re looking at a planet, you’ll need less detail on the limbs than what you’re facing because they’re edge-on. And so on. And then there’s stuff you can pre-render and cache.

So you have a ton of logic there just to figure out what you’re rendering, moving geometry and textures in and out of memory (both main and VRAM). This is also costly and it might turn out that your clever algorithms are taking too long while the GPU is sitting there waiting for stuff to show. So then you need to go back and find ways to speed up your algorithms even if it puts more load on the GPU.

Then there are the assets themselves. You can optimize them so they look great with as few vertices and as small textures as you can get away with. There can be really big gains there.

Finally there are the shaders and FX — the magic that give things the “look” you want them to have. These are also code and can be faster or slower.

So, your objective is to get all of this dancing together so that the data flows through as smoothly as possible, and gives you beautiful scenes with as few dropped frames as possible, while still leaving CPU cycles for… the actual game logic.

Game engines do a lot of this for you on their own if you’re dealing with something it’s built to handle — like a level you can load into memory, or a larger world set up in a particular way that makes it easy to stream. Unfortunately a lot of this doesn’t hold for KSP — no game engine is designed out of the box to stream a world that’s in interstellar scale but where you can see an individual kerbal’s eyelashes. A lot of stuff has to be built specifically for it. And with anything like that, the first step is to get it working without painting yourself into a corner, after that, you start optimizing, and it’s not at all uncommon to find efficiencies that make things 10 or even 100 times faster when you get into it.

(It’s also super satisfying when you do that!)

[Vl3d]

3 hours ago, Spicat said:

Then what I wanna know is if the current terrain will get better textures and more details (scatter etc.).

Edited October 25, 2023 by Vl3d

[Cracha]

Did they mention radiators with the for science! update? I know they’re probably most useful when we get more heat generating parts, but I’d like to play around with procedural radiators with re-entry heating.

[dansiegel30]

Certainly more instruments than they showed.  I believe only 7 were shown when they showed the VAB Science filter?  Also, was looking forward to Kerbal specialization.  Pilot, Scientist, Engineer - but with sub specialties for future features like colonization.  This would have been a good time to do this, perhaps 0.3 with Colonies.

Edited October 26, 2023 by dansiegel30

[Vl3d]

It's not clear for me if we're gonna see a "capabilities progression" mechanism like unlocking maneuver plans or more advanced SAS features as we do missions and gather science. I really enjoyed this feeling of progression in KSP1 and I've always wanted it to be extended by the gradual discovery of celestial bodies with telescopes and missions. I think that having a certain tempo of exploration and being rewarded with new information, parts and capabilities after successfully fulfilling key missions is critical to the flow of the campaign.

[Periple]

6 hours ago, Vl3d said:

It's not clear for me if we're gonna see a "capabilities progression" mechanism like unlocking maneuver plans or more advanced SAS features as we do missions and gather science.

I thought that was a really bad idea! It made the initial learning curve so much steeper and I’m quite sure that a lot of people just gave up because of that!

6 hours ago, Vl3d said:

I really enjoyed this feeling of progression in KSP1 and I've always wanted it to be extended by the gradual discovery of celestial bodies with telescopes and missions. I think that having a certain tempo of exploration and being rewarded with new information, parts and capabilities after successfully fulfilling key missions is critical to the flow of the campaign.

I think that’s a totally different thing! Gradually unlocking parts is great and it can make the learning curve easier too as you start out with less complexity to manage and simpler objectives to meet.

I do hope that discovery with telescopes and such will be a part of the game too but it will feel silly if it applies to everything. I shouldn’t need to discover Duna. Maybe Gilly, Dres, Eeloo, and the smaller Jolian moons?

[magnemoe]

On 10/25/2023 at 8:09 PM, Cracha said:

Did they mention radiators with the for science! update? I know they’re probably most useful when we get more heat generating parts, but I’d like to play around with procedural radiators with re-entry heating.

Don't think radiators works well with reentry heating, realistically and how it worked in KSP 1 where the heath shield either overloaded or blew. Exception is probably the 10 meter inflatable shield and circulation into low Eve or Jool orbits. 

 

[dansiegel30]

On 10/26/2023 at 11:22 PM, Periple said:

I thought that was a really bad idea! It made the initial learning curve so much steeper and I’m quite sure that a lot of people just gave up because of that!

Kerbal capabilities could really be expanded well beyond what it did in KSP1.  In KSP1, the only pain was pilot training, however you could always use a probe core instead, to get maneuver node SAS, etc.  But yeah, I never did like having to get my pilots to the Mun and Minmus (IIRC) in order to get enough experience to use manuever node locking.  

 

However, the main reason for Kerbals skill progression in KSP2 I think is for engineers and scientists.  It would give a REASON to have an army of kerbalnauts to hire, requiring the keep high skilled engineers at orbital construction facility stations or Fuel manufacturing facilities, in order to provide speed bonuses.  High skilled scientists could generate more science or faster science on long distance missions or perhaps some experiment requiring a huge amount of time to complete (ala KSP1 science lab or Kerbalism).  The engineers will definitely have the biggest role in the Colonization and Orbital Construction update.

 

This is a small feature in the big scope of things, so hopefully it is being planned, but isn't happening soon (or in 0.2) but will happen one day.

[dansiegel30]

On 10/25/2023 at 1:09 PM, Cracha said:

Did they mention radiators with the for science! update? I know they’re probably most useful when we get more heat generating parts, but I’d like to play around with procedural radiators with re-entry heating.

No, they did not mention it, however you can see radiators in the Action Group Manager.  There is no button yet, but they accidentally left that in the code in the AGM.  So, it is indeed coming, bit in all likelyhood 0.3 with colonies and ISRU which will indeed require them.  

 

I never liked the hack in KSP1, using radiators with no heat shield during aerocaptures.  Completely unrealistic.  For infinite aerocapture capability, I just normally make a space-plane.

[Vl3d]

I'm really worried about the fact that CommNet implementation has not been shown or announced. We have not really seen the occlusion system, nor the DSN, nor relays being used, nor an UI for connections and range, let alone an antenna planner. Does the reentry heating system also cause plasma blackouts?

I don't even want to go into details about any innovations that could have been added to the game, like requiring 2 antennas to make a relay instead of separating the functions into 2 antenna types.. or boosting signal at the expense of EC etc.

And why are we expected to know how far 100 Mm or 10 Gm is? Can't there be a visualization tool for this?

I hope there will be some alternative to spamming 12 antennas on a craft to get enough range. That's just not realistic or logical in any way.

CommNet is a system that really has room for improvements and innovation - doing it the same way KSP1 did would just be disappointing.

Edited October 31, 2023 by Vl3d

[jost]

On 10/24/2023 at 8:06 PM, shdwlrd said:

It depends on the reasoning for the telescope. If you need a telescope to discover a planet in your solar system, that's stupid. You should see the planet transiting across the night sky. You know it's there. If you want to use the telescope to get a better view of the planet, that's cool. If you want to use a telescope to do a ground survey before a mission, that's up to you. But using telescopes for anything in the Kerbol system should be optional and not a requirement for any discoveries.

Hm, the outer Planets ( Neptun, Saturn, Uranus, Pluto), Jupiters moons and Asteroid belts were discovered with telescopes so I don't think that this argument is  a good one in any case. I agree however that most planets of the kerbol System shouldn't need to get discovered by a Telescope. 

Eeloo, Dres and maybe Gilly, Pol and Bop are a different beast though.

In the end I don't think KSP2 will have it since it would make it more difficult for new players

[Astroneer08]

I would love a more detailed version of KerbNet, like instead of just watching a screen refreshing, and an easier way to close the menu.

[Dakota]

Happy to share that you can expect buoyancy improvements in For Science!

[Strawberry]

1 hour ago, Dakota said:

Happy to share that you can expect buoyancy improvements in For Science!

Once again my blindful hopeposting proves to be prophetic visions delivered psychically from Nate Simpson

[MechBFP]

On 10/31/2023 at 5:14 AM, Vl3d said:

 

I'm really worried about the fact that CommNet implementation has not been shown or announced.

 

What exactly is your worry?

[Vl3d]

12 hours ago, MechBFP said:

What exactly is your worry?

I provided some details in the post.

On 10/31/2023 at 1:14 PM, Vl3d said:

We have not really seen the occlusion system, nor the DSN, nor relays being used, nor an UI for connections and range, let alone an antenna planner. Does the reentry heating system also cause plasma blackouts?

I don't even want to go into details about any innovations that could have been added to the game, like requiring 2 antennas to make a relay instead of separating the functions into 2 antenna types.. or boosting signal at the expense of EC etc.

And why are we expected to know how far 100 Mm or 10 Gm is? Can't there be a visualization tool for this?

I hope there will be some alternative to spamming 12 antennas on a craft to get enough range. That's just not realistic or logical in any way.

CommNet is a system that really has room for improvements and innovation - doing it the same way KSP1 did would just be disappointing.

I also have some ideas in the Grand Unified Wishlist:

Spoiler

• visualization of antenna range in map view
• antenna planner
• remove dedicated relay function and require 2 antennas to make a relay
• discourage antenna stacking to increase range
• add antenna gain/power slider (consumes more EC)
• data transfer speed

There's a lot of cool innovations that could be used to improve CommNet. My worry is that it will not be improved because "it just works".

Also, CommNet is joined at the hip with the Science system. You can't really have one without the other - and most ideas for Science mini-games involve communications: data transfer speed, data size, transmission EC consumption, relays etc. Comms make or break science - just check out Kerbalism Science!

[MechBFP]

1 hour ago, Vl3d said:

I provided some details in the post.

I also have some ideas in the Grand Unified Wishlist:

  Reveal hidden contents

• visualization of antenna range in map view
• antenna planner
• remove dedicated relay function and require 2 antennas to make a relay
• discourage antenna stacking to increase range
• add antenna gain/power slider (consumes more EC)
• data transfer speed

There's a lot of cool innovations that could be used to improve CommNet. My worry is that it will not be improved because "it just works".

Also, CommNet is joined at the hip with the Science system. You can't really have one without the other - and most ideas for Science mini-games involve communications: data transfer speed, data size, transmission EC consumption, relays etc. Comms make or break science - just check out Kerbalism Science!

You have seen the game in EA. You should lower your expectations to the game not getting improved so that you can be happy if anything does. 

[Periple]

2 hours ago, Vl3d said:

There's a lot of cool innovations that could be used to improve CommNet. My worry is that it will not be improved because "it just works".

Then I think you need to recalibrate your expectations, as the KSP2 general design paradigm has been to go broader rather than deeper. Why do you think CommNet would be any different?

2 hours ago, Vl3d said:

Also, CommNet is joined at the hip with the Science system. You can't really have one without the other - and most ideas for Science mini-games involve communications: data transfer speed, data size, transmission EC consumption, relays etc. Comms make or break science - just check out Kerbalism Science!

There was a suggestion that instead of being able to transmit/retrieve anything, science experiments will have a flag for "can transmit." If I had to hazard a guess, I'd expect that experiments that involve taking readings can be transmitted back with no loss of reward, but experiments that involve material samples will have to be returned. If that's the case, then the only thing about comms is that for them to work, you need to have a connection. 

If you're expecting something deeper or more detailed than that, I think you're setting yourself up for a disappointment again.

[Bej Kerman]

On 11/4/2023 at 11:44 PM, jost said:

Hm, the outer Planets ( Neptun, Saturn, Uranus, Pluto), Jupiters moons and Asteroid belts were discovered with telescopes

Ground telescopes. Didn't need to wait for JWST to discover Neptune.