Відео

True I actually missed the line with killing my own doorkeeper

True, looks like I forgot to update the title on this one. Thanks!

Hmm I think if you are using local position, you can probably just rotate the mesh itself without changing the offset. I think you'd only need to take offsets into account if you are using world position. I could be wrong though I havent tested it.

Thanks, looking forward to getting some interactivity working

Unreal Engine + Fast Noise 2 library, I prototype the noise formulas in the fast noise editor that comes with the library then convert to c++. rest of the planet mesh gen stuff is custom code I wrote which makes use of the Realtime Mesh plugin for unreal engine. Fast Noise 2 is a cmake library so I am also making use of the ue4cmake plugin to get it to build properly.

The material contributes to the surface color but there is also a directional light

Dunno, I don't do any moderation on the comments so if it was it wasn't by me.

was able to take the unreal engine perlin noise and replace the function with fast noise 2, I guess at this point it's all just generic noise terrain math stuff. I should probably bring the object creation out of the lambda though. auto GetDisplacedPosition = [&EditMesh, &Options, &Offsets, &Normals, bAlignWithNormal](int32 VertexID) { auto PerlinGen = NewObject<UFastNoise2PerlinGenerator>(); FVector3d Pos = EditMesh.GetVertex(VertexID); if (bAlignWithNormal) { FVector NoisePos = (FVector)((double)Options.BaseLayer.Frequency * (Pos + Offsets[0])); float Displacement = Options.BaseLayer.Magnitude * PerlinGen->GenSingle3D(Options.BaseLayer.Frequency * NoisePos); Pos += Displacement * Normals[VertexID]; } else { FVector3d Displacement; for (int32 k = 0; k < 3; ++k) { FVector NoisePos = (FVector)((double)Options.BaseLayer.Frequency * (Pos + Offsets[k])); Displacement[k] = Options.BaseLayer.Magnitude * PerlinGen->GenSingle3D(Options.BaseLayer.Frequency * NoisePos); } Pos += Displacement; } return Pos; };

Nice, so in order for me to use the fast noise 2 library in C++ I used a plugin called ue4cmake, which allowed me to integrate the cmake fast noise 2 library. The fast noise 2 library comes with a noise editor which will allow you to play with the noise nodes in a more visual scripting environment, and also helps to see how to combine the nodes and come up with different noise formulas. Then after coming up with something that you like, you can go generate the same noise formula in C++ by chaining the nodes together. Currently I generate reasonably convincing terrain by focusing on 4 different noise layers, 1 for continents, 1 for mountain ridges, 1 for mountain peaks, and 1 for valleys. The full noise formula is around 200 lines in C++ but building them is pretty verbose, I think I ended up with around 35 nodes total, that includes basic operator nodes like remap/multiply/add etc. Also I will note, the hardest part was not actually even making good looking terrain, it was actually combining all the nodes together in a way that had A) Consistent output value ranges, and B) Predictable control over sea level

I've done a code breakdown before on the C++ components, and eventually when everything is done I may look into putting it on the marketplace... there are probably too many moving parts to easily duplicate it from just like a tutorial or description though. Several plug in library dependencies, somewhat pain in the ass project and build setup to use the fast noise 2 library and some other things like that. That said, the concepts I use in the noise formulas should be universal, so a similar formula can be constructed in anything with coherent noise generators. The fast noise 2 formula I'm using currently was actually adapted from a procedural planet material I made a while back. Basic concept is constructing separate layers for Continents, Mountain Ridge, Mountain Peaks, and Valleys and then composite them together into a final height map.

All of it, I set up a prompt in chat gpt to generate a selected style, tone, title and lyrics based on a selected subject matter, then feed them to suno ai

Yes it's a solid combo deck, especially with the new MDFC lands

Thanks!

Hmm I mean, I did climb to top 200 with it so it can't be THAT bad. Probs just not for everyone. Frustrating into certain decks - wizards, rw energy if you don't get good luck on mills.

@@snznz Well, Idk what luck is. But after 20 games i did't find a single Omniscience.

Hmm the main issue is that at large scales the raytrace sphere logic creates jagged steps/artifacts in the mesh itself at close viewing distances, so while you are right that I could work around the problems when you get far away, the main issue is the offset precision issues up close. It is an interesting technique though. If i could get a double precision ray sphere intersect within a material it might work, but I haven't seen a good way to do it. Other option could be a compute shader but I'd have to look into it.

It's all ai gen shlock, I just told chatgpt to make lyrics for different planeswalkers and fed it to suno ai

Deck 3 A-The One Ring (LTR) 246 2 Search for Glory (KHM) 27 3 The Book of Exalted Deeds (AFR) 4 4 Chance for Glory (GRN) 159 4 Gideon of the Trials (AKR) 19 3 Surge of Salvation (MOM) 41 4 Needleverge Pathway (ZNR) 263 4 Sacred Foundry (GRN) 254 4 Alpine Meadow (KHM) 248 1 Faceless Haven (KHM) 255 4 Mutavault (M14) 228 3 Sundown Pass (VOW) 266 4 Snow-Covered Plains (KHM) 277 1 Approach of the Second Sun (STA) 1 2 Settle the Wreckage (XLN) 34 1 Farewell (NEO) 13 3 Karn, the Great Creator (WAR) 1 3 Divine Purge (Y22) 4 4 March of Otherworldly Light (NEO) 28 3 Temporary Lockdown (DMU) 36 Sideboard 1 The Book of Exalted Deeds (AFR) 4 1 Tormod's Crypt (M21) 241 1 Pithing Needle (MID) 257 1 Liquimetal Coating (BRR) 28 1 Cityscape Leveler (BRO) 233 1 A-The One Ring (LTR) 246 1 God-Pharaoh's Statue (WAR) 238

So we are making use of Priest of Titania, so we actually care about our elf tribal-ness, which is why I opted to drop the halfling to fit more elf dorks

IKR its nuts

Yea, if I was going to do it again I'd probably use some form of raymarching

Was messing around with different ways to integrate the depth into the wave calculations. Turns out if you base you wavelength on a fraction of the sphere radius, and you lerp that a very small amount with the vertex depth, you end up with this. So something like: wavelength = lerp(radius, depth + 100, .1) * scale. Still trial and erroring other ways to integrate depth into the waves.

@@snznz Interesting

maby ur right maby i dont now wath it means

It's generated at spawn time currently. It is possible to write functions that would modify it at run time, but to get true voxel style editing you would probably want to convert the lowest level chunks into a voxel format using something like dual contouring. It's something I'm interested in doing but probably won't get to for a long time.

Thanks dude much appreciated 👍

Nope, I am using scene depth - scene depth without water nodes

@@snznz you might get better directional controls without the depth issues as a bonus if your terrain generates distance fields.

Yea, I haven't been able to do it in a way that is performant enough at this point, but I may look into it more eventually.

@@snznz if only there was a way to precompute the DF and then mask it out over distance.

Ended up calculating the depth at mesh construction and encoding it in the vertex color channels, seems to work well

hey im have problem how to fix im have everyting what in need but still dont working [SM5] /Engine/Private/PostProcessMaterialShaders.intermediate.usf(421,9-16): error X3511: forced to unroll loop, but unrolling failed. [SM5] /Engine/Private/PostProcessMaterialShaders.intermediate.usf(489,21-85): error X3511: unable to unroll loop, loop does not appear to terminate in a timely manner (477 iterations) or unrolled loop is too large, use the [unroll(n)] attribute to force an exact higher number [SM5] /Engine/Private/PostProcessMaterialShaders.intermediate.usf(492,39-124): warning X3570: gradient instruction used in a loop with varying iteration, attempting to unroll the loop

Yea hard to say, I'd double check that all the variables are named correctly and have values around what is in the video

Yep

@@snznz ty, do you have a link from where you started the atmosphere? If not I can find it myself ty

@@JesfNwje Not sure what you mean, the custom node code is linked in the description and the video is a guide on how to set it up

@@snznz oh ty

Sure, I'll take a look! My videos here are a bit behind the stream, I have fbm sphere Gertsner waves working now but haven't done any work on making the shader actually look like water yet... Also been having a difficult time getting distance fields to generate properly for the mesh planet which has been a blocker for things like shoreline foam.

​ @snznz distance field mask is very imprecise at high distances, you can use Depth Behind Water node to create a proper mask (available in Single Layer Water shading model), but that's only pixel shader (you cannot fade off waves based on it). There's DC on the marketplace page, if you join I can show you the mask. My comment keeps being removed by UA-cam for some reason

Thanks dude that's actually really helpful info, since even if I could get it working, its expensive to build the distance field during mesh gen... I'll check out your discord when I get a chance. I did try just using SLW shader to see if I could just throw it on as is, but it has weird Z axis issues when I wrap it around a sphere.

Yea his video is very good but did not go into LOD so it really only works for "toy" sized planets where you can get enough detail. I can do an updated planet generation video at some point, but... It is not a simple process and you will need a couple plugins, I use the Realtimemesh plugin to handle mesh generation, so you would want to get familar with generating mesh with it. You may want to start with just 1 flat quadtree. If you can generate 1 flat quadtree, the planet is just 6 quadtrees composing a cube with their points normalized.. Each node should generate its mesh data off the game thread, then submit a delegate to a queue that is processed on the gamethread to generate its actual mesh component. I have a code review video of my initial implementation here ua-cam.com/video/E8a1WddwQXs/v-deo.htmlsi=0LNAW62TbYjG4UN- but it has changed somewhat.

I just do it for fun. One day Ill finish putting it together into a game. As for the channel, it is pretty much just a stream dump in case I ever want to have the raw footage to edit together. Although I do occasionally put together more targeted videos when I finish something I think is worth one.

@@snznz Thanks for the reply. Any interest in doing paid/contract work?

Potentially, it's tough to break into game dev without a professional background in it

Not sure as I haven't done much research on it but I'm sure I'll come up with something

@@snznz keep me posted

Thanks, dude. It's on the radar

Appreciate it dude!

I use fast noise 2 library for my noise formula and RealtimeMeshPlugin to render the chunks, the rest is a massive amount of grinding my face on C++ code to get it working correctly.

I am building a procedural universe generator with no loading screens/instanced areas. Sort of like a stylized space engine.The planet generation tech is part of it.

@@snznz impressive, good work 👍

Creating a volumetric procedural cloud layer for spherical planets

@@snznz what did you study, are you phd hder or sth like that

I have a BS in computer science and I've been a software dev for a long time, but all my game dev stuff is self taught

Maybe, at the time I was only trying to get something that looks good from space so I never bothered. IIRC there was always a trade off. For example you could make it a masked effect with dithered transparency, and then you should be able to get accurate shadows... but you trade off having dithering translucently. Could try playing with the material settings, trying different translucency types, turning off/on raytraced shadows etc. Alternatively, I am currently working on a volumetric raymarched cloud shader which should be dramatically superior to the shader here (can fly through clouds, they are volumetric). That one also has no surface shadows yet, but I will eventually integrate it into my atmosphere shader and try to figure out how to get the cloud shadows integrated into the atmosphere surface lighting. Whenever I have the basic effect done I will probably put out a summary video, then again whenever I figure out shadows for it.