NOPE

I wrote the software where I can drop in any ship model and it proceduraly sets up all of the colliders, cameras ect. literally takes less than a min to be flying a new ship. Also can put ships on the surface already and traverse multiple ships even though I don't have that in that video.

 

Why am I suddenly reminded of No Man's Sky?

 

Its similar in some regards the difference being mine will have a story. NMS tried to do too much and failed at everything what I have in mind is much more limited in scope. NMS is a good reminder of what not to do. Feature Creep.

 

Since you tagged me so much, here i am responding. I don't want to start comparing games because i know where it always leads to.
KSP does have some pretty nice visuals for the game that it is (heavily focused on physics and spaceships building). Sure there are space games out there that are much better good looking, like star citizen, but KSP isn't as bad as you claim.
It does have effects, atmospheric scattering, re-entry effects and also has a cockpit fps view.
The planets don't need to always occupy 1 million units in a game to be considered planet size. That's where visual tricks come in, to improve performance of all these systems.
You can walk or drive around any planet in KSP, and you will see the size of the planet is real.

Here are some images of KSP





Now, before posting the first time, i didn't see that you already had progress on your game, and kudos to you, it is looking promising for such an early state, and i see that you're on a good way and hope you can do an awesome job with it, but for now, in my opinion, it doesn't reach KSP's quality.

 

It's so easy to look at someone elses work and nitpick. Actually sitting down and doing is an entirely different ballgame.

OP will not come close to KSP any time soon. You know this, look at how long you worked on No Heroes. Last I saw it was looking great, but that didn't come easy and the learning process wasn't overnight.

And I mean - come on - look at how crappy No Heroes is!



That grass is in no way as good as infini grass!

And look at those fake clouds! It's so 2d!

 

I agree, I need super hi res textures. Also I'm one man they are many who made that game. I wouldn't even try to build that simulation solo.. it would take to long to test for one.

It is possible to do better than kerbal if that is ALL you are doing, is mostly graphics (solo). My simulations won't even be nearly as complex, but they will be more "cinematic".

Also, I will be giving away my game in early stages for free so who will be able to complain.

There will be "shooting" in my game, likely on ground and in space, but it will be arcade style and simplified.

All I'm telling you to expect from my game is an original "short" story, and more of a cinematic experience.

Even if I don't end up having full planetary terrain, I can still do what I was doing which is to have key locations on planets surface terrain-ed using unity terrain. the rest of the planet can be tessellated with a custom tess collider.

That would save a lot of time if I went that route, it just wouldn't be as impressive as far as seeing the full planet terrain as you approach the surface, but it would free up more time to work on the controllers for sure and the weapons and fighting systems in space, possibly even some type of ground fps where you fight hordes of enemies almost like a tower defense game .

And im thinking there will be only one resource in the game Energy, so yah I will be keeping it super simple to start with but fun! In a sense im thinking a type of space survival game where your energy is always running out and you have to constantly explore to find more of the energy resource to power your ship, to have ammo for your weapons and hostile AI's are constantly hunting you, in space and on the ground.

I have no plans for feature creep beyond that unless people like the concept of the prototype and think it has a shot to become more than that.

Which brings me to this post, the reason im going back to doing scenes, to get at least decent at building scenes because I can use these on the planets surface and the skills I gain for making the "key locations" on the planets surface at least decent for the prototype.

 

And I only count 2 or 3 terrain textures in that screen shot.... it would be so much better if it had like 70.

 

Yeah for real!

Seriously though, @eskovas do you have any posts talking about your techniques for the recent terrain work? The recent stuff you're doing is crazy and really is top shelf quality.

 

It's not that tricky. It's easy with a flat surface. I wrote a huge tutorial series on it years ago based on XNA. The basis is that it used a Quad Tree to represent the terrain and a heightmap to generate it.. The camera position was fed to the quad tree which subdivided the terrain by quad and was most dense around the player. It stepped out and eased the subdivision the further away from the player you got.

It automatically LOD'd the terrain in real time and as you moved it would recalculate the terrain mesh, including terrain normals. It was configurable to specify minimum LOD levels and it also did view frustum culling. The same principles apply in Unity and you could accomplish the same thing using an Octree instead of a QuadTree for your geospatial partitioning to get a curved terrain. You would want to transition to a sphere mesh at a certain distance... using the above method to try to generate the entire sphere would result in a lot of pinching and stretching, but it's doable.

 

I always feel weird when someone brings up my game
That image does have plenty of issues, like very visible grass patches, not enough grass variety, "weird" transition between grass and not grass, too much "fog", very few detail objects, etc.
It's a constant learning process, especially on the environment and level design, which is pretty difficult to get it right.

In this shot you can probably distinguish between 3 terrain textures. I'm using a total of 8 for the terrain (with RTP). But for the most part, and for what i'm doing, 8 is enough.
I'm still contemplating MegaSplat with it's awesome 256 texture limit and nice shading, but for now i'm pretty happy with RTP. Would be awesome for more variety.

 

I talked to the megasplat guy - it sounds very, very impressive. I am still torn though, I have so much stuff integrated into RTP - it would be a nightmare swapping out. He's apparently working on foot placement for tesselated terrain, which can create some weird problems with sharp terrain features and shows that he's not just looking to make a cool demo, but really make it game usable.

 

Oh man, this thread made me happy today

I remember my cryEngine days: drag and drop, drag and drop, Mess with terrain, drag and drop, lightning, drag and drop... and after 6 months, I had an beautiful nothing. I had 13 years old, and I thought was beautiful...

You should read every word from this community, and be realistic(in real life): you need to learn!

I know that it's funny creating this scenes and bla bla bla, but what makes these scenes... YOUR scene? Rearrange assets is not art, it's spend time with other people's work.

I'm not saying to you stop making this kind of scene, but you need to walk before run an jump.

You can't just use assets to make a NEW game. That's the first law. You need to put your identity, something that the player will know what game he is playing when see the game. And when you discover your identity, you will be ready to create a game.

For now, forget about reinvent the wheel. I suggest you to create a low poly scene, creating your owns assets made in blender(you are a programmer but why can't learn modelling?). And put on a mobile device, an SLOW mobile device, making it runs smoothly. You will discover a lot of things by just building for mobile, I'm sure...

You broke my heart when said that unity is prevent solo developers to create creative works. Man, this is wrong! I didn't have any problems with unity's limitations. The terrain tool is not made for creating a high detailed scene(my game speeds up a lot when I deleted my terrain and replace with an mesh with the same format, color, texture, etc...). You can make incredible things on unity, and the majority of developers of unity are indie, like you and me.

Sorry, but your scene is far from the witcher 3(or 2). It needs an essence, and essence that a bunch of assets packs won't give you. Do you thing that my words are wrong? Take a look on the best games made with unity on 2016. Start from scratch again with an modest dream, and maybe you can create an game as funny as flappy bird, or as perfect as shadow of the colossus are. But if you will keep your objective as it is now, just don't judge unity as an bad engine or with bad tools.

Remember: we are here to help people that wants to listen what we are talking, not what they want to listen.

 

Ive learned quite a bit about modeling enough to know that it isn't my thing I don't enjoy it in the slightest. It is very tedious and boring, and I am not an artist.

Building controllers is fun, especially space sim controllers, building the framework to run the game is fun.

Here is a question for you, how exactly do "programmers" that work for studios "get to where they are" because they obviously aren't 1 man engineering teams.

Its kind of ridiculous to think someone can be good at modeling, programming and level design unless they have been doing it their entire life and had 10-20 years to master all of the trades.

I want to transition from web development, information database systems into game development because I am just burned out with it. I have been writing web based software for 15 years.

I don't have "years" to spend tinkering with some game and modeling program and I have no interest in becoming a professional modeler. I want to build out a bunch of "mini" games that are doable with other peoples assets mostly to get the "coding experience" with controllers and eventually multiplayer architecture (APIs).

I was thinking that a coder would need a decent understanding of level design, but not a "mastery" of it.
A decent understanding of modeling but not a "mastery" of it.

If someone said hey I want you to code a controller that interacts with this level in a certain way that is where "my specialty" would come in as a controller programmer along with eventual knowledge of multiplayer network API's.

So all I'm saying is that what I intend on doing in unity is more for the purposes of getting to a certain place on a coding level. I just want decent backdrops for my programming work to shine if you get what im saying but Im not sure if I have the intention to be a 1 man engineering team.




Early on I realized that is the first skill you would need to build a full and complete game, and a few years if your working alone. Some models can take months to build especially if you arent a professional modeler with 15 years of experience.

Realistically I am only interested in seeing what is possible in unity and building prototypes.

 

And Jesus Christ guys I was only REFERRING to the GRASS in that scene.

Witcher 3 does not have realistic grass rendering 200-300,000 individual blades of grass.

In fact very few games deploy direct11 grass. That is all I was really referring too.

Sure I could of spent another day just working on improving the grass but I am moving on. My only goal there was to figure out how to implement the direct11 grass onto Unity terrain, easily.

 

Did you use that DirectX 11 grass asset on the store? I keep seeing it but never get around to trying it out.

 

Yup, the thing about it is that it does use the hardware for it to be useful you need a lot of grass for it to look realistic and therefore will put some limitations on what you can do with the rest of your scene if you are wanting it to run on low end hardware.

I want a to build some scenes using it and I could care less about the hardware limitations. It does look way better than unity grass. Though after seeing how Gaia places its unity grass which is pretty good i can say that unity grass can be good but it will never look as realistic as that grass shader no matter what you do to it but unity grass is much faster hardware wise.

 

Just remember, the fastest way to improve always involves being realistic about what you're doing and being open to learning from examples that outclass your skill.

I think one of the biggest skills tiny indie efforts need is understanding and identifying things that matter.

You will not be able to impress gamers with visual fidelity. Only the AAA space can really do this. The most you can hope for on the tiny indie level is that screenshots or videos look decent and that it doesn't remind people of asset flips or other bottom of the barrel efforts.

Gamers won't care if you have a half billion blades of individually modeled grass. Truth is, after about 30 seconds, they won't even notice. Except for noticing problems:
- "the grass is cool, but it looks really stupid when a guy steps on it and it sticks through his leg"

Higher fidelity means higher expectations. It's a weird thing, but basically - I think it's better to have even levels of fidelity and polish. If you can't polish the behaviour or the game play to the same level of fidelity of the visuals, there's a dissonance for your players.

In your space ship example, I think that having a low poly ship interior dramatically helps the presentation. It kind of fits with the lower resolution textures on the planets, and it makes some of the glitchy stuff more forgivable.

If the ship interior was super high fidelity, many of the other problems would suddenly be more glaring and more distracting. Your demo video would feel much worse if the ship had higher quality textures or models.

This is one of the big give aways with mid level asset flips, many of them are using some pretty nice looking environments (these are often cheap and easy) - but all that those nice looking environments really do is highlight the fact that everything else in the game is a f'n mess: the ui sucks, the animation is weak, the controls aren't responsive, etc.

I think many, many people misunderstand the real cost of higher fidelity. Each step up in detail isn't isolated, as you end up raising the expectations of your players for everything else, and all the different parts of the system suddenly need to be better, smoother, richer, higher fidelity and more complete.

 

No, the first law is if you try to do everything yourself, you will fail. Assets have their place as long as you use them properly.

 

Well, I tried the DX11 grass and actually, while a clever solution, I switched to using billboard grass for 2 reasons:

a) big performance drops - this has to do with my scene being isometric and thus without a large grass distance the "disappearing grass" would really stand out.
In a Firstperson or thirdperson view you could lower the grass distance. Still, without a good system for far grass this is going to look weird.

There are good solutions for "shell based" far grass systems, but the DX11 grass, if it is the same system I tested, currently does not do that.

b) horrible aliasing even with SSAA. Of course, this again is not so much a problem if you can use forward and MSAA, but in deferred the AA options are more limited (SSAA can be better than MSAA, but only if you can take a big hit to your framerate).
Also this is a general problem with transparent or cutout geometry. But the DX11 grass was worse in this regard than Unity grass, most probably because of how the geometry is created with polygons instead of using billboards. Small polygons produce horrible aliasing, especially if combined with specular highlights. Now add vertex animation to this and the grass quickly becomes a horrible aliased mess.


Add to this the fact that you need to duplicate the terrain geometry, and the DX11 grass suddently isn't so hot anymore outside of specialized scenes and tech demos. Cool if grass plays a center role in our game and you can waste most of your framerate on good grass... if you want lots of animated characters, and tons of prop meshes in the scene, you probably are going back to simpler grass if that gives the more important parts of the scene more breathing room.



SLI -> it is dying. Multi Card with DX12 / Vulcan, yeah, might be a big thing in the future. As long as the devs are not required to write special profiles just for the few rich guys that run triple SLI setups for their benchmarking e-peen and being able to complain online how bad SLI support is, it might be a thing.


Terrain -> if the stock terrain shaders are not good enough, and you are already spending big in the asset store, why not look at one of the MANY terrain shaders that actually improve the look of Unity terrain? Have a look at RTP and tell me again that Unity terrain looks like crap. Currently supports up to 8 Textures in a single pass (if you can live with no heightblending between layers 1-4, and layers 5-8), and up to 12 in two passes.
Does heightblending, parallax shading, procedural snow, wetness, and about a 1000 things more that turns the dull old Unity terrain engine into a next gen monster.



You currently seem to be EXTREMLY focussed on the graphics side. Its a common mistake of newbies to be so graphics fixated, and try to outdo AAA games as lone wolves. A lot of us were there, done that.
Just keep in mind that a) a game is more than just pretty grass, if you spend years fixated on perfecting the look of your grass you will never finish anything, and b) AAA games are built by 100's of expierienced devs labouring away at it for 4+ years, not a single inexpierienced dev that is talking about how he spent 20 hours building something (a professional might spend multiple months on a single level, full time).
If you like lofty goals, sure, take Witcher3 as your inspiration. Just be prepared that at some point, either you hire the big team needed to really get into the reach of a Witcher 3, or scale back your ambitions to be able to finish your project in a single lifetime.

 

-deleted image of my grass cuz i think it still sucks too much- ;D

 

It's great to post early works and compare yourself to higher levels of quality. Don't get discouraged! The only way to get better is to gather feedback and set achievable goals.

 

Agreed.

Gonna stick this in spoiler tags since it's OP's thread about his scene, not mine.

Spoiler

The way I approached the 'grass problem' was not to use higher fidelity grass or better techniques, but instead to throw tons and tons of quantity at it. This grass is pretty high density, and it's everywhere.

This is the picture that I posted above and deleted:


Looking at my screenshot above, it's just a blurry mess (@gian-reto-alig mentions bad aliasing), and the different grass types look terrible together (@eskovas mentions how his grass needed more varieties). Maybe it's not that bad, but it's not good.

So I tweeked it a little after posting:

This is the same density in the same scene, just with a texture swap and some minor tweeks to color, etc.

The big improvement was using a much, much lower resolution grass texture. Originally I was using higher res 1024 grass textures, but the fact is - you aren't looking at grass close up, you're looking at tiny, scaled down grass. The extra details (at the sizes I was using) end up just making it blurry.

So I went with a 256x256 grass texture instead, and I think it dramatically improved. Counter intuitive for sure. But also a little softer on the gpu, so a win all around. The biggest win though, will be in youtube video, since the video compression artifacts completely obliterated the previous grass (this should hold up much better).

edit: mipmapping should mean theres no real performance difference, and the resolution isn't meaningful - the real difference is the grass itself in the low res version is hand drawn and includes sharp clean edges and the like - so things stay crisper when scaled down more - kinda like a good icon - you need clean edges and simple clear shapes.

Still needs work, still isn't perfect, definitely isn't AAA, but hey, I got something positive out of this thread

 

Oh I just got Colormap 3.0 terrain Parralax and Tesselation shader along with Gaia.

Im building a terrain right now that will use heavy terrain tessellation on slopes as well as directX 11 grass. I am not building it to run on your grandmas Pentium 400mhz.

I will post it back here when im done probably by the end of the week, after all making a "top tier" terrain takes work even using procedural tools.

 

No. You're just targeting SLI which even the PC Master Race and NVIDIA fanboys have started recommending against.

 

No you will be able to run it on a single Gtx 550+ at probably around 20-30fps, Im going for minumum 50FPS on a Gtx 770, so I will optimize it to run decently on a mid grade "budget" card and very well on a high end card.

 

I happen to believe there isn't enough content to push your graphics card to the limit. And so when I create this terrain I will be tesselating heavily, also watched a video on all the techniques fallout used to place mesh around thier height map. Fallout uses so many meshes but ALOT of them repeat to hide heightmap slopes, thats pretty much where the "work" is after the proceedural generation of the terrain.

I will be using free textures however, but I am making my own custom vegitation by having my girlfriend which is a photographer take pictures of flowers and plants with a backdrop behind them so I can crop them out in photoshop.

So, I will be adding some my own "assets".

 

If you're not pushing your graphics card to the limit it simply means your CPU isn't able to keep it fed.

If you want to push your hardware the furthest you may want to reconsider your choice of engine.

http://answers.unity3d.com/questions/298732/does-unity-support-crossfire-or-sli.html

https://answers.unrealengine.com/questions/21746/does-the-ue4-engine-support-sli.html

CryEngine 5 appears to be the only one gaining support for it.

http://www.tweaktown.com/news/53928/cryengine-gets-multi-gpu-support-dx12-feb-2017/index.html

 

Thanks for the info but I am developing it using one single GTX 770, I have SLI and I will test it out but I'm not going to make it an issue if it doesn't work. If I can get 40+ fps after I'm done on a 770 I will be satisfied.

 

Huh. I was under the impression you were targetting a higher end card than that. That's actually pretty reasonable.

 

It will be occlusion culled in linear deffered rendering and if the tessellation becomes and issue I will setup some custom LOD systems to control the tessellation in real time on various textures and surface mesh.

 

Im very interested in creating a system to optimize and manage the assets I am deploying.


Volumetric fog and mist
ATS colormap shader 3.0 for tesselation and terrain enhancemnt
DirectX 11 tesslated grass shader
Suimono Water System

And Im going to build my own day night system to tie into these assets with a custom editor extension to manage it so I could "sell" it possibly later on, also an API to easily tie the system into a UI options menu + one that will come with it.

That is my goals for the project, the best looking terrain I can come up with, plus creating some of my own mesh and vegetation assets.

 

This is what I am attempting to duplicate as closely as possible. I decided its best for me to have an image for a frame of reference so I dont waste so much time trying to decide on what I want to do with the terrain to begin with.

 

they have some great textures here for anyone looking for them for non commercial project. Im going to be using them in this mountain landscape replica.

http://gamebanana.com/textures/4230

 

Suggestions/recommendations:

• Don't stop at one reference picture. Get a variety of different angles and stuff (what does it look like standing within the wooded area? How does the water look when youre close to parallel?)

• Download cinematic post effect pack. Experiment. Color grading & tone mapping is a big deal. Good AO helps too. Abusing bloom & sunshafts is a red flag for noob work, every noob goes through the "omg sun shafts!" phase (I sure did - and you see it again and again).

• Speed Tree modeller is a good deal, a month is affordable for anyone, and you can crank out a bunch of stuff (warning: very high learning curve if you're doing more than just variations or tweeks).

• Don't neglect details. At minimum: dump a bunch of rocks. Tessellation is cool - but big ol rocks add variety.

• Avoid flat ground. Look at witcher screenshots, the only time the ground is flat is when there's combat or buildings. Peaceful nature areas are covered in slopes. Flat ground is boring and unnatural.

• More & bigger isn't always better. Nobody cares about 24 sq miles of boring empty nothing. Details, landmarks, interesting side spots, abandoned camping grounds, a burned down farm house, the details are what capture peoples imagination (your lawn furniture on the moon is more important than high res planet textures - this is what ppl will remember).

 

Prepare to feed the garbage collector. It'll get your 50fps.... in between stutters.

 

Yup, not the terrain is 2,000,2,000, just large enough for the mountains and lake valley.

Im not trying to be 100% here, just very close, as far as the woods I will improvise and I was thinking of making trees with unitys system first before spending more money on it.

I have a pretty good understanding of tone mapping and color correction curves ect played around with them a lot.

 

Not with a custom occlusion culling system, not generating the tesselation on the fly instead setting up custom occlusion zones for tess heavy areas of the terrain, like areas of high altitude (mountains).

 

I will have something to show you in a few hours ! the beginning really but ill have the terrain textured the way I want it and probably the water done like in that image with a matching reflection.

 

This isn't necessarily representative of a scene in an actual game, though.

If you just need to represent a shot from a fixed camera location in a game then you pre-render your image and draw it on a quad. Or if you need a little perspective you use massively simplified geometry and some appropriate sprites. If your player needs to be able to move around within the environment, though, you need to make a lot more than is visible in any one shot of that environment.

In reference to the thread title, "Can your GPU handle it?" is a nice attention grabber, but it's not an attitude I'd want any artist or level designer working for me to have. It's their job to make sure it supports whatever the target hardware is, and I expect them to build things efficiently and use any trick necessary to make it happen. Just making stuff that looks good isn't enough - it has to look good and run fast.

That said, I don't want to be discouraging. You're doing cool looking stuff and clearly have a strong interest in environmental art. That's great, and practice will get you far!

 

Well you have to learn how to build it before you can learn how to optimize it .

I would rather go for best look and optimize after, seems like the most efficient workflow. I am not saying I am not going to optimize I just mentioned before im building an optimization system to work with all of the assets im deploying.

You should see what im working on now, tessellated terrain, with color map + directx11 grass. It is taking me a bit longer than I thought to work out some of the tessellation specular issues.

Probably in day or two ill post a new thread showcasing my Colorado mountains close replica scene, its 2K/2K terrain.

I may share it before I add mesh to the terrain. It is running extremely fast as well already.

I will be putting the controls to manage the tesselation in an options menu so you can adjust the render according to your hardware.

 

Research and consider photogrammetry in addition to and possibly instead of this photoshop crop technique. Unless you already have experience in cropping out stuff to add to a scene as billboards - this technique might reduce the overall visual quality.
Photogrammetry can improve the visual quality of an environment a lot and can be done with free software as long as the photographs required are available. This process adds unique content to 'your' scene that no other environment will have, that is unless your girlfriend takes photos of the same dandelions that other people use in there scenes.
The only negative I can find with this technique is (a technique I've only explored in concept) this process only allows the artist to create what has already been created in the real world or in another medium. Custom modeling/painting is required to include exotic fantasy/sci-fi content, but this is not what you are going for.

@frosted is spot on with this. I'll add (my opinion). An artist has to tell a story with an environment. If there is no story to tell besides look how high resolution this environment is (same story - different day) the environment is forgettable.
Environments without life or story are only good for a 20-30 second fly through at best or a still image that is forgotten 10 seconds after it is shown.
Without moving parts, dynamic "life" in the scene the bestest, greatest terrain environment with realistic everything is nothing more than a wall paper.

Show the viewer a functional alive environment that has life happening with or without players in the scene and you create interest in these environments. Make the environment ALIVE! Make the environment a character rather than a place the character occupies.
Tell a story with the environment - without any character in a scene will be remembered.
I would rather see an environment with a purpose or that tells a story with life, moving elements (not just water and wind) with lower resolution models and textures, than yet again, another static ultra realistic 20 fps high resolution, static dead scene.

At this point you aren't doing anything 'different' from 500 other noob environmental artists with access to the asset store. No offense intended.

 

That's Moraine Lake in the Canadian Rockies.

 

Well I did a search for colorado mountain lake in google images and that is what came up but thank for letting me know.

Man getting the colormap ultra shader to work right is a pain. Im having issues getting it right and there is very limited examples and tutorials. I think my issues have to do with setting up the textures normals.

I am using crazy bump and it looks right but cant get it right in unity. I am going to keep at it though I dont give up.

 

All in good time, I'm still trying to master the colormap ultra shader and setting up the textures properly for the tesselation to work and look good.

Right now my only focus is to learn how to make the tesselation look good with the textures I have, after that im going to get into making seamless textures. I have a lot of experience in photoshop being a web developer for 15 years.

Thanks for the useful advice, though. I know the terrain itself or "heightmap" as its called in the pro world isnt the end all be all. All of the big AAA games, you dont really see the terrain heightmap much as they are covered with "sign posts" and hard and soft barriers. Which is where the real work and time will be spent after the terrain is ready.

I fully intend to make my own mesh assets for this terrain too. Im not a modeler but I do have autodesk with a rock generator with noise algos .

I will share my terrain though before I start adding the mesh and accept the "criticism" from you pros.. after all I'm a noob.

 

I'm no pro yet. But I've been hard at it for a long time, I definitely started with the same "I can do better" attitude you have. Years later, reality beat that attitude out of me. I am really critical of people over reacting to those first few moments of scene building where you make insane progress in a few hours and think "if I did this in a day, what could I do in a month!?" - because I am one of the guys who fell for it

That said, a solo dev can really do amazing stuff these days with a tiny budget. But it takes a ... crazy ... crazy amount of learning and it's not remotely linear progress.

 

Are you using Windows or Mac byte order? It makes a difference.

 

After getting into terrain tessellation shading im starting to get that. It is definitely not something you can do do quickly unless you have alot of practice.

Its not that "i can do better" its more like I didn't invest so much time into this to produce mediocre results. I'm willing to do what it takes to create fantastic results even if I have to put in a hundred hours a week learning and "doing", creating various scenes.

 

this is indeed a funny thing, if you start out with the engine and all the prefabs you can bring together an okay looking scene in a day, easily
as soon as you start to create your own prefabs you start to understand the scope of the things you want to do...

 

Just keep in mind, fantastic results may take many thousands of hours. Although it depends on "fantastic" and "results" for sure.

In my personal experience, the thing that gave me the most trouble was learning UX design, typography and layout. The big stumbling block for me there was incorrectly believing that I was good at it when I started. Since I incorrectly thought I was good at it, I never really gave myself the time to learn it properly, and my results kept sucking. Finally I started to improve, and now I can do a decent layout.

The single biggest problem I had there was that I kept thinking "but I'm good at layout and ux!" - this poor self assessment cost me an insane amount of time where I produced terrible work but at the same time, didn't invest the time in really learning the fundamentals I needed to in order to get better.

It's kind of like those 12 step programs, "the first step is accepting you have a problem", with learning the first step is accepting that you don't know what you're doing.

 

That is something I also had to find out. For my usecase, where because of the camera angle you couldn't get close to the grass, using a lower resolution texture works BETTER. Going with 64x64 textures for individual strands of grass or flowers did the trick in my case. Looks good with the isometric view I am having, and is much quicker to paint by hand than a HD grass texture (takes me about 15 minutes to paint a quite detailled low res flower in Photoshop thanks to the low resolution... "quite detailled" sounds stupid for a 64x64 pixel texture, I mean with full shading, and detailled leave shapes and all).

Always find out what looks good enough for your usecase, and create just that, and nothing more. Overengineering is the root of all evil, that seems to hold true for art just as much as for tech stuff.



Anyway, its derailling the thread a little, so I will stop here (even though the "KISS Principle statement" of my post seems to fit the thread and the general focus of the TO on a single aspect of his game over all others).