New Planet Tech is KILLER of Exploration (all terrain is tiling/repeating/not procedural/random)

LOL! I wonder what's happened to

Not sure I see how that follows?

An improved procedural engine, creating more diversity and less repeated tiles could surely also create superior outcomes? All depends on the quality of the rules used?
It's always a trade off between quality and computer power. And you can only get so detailed if you want to have the proc gen running on a normal PC.
So hand crafted tiles can be used to "modify" proc gen terrain to give it more detail without requiring more power.

The trick is: Can you make enough prefabs and transform them right so nobody realizes the sleight of hand?
 
Eh, generally the 10-15% FPS boost you get from switching to a dedicated RT model is enough of an improvement to make it worth it even as the AMD cards lag behind the Nvidia ones by around the same amount. With Metro Exodus enhanced, the 6800 handles 60 at 1440p just fine, with the big gap with the 2080 being in certain scenes where it drops up to 50% from the usual 15-20% it normally lags behind. Sure those settings suck if you have a 120hz 4k monitor, but it's not really a big deal, hell, I still use 1080 monitors rendering at 4k for the supersampling. Of course CPU is always the bottleneck for RT, given how much it offloads there from the GPU, so make sure you have a good amount of threads available for those games too.
RTX suxx and actually slow down progress of photorealistic graphics. Because it took too many resources to make lighting like 2% better. Umm, thanks, but no.
rtx.jpg
 
Last edited:
LOL! I wonder what's happened to

Not sure I see how that follows?

An improved procedural engine, creating more diversity and less repeated tiles could surely also create superior outcomes? All depends on the quality of the rules used?
Have to say I do find the idea of predefined height maps being used a big disappointment. You need some handcrafted features that you can then scatter, like boulders, but on the overall geography? You'll need specific features that the base procedural generation won't come up with alone, like craters, but I assume a procedural crater generation algorithm is perfectly possible for those. Ditto with mountains with specific features; even less need to use it for them, unless you're going for features like long broken escarpments, which might be tricky to generate in any other way (not seen any of those).

Still think on average Odyssey planets look rather better though.
 
Have to say I do find the idea of predefined height maps being used a big disappointment. You need some handcrafted features that you can then scatter, like boulders, but on the overall geography? You'll need specific features that the base procedural generation won't come up with alone, like craters, but I assume a procedural crater generation algorithm is perfectly possible for those. Ditto with mountains with specific features; even less need to use it for them, unless you're going for features like long broken escarpments, which might be tricky to generate in any other way (not seen any of those).

Still think on average Odyssey planets look rather better though.
Actually that's what Odyssey does:
Proc gen heightmaps + handrafted prefabs for details.
There are just not enough and varied prefabs to keep the illusion alive atm.
 
Actually that's what Odyssey does:
Proc gen heightmaps + handrafted prefabs for details.
There are just not enough and varied prefabs to keep the illusion alive atm.
The problem is that I can't see it using the handcrafted ones where there's any obviously clear need for them. Whilst I've seen individual bits of terrain that "feel" a lot more convincing as a hilly but not particularly exciting average planetary location are handcrafted maps necessary for that? Perhaps they are to get some of the nicer features (collections of boulders accumulated in the bottom a small valley for example), but not for the same mountain which looks like it's something that could be produced procedurally anyway.
 
The problem is that I can't see it using the handcrafted ones where there's any obviously clear need for them. Whilst I've seen individual bits of terrain that "feel" a lot more convincing as a hilly but not particularly exciting average planetary location are handcrafted maps necessary for that? Perhaps they are to get some of the nicer features (collections of boulders accumulated in the bottom a small valley for example), but not for the same mountain which looks like it's something that could be produced procedurally anyway.
I bet - just ansumption extrapolating from my experiences in Horizons - the old proc gen only variant looks really not up to 2021 if viewed from a players avatars perspective. So they had to do anything and rising the bar on the proc gen they used would probably not have worked out.

The mountains we see in the landscape today are pure heightmapping, that's why they look like melting ice a bit.

And to get a superduper view from space they also opted for tiles and here is the point where I object to their decisions. At that scale either more handcraft would have been needed or they could have stayed at the old system. But that would have meant possibly running two proc gens depending on your viewpoint ingame.
 
I bet - just ansumption extrapolating from my experiences in Horizons - the old proc gen only variant looks really not up to 2021 if viewed from a players avatars perspective. So they had to do anything and rising the bar on the proc gen they used would probably not have worked out.
The on foot perspective is pretty similar to the SRV one, and when are most players going to be on foot? Hopping out for a quick genetic scan, or wandering around setllements, which are handled separately anyway (predefined is understandable for settlements). I don't see the need for handcrafted heightmaps for either of those. Better ground textures, sure, and we've got them.

Advancements in the planet tech should be welcomed - I believe procedural generation should now be able to do more than Horizons did, I agree it's not up to 2021 (even though in the world in general "not fit for the modern day" usually means "not ridiculously OTT", but I'm getting distracted by my non-game opinions there :) ), but I feel this predefined heightmap method was the wrong way to go about it. On the basis of no experience designing such algorithms of course!
 
The on foot perspective is pretty similar to the SRV one, and when are most players going to be on foot? Hopping out for a quick genetic scan, or wandering around setllements, which are handled separately anyway (predefined is understandable for settlements). I don't see the need for handcrafted heightmaps for either of those. Better ground textures, sure, and we've got them.

Advancements in the planet tech should be welcomed - I believe procedural generation should now be able to do more than Horizons did, I agree it's not up to 2021 (even though in the world in general "not fit for the modern day" usually means "not ridiculously OTT", but I'm getting distracted by my non-game opinions there :) ), but I feel this predefined heightmap method was the wrong way to go about it. On the basis of no experience designing such algorithms of course!
As "Legs" was the main selling point of the whole DLC, one might think they would make that part look as good as possible. :)

Again, sorry for being a bit unclear with too long sentences on my part.
I think we are both in the same ballpark, I just fail to bring a point across that's not too important.
 
Still they did. On stations and from local lights I see multiply shadows. Doing sources and lights is easy, but casting multiply shadows is what gonna tax your system hard.
( Off-topic so hidden: )
I need to do some comparisons in various indoor locations but I've seen at least one place where this didn't work.

One of the Odyssey science lab modules I visited had transparent glass panels with a triangular patch of frosted dots in each corner, and one of these corners happened to be right in front of a point light source -- a little wall-mounted LED-ish thing -- on an adjacent wall. It cast a really beautiful pattern of large circles into the room on the other side of the "glass", falling on pillars and consoles, and I was really impressed with the lighting until I realised the only reason I could see it from where I was standing was because my avatar wasn't casting its own shadow. In fact, looking down at my own faint shadow gave the distinct impression that it was being created by the local star and not by anything inside the structure, although I guess it could have been from a higher-priority light source, maybe a ceiling-mounted panel.

Not a game-wrecker by any stretch, but a curiosity. The sort of thing you don't notice until you notice you've noticed it. In fact until seeing the shadow-circles I hadn't even thought about whether or not I was casting interior shadows even as I skulked around in the darkness with my flashlight, and there are obviously times when performance needs to take priority over prettiness. Just trying to imagine the number of light sources the game would have to process for "full" realism inside structures -- suns, stars, exterior and interior lighting, consoles, fires, flashlights, weapons fire, hovering ship lights -- boggles the mind.
 
The problem is that I can't see it using the handcrafted ones where there's any obviously clear need for them. Whilst I've seen individual bits of terrain that "feel" a lot more convincing as a hilly but not particularly exciting average planetary location are handcrafted maps necessary for that? Perhaps they are to get some of the nicer features (collections of boulders accumulated in the bottom a small valley for example), but not for the same mountain which looks like it's something that could be produced procedurally anyway.
Ultimately, if/when we get atmospheric worlds they're going to want to do better than (cleverly used) perlin noise style height maps. If you look at Earth height data you can't avoid noticing the paths of rivers/glaciers carving through things -- there's a sense of flow to stuff. Although I believe there's a mention of the Odyssey maps being twisted and distorted, that might give you a nice rough version of that.
 
( Off-topic so hidden: )
I need to do some comparisons in various indoor locations but I've seen at least one place where this didn't work.

One of the Odyssey science lab modules I visited had transparent glass panels with a triangular patch of frosted dots in each corner, and one of these corners happened to be right in front of a point light source -- a little wall-mounted LED-ish thing -- on an adjacent wall. It cast a really beautiful pattern of large circles into the room on the other side of the "glass", falling on pillars and consoles, and I was really impressed with the lighting until I realised the only reason I could see it from where I was standing was because my avatar wasn't casting its own shadow. In fact, looking down at my own faint shadow gave the distinct impression that it was being created by the local star and not by anything inside the structure, although I guess it could have been from a higher-priority light source, maybe a ceiling-mounted panel.

Not a game-wrecker by any stretch, but a curiosity. The sort of thing you don't notice until you notice you've noticed it. In fact until seeing the shadow-circles I hadn't even thought about whether or not I was casting interior shadows even as I skulked around in the darkness with my flashlight, and there are obviously times when performance needs to take priority over prettiness. Just trying to imagine the number of light sources the game would have to process for "full" realism inside structures -- suns, stars, exterior and interior lighting, consoles, fires, flashlights, weapons fire, hovering ship lights -- boggles the mind.
Well spotted ;)
 
Still they did. On stations and from local lights I see multiply shadows. Doing sources and lights is easy, but casting multiply shadows is what gonna tax your system hard.
Because the engine is basically creating a planet from scratch 60 times a second, I wouldn't be surprised if there's some sort of clever trade-off in the terrain shaders to make this possible. Like if you only have one main light source it renders 10x quicker or something silly. But I don't really have a clue.
 
Ultimately, if/when we get atmospheric worlds they're going to want to do better than (cleverly used) perlin noise style height maps. If you look at Earth height data you can't avoid noticing the paths of rivers/glaciers carving through things -- there's a sense of flow to stuff. Although I believe there's a mention of the Odyssey maps being twisted and distorted, that might give you a nice rough version of that.
Yes, those are a challenge. You can either seriously cut back on the rivers (make them very rare, with few or no tributaries), or have the same sorts of patterns, which will be obvious very quickly. I don't know what the solution will be there. But if predefined heightmaps are the only way for that, well, use them there. Don't use them where you don't need them.

It's something I've occasionally pondered about but haven't managed to come up with anything any good, but I'm sure there are people with a lot more of an idea and experience than I have looking in to it.
 
Talking of shadows and light sources, if you hold up a candle in a room with the main light on, how many shadows do you see?

It's not very often that you have two stars of the same luminosity in the sky. If they are, they are close enough that the shadow edge would be softened, not separated. In other cases, the additional of the shadow would really add very little to the scene. You might see a slight darkening if you look really closely.

There is also something not quite right with the apparent luminosity of stars - which then produces confusing issues where the closest star appears light but is not the primary light source. If you look at an Type L Dwarf through an atmosphere, it looks as bright as a Type G star, but then leave the atmosphere and it turns into a ball of red/purple.
 
Back
Top Bottom