Separate names with a comma.
Discussion in 'General Discussion' started by crazikyle, Jan 26, 2016.
So far we've only seen a tiny snippet of PBR in BeamNG with the in-game time probably set somewhere later than afternoon to make the screenshot look more like eye-candy, so judging the whole technology from that is like saying you don't like an entire car because the colour of the wheels isn't silver
I'm pretty sure all the materials, especially car paint will look better & more realistic, not necessarily overdone like some shaders for other games
well you got me there, i suppose metallic paint is something i would like i guess. I gotta wait until i see more of it until i can make my mind up
Forgive me but it appears that PBR is less graphically intensive and also more realistic/prettier, so that leads me to wonder, what's the downside?
Look like there isn't really one, except that it needs to be implemented.
Needs way more textures so the game file size increases a lot, as far as I know. For now we have just the Diffuse, Normal Map and Specular Map, with PBR we'll have Albedo, Ambient Occulsion, Normal Map, Roughness, Metalness and ClearCoat. 2 times more, for each car, map, etc. If I'm wrong correct me.
Clearcoat doesn't need a texture, it's just a slider for how much clearcoat there is usually. A metalness texture is only needed for a material that has both metal and dielectric parts, so not necessary for most materials.
I guess in practice, you pay for Adobe for Substance or try to get by with free offerings like:
As with all realism improvements bar goes up and all that comes with that is bad part, for some probably just good higher the bar gets, depends from people and their perspective of course.
It is somewhat natural development though, not fan of it though, but tracks of GPL are not much to look at of course, tracks of rFactor1 are not probably getting much love after bar gets high enough either.
Soon everyone wants only maps with 40 000 000 polygons like in F1 games and you need a modding team to produce any decent content, that is bar getting higher.
This game has been in development 10 years soon enough, I think we are at year 8 now?
So at the beginning game was marketed as one with great physics being focus and it was even mentioned graphics won't be big thing, but seeing game focusing quite lot on keeping sofa players with gamepads happy and bringing new levels of visual quality in, I think they have shifted direction a little from their humble beginnings and it is good as game might be more successful with pretty pictures.
Physics realism is still of course part of development, but lot is done to have game easy to access and very friendly to play on tv with controller, with Linux support maybe even multiplatform possibilities are happening and game in future might have potential to enter on console markets as modern consoles are quite powerful.
It is logical step in my opinion, but time to create a mod that people will accept has been very high with this game and I'm not sure if that is going to become less even there are wonderful tools for PBR, there is so much more one needs to learn and build, how many people will have then time for such?
So I guess team sees also this and they have been building tools to reduce time needed to create, but I think that subject is something we see focus on future updates in form of helpful tools and especially information.
300 hours can be more than 1 year building a mod for single person and 300 might not be even a lot, especially if you are just starting out, I do miss times when 10K polygons were though as high end, modding was not a freaking job back then.
e. bah, discussion with modern persons is pointless as they just see one step right in front of their face and no deeper thinking of whole chain of effects never occurs.
Sounds like youre reading to much into it when it comes to effort from modders. First off you probably wont need to use every texture type if you dpnt eant to. Another thing is im running a gfx mod but ive made hundreds of 3d modeled cars and one of the big things that has kept me from finishing my 3d models for beamng is that i know itll look dull ingame due to lack of supporting components.
Not enough clearcoat?
It's the big picture I lost in my thoughts once again, not so much of this exact upcoming change.
I have played with Substance and when you get all the quirks down it is much faster to get materials done to incredible level of visual quality compared to old methods, these wonderful tools do help a lot, but then bar rises again and savings vanish when people are excepting more and more.
Please add the firetruck, then people ask make it squirt water, make water put out flames, I really love that Diamondback's reply, because that is how people really are.
You have put your bar high enough for not putting your cars to game, but I'm quite certain there is still quite lot of people who would not see your cars as dull, but interesting, likewise I'm not releasing anything because I don't have enough time to spend on building skills and content to expectations of others and perhaps we both are just putting our bars too high or maybe standards have just gone beyond sensible levels already.
It is like dreaming of that perfect apple and not having any apples because dreaming too high, not sure if in your culture there is similar old wisdom, but hopefully you understand what I'm trying to say here.
Accidently mixed matte additive in it
I guess its up to each individual, back in the day i used to make cars for Racer(.nl) which was more or less meshslapping compared to BeamNG but that did make me put more cars ingame.
I have been dreaming of a Beamng specific vehicle editor thing with easy to use UI for importing a mesh, choosing a template jbeam, adjusting the jbeam of said template from body work down to suspension arm mounting points, setting up drivetrain and go. One can dream atleast
As someone who used to work in apple orchards, I can 100% agree on the apple side of things...
In all seriousness though, having high standards is a blessing and a curse. It's good to always push yourself and raise your bar, but only as long as your expectations are sensible and grounded. If a skilled modeler might be disappointed and feels like his creation is not living up to it's potential, it's likely because:
(1) he envisioned it in his head different, and the "dream meats reality" can sometimes be hard. Spending a lot of time on something can also impact this: a 10 minutes sketch is not comparable to hours and hours of hard work.
(2) He expects too much of himself, i.e setting the bar to unrealistic standards.
In both cases, the best way to counteract this is reminding yourself that people would probably still find it amazing, especially since they don't have the same picture in their heads, and since they don't notice all the tiny mistakes that might really bother you, instead they are just enjoying the (whatever it is you created). Just like thousands of people can go watch a movie and have a blast, yet the director would hate the movie and eventually release a "directors cut".
I also read what the devs said about the fire engine and have to say I respectfully disagree. Not that I think that a fire engine has to be made (actually- quite the contrary), but the reasoning behind it seems to be grounded in this very topic of "what will people think", which sometimes is more of "what I think that people think" and less of what they actually think...
At this day and era, it would not be unreasonable to except machine learning to be used to figure out optimal locations and parameters on nodes based on total mass, weight distribution CG, suspension type and parameters as well as mesh, after all it is pretty much just finding the optimal parameters trough series of tests.
Considering BeamNG.tech does have machine learning stuff already although in completely different purpose, such NB structure builder for me would be something team might at least give a thought.
Automation team made rather simple prefabrication based (I guess?) NB structure exporting, to have some premade starting points like vehicles currently on game and button find solution would be fairly possible, but do they have resources to spend something like that?
This is one machine learning project for BeamNG tech, today machine learning is happening so much and there are so good algorithms that I'm somewhat impatient to see more use of the stuff in everyday software.
But as team is clearly wanting to be in current high tech for self driving cars, they might come to have enough knowhow so that they can bring this kind of new methods for tools to make content, or then it is just my wishful thinking, but perhaps at least internally thought have crossed their minds of how to have community to produce more quality content and this machine learning area from my point of view could bring huge leaps to tools for vehicle creating.
Why is raising the quality bar a problem when most people make Russian meshslaps anyway? But really, most mods are low quality and everyone is fine with that, there are maybe 5 mods in the history of the game that exceeded the dev quality and most people don't even download them but prefer stuff like Gavril Vertex that is mid quality (at least model and texture wise) with a lot of quantity, or WBIMP which had a ton of content of varying quality, or small mods that add parts to vehicles that don't look amazing visually but are good enough... Only the dev quality standards will be raised and they are already over the top Jbeam-wise so maybe it's time for textures too? High quality mods have always been rare and this won't make them that more rare but just better, it will also improve mid-quality mods that will be able to stand out texture-wise. It's not like everyone will suddenly expect every single mod to have the best PBR textures possible made using paid software that some modders can't even afford, decent-ish PBR textures could probably be made quickly from edited non-PBR ones using GIMP or something and they would probably look good enough for most people. That's how I see it at least, I don't mind mods that don't have the best textures.
Tbh, the vertex wasn't even mid quality. Granted it was an automation model, but it was really rough around the edges and was plagued with bugs. Not to mention that the jbeam was pretty awful and it used to almost crash my game every time I loaded it in. I would put it at the level of Modsgaming.ru in terms of quality.
It's not THAT bad, it has an edited jbeam that fits the car properly, and a lot of work put on it. And anyways it's MORALLY better.
I think you are only talking about the first ever version of it, the mod has come a long way since then, but either way, don't you dare compare it to Modsgaming quality, there is one big difference, that the Jbeam fits the shape of the car. And remember that it was worked on for years at this point while Modsgaming meshslaps are made in 1 day. Modsgaming mods are below Automation quality and this was way above. Also it's getting remodelled and retextured now. I could give way more reasons why it's better but this is not the thread for this.
With how beamng will likely implement it we might, but these days most games do what is called 'packing'
AO/Roughness/Metalness are grayscale maps so what a lot of places do is have one map with all 3 (and sometimes 4) by putting them in the RGB(A) Channels and then pulling them back out. Most of these maps tend to be small size anyways, the biggest is always the normal map.
That's clever and very BeamNG-like, I hope this is what happens, but it could make texture creation a pain unless there's an automated tool for it
tools like substance and quixel mixer (which is free!) allow you to set up how your channels are exported so it would just be the people who are doing photoshop that would have problems, and you cant really do pbr in just photoshop correctly anyways so.