Wheres the new snapshot @jlv

[ddmx]

...would be inaccurate (sure, there isn't a standalone track editor)...

[Atom6246]

In a lot of ways I agree that content creators are the backbone of MX Simulator. However, to infer that JLV hasn't provided enough tools for creators wouldn't be inaccurate (sure, there isn't a standalone track editor).

The platform that MX Simulator is built on is the single greatest 'tool' the game provides. It's about as open source as any video game can get. If you don't believe me then try making mods for a AAA title where file formats are wonky and you need to decrypt everything using 3rd party tools. Aside from what's hard coded, nearly every visual aspect of the game is incredibly easy to alter. The hardest part of getting a new model into the game is learning how to use Blender.

As far as tools, I'd also suggest that some of the 'tools' he's provided over the years have been normal maps, environment sounds, animation, and many others. None of those existed in 2007 and a decent bit of the game you see today is driven by not only JLV's vision but also the Suggestion Box on these forums.

You're right. The tools are there but what I'm trying to convey is what we are using could be more efficient and better overall. Guess that's a can of worms I shouldn't have opened because there will always be bigger and better engines and tools to have.

Environment sounds, animations, etc are nice to have sure, but IMO I'd rather see more optimization and gfx updates. I don't understand how those aren't the highest priority. Thats the main reason people get turned away from this game, it's looks exactly like it did when it released (without the community's content minus a scoring tower). I do understand JLV had to take care of things behind closed doors but now those are seamlessly fixed, now the focus is back on animations that are in no way beneficial to the games performance. If anything it will be the opposite. I don't think people realize how hard it is to run this game. I'll give you an example, I just spent $700 on a new CPU/mobo to be able to get 125fps in outdoors with track objects and a stick figure replacement pack. A i7 10700k, and a 2080 is finally enough to get a stable 125. That's not good.

Once again, cool to have, but theirs bigger fish in the sea that should be looked at first.

As far as erode goes, I'm pretty sure we are limited from what we have rn. Not much more to learn until we can get higher scale resolution tracks if I'm not mistaking.

[ddmx]

I think we'd all like to see varying degrees of graphical improvements (dynamic shadows, please ).

The game as it is today is a perfect representation of simulation first, prettiness second. I think that's always been the priority stack.

[Bradclay306]

+! on wanting to see graphical improvements to the current engine.

Would there be anyway to implement 4k terrains being usable? Building an outdoor track on a 2k terrain's with more 'realistic' sized ruts is virtually impossible without creating a jagged mess. go back and ride some of the amazing gp tracks by haggqvist back 7/8 years ago. the ruts are just a mess. i personally think this would be a massive benefit to the game

[Luca Tirone #48]

I agree with Atom. To me this year I would have really liked to run the whole national rf championship, but unfortunately I can't do it because my pc in a track with more than 20 people turns me at even 30 fps and therefore it becomes unplayable and my desire passes to run a championship like this. My pc will not be very fast, however, it still manages to make me run recent titles in even high resolution, but on mx simulator even lowering the graphics to a minimum, in a race it becomes unplayable.
In my opinion it would be better first to improve these problems, because many people stop competing for this reason too.

[Jones221]

Everything posted on the previous page was not in-game. After reading his interview, he has a high poly cad from the manufacturer used to bake. Please find in game screenshots showing details.

@jeremy : still curious to see LOD popping examples as well.

@alex : PC’s have upgraded but the game never changed the way it reads/renders.

@atom : kellen can’t load without crashing due to models/skins used, this is obvious.

This is a 1 man operation still getting updates 13 years after the release. Be happy it even gets worked on..

[TeamHavocRacing]

Ya, but when will we get an update?

[Hi Im Skyqe]

@atom : kellen can’t load without crashing due to models/skins used, this is obvious.

I would attribute Kellen's crashes to SCRAM files, everyone uses them for pretty well everything now.. We had tons of 4k packs being used with 'bloated' models years prior to the SCRAM file without issues. The reading of the file takes long and with the short timeout it is bound to take multiple tries to load in..

[Atom6246]

@atom : kellen can’t load without crashing due to models/skins used, this is obvious.

I think the obvious reason of this (which you can even hear Kellen say in his streams) is that this is attributed to scram files. BUT, from my knowledge, isn't this game capped at using 2gb of VRAM? Which could potentially be an issue as well? If it used more VRAM it would be able to load more without running out of virtual memory, and or timing out? Right?

That's another thing, content creators like Youtubers, Photographers aka "screenshotters", etc can't even make the game look it's very best because of running out of virtual memory. Discard frames for a moment and think about that. They can't even load into the game.

[motokid499]

+! on wanting to see graphical improvements to the current engine.

Would there be anyway to implement 4k terrains being usable? Building an outdoor track on a 2k terrain's with more 'realistic' sized ruts is virtually impossible without creating a jagged mess. go back and ride some of the amazing gp tracks by haggqvist back 7/8 years ago. the ruts are just a mess. i personally think this would be a massive benefit to the game

Personally I love Haggqvist ruts. I'd rather fucked up realistic sized ruts than trying to design ruts in a way that optimally works for the graphics engine. Ideally, 4k terrain would be the answer, but until then am I the only one that genuinely isnt bothered by the chatter and shit? Just go slower. It used to be a huge pissoff for me and the main motivation behind my "erode sucks" post, but i've been able to modify my advanced stability in a way that works better for the chatter and now I just kinda want realistically sized ruts I really dont care if the game dont support it Ill gladly slow my pace down to make the rut work. The way I see it now, the chatter is a feature that should be used to stop people from going 5th wicked in every rut

[jlv]

Absolutely.



I'd probably estimate theirs is 10x the poly count, considering my structure uses 4 sided bars, and theirs is using 6 sided, plus plenty of extra geometry for the bars to connect. But that's how it should be, we don't have as powerful of as engine as of right now. I'm not arguing that our models are too high poly, I'm arguing that other games can get away with way less model/texture detail due to better lighting/postprocessing/etc.

Save 4 triangles by using 4 sides instead of 6 sides on some tubing, waste 10,000 triangles on a bike on the far end of the track by not having an appropriate LOD. Not a good trade. No game engine can afford to blow tens of thousands of triangles on stuff that is 10 pixels on the screen. I admit there's lots of stuff I can optimize more, but the mesh rendering code? Here it is, tell me how to optimize it:

Code: Select all

                glCallList(model->list);

That's how the game renders meshes. It is literally one line. The rest is all done by your GPU/driver. Maybe I'm doing something retarded making the display list? Here's the code. It basically says, "Here are some triangle indices, vertices, texcoords, normals, and possibly tangent vectors, bone weights and bone indices. Compile them into a display list." Very straightforward, by the book stuff. (If you don't know OpenGL, this is the initialization code, run once when the track loads.)

Code: Select all

        ...
        glNewList(model->list, GL_COMPILE);
        draw_model_array(model);
        glEndList();
        ...

static void
draw_model_array(struct model *model)
{
        glVertexPointer(3, GL_FLOAT, 0, model->verts);
        glTexCoordPointer(2, GL_FLOAT, 0, model->texcoords);
        glNormalPointer(GL_FLOAT, 0, model->norms);
        glEnableClientState(GL_VERTEX_ARRAY);
        glEnableClientState(GL_TEXTURE_COORD_ARRAY);
        glEnableClientState(GL_NORMAL_ARRAY);

        if (g_options.shaders) {
                mxglClientActiveTextureARB(GL_TEXTURE1_ARB);
                glTexCoordPointer(3, GL_FLOAT, 0, model->tangents);
                glEnableClientState(GL_TEXTURE_COORD_ARRAY);

                if (model->weights != NULL) {
                        mxglClientActiveTextureARB(GL_TEXTURE2_ARB);
                        glTexCoordPointer(4, GL_FLOAT, 0, model->weights);
                        glEnableClientState(GL_TEXTURE_COORD_ARRAY);

                        mxglClientActiveTextureARB(GL_TEXTURE3_ARB);
                        glTexCoordPointer(4, GL_INT, 0, model->bones);
                        glEnableClientState(GL_TEXTURE_COORD_ARRAY);
                }
        }

        glDrawElements(GL_TRIANGLES, model->ntris * 3, GL_UNSIGNED_INT, model->tris);

        if (g_options.shaders) {
                if (model->weights != NULL) {
                        mxglClientActiveTextureARB(GL_TEXTURE3_ARB);
                        glDisableClientState(GL_TEXTURE_COORD_ARRAY);

                        mxglClientActiveTextureARB(GL_TEXTURE2_ARB);
                        glDisableClientState(GL_TEXTURE_COORD_ARRAY);
                }

                mxglClientActiveTextureARB(GL_TEXTURE1_ARB);
                glDisableClientState(GL_TEXTURE_COORD_ARRAY);

                mxglClientActiveTextureARB(GL_TEXTURE0_ARB);
        }

        glDisableClientState(GL_VERTEX_ARRAY);
        glDisableClientState(GL_TEXTURE_COORD_ARRAY);
        glDisableClientState(GL_NORMAL_ARRAY);
}

You could argue I should use vertex buffer objects instead of display lists, but guess which one tends to have more performance problems.

Now, you might say, "Doom Eternal draws 90 million triangles! How do they do it? It's because Vulkan is 100 times faster than OpenGL isn't it?"

I hate to break it to you but...

• This is coupled with a new LOD system that is designed to effectively switch between detail levels on models without interrupting the presentation. This works by determining how much of screen-space an object occupies on-screen at a per pixel level, then using the factor generated to decide uniformly when to switch LODs.

So the way they draw 90 million triangles is ... avoiding drawing 90 million triangles. This is basically how optimization always works. The fastest way to do something is figure out a way to skip it entirely.

This isn't to say that there aren't things that need optimization in the game. But triangle throughput in the mesh rendering? There's nothing I can do about that. Other games do better by not wasting their triangles on stuff half a mile away. Pro modelers do stuff like this for a reason.

[jlv]

Environment sounds, animations, etc are nice to have sure, but IMO I'd rather see more optimization and gfx updates. I don't understand how those aren't the highest priority. Thats the main reason people get turned away from this game, it's looks exactly like it did when it released (without the community's content minus a scoring tower). I do understand JLV had to take care of things behind closed doors but now those are seamlessly fixed, now the focus is back on animations that are in no way beneficial to the games performance.

Just a little explanation on that. At the time I had to add some code that I didn't want certain people to notice. Adding a JS interpreter was a great way to hide what I was doing in a huge amount of new code. But now that scripting stuff is there I might as well use it. The moonwalking JLV starter is such an eyesore. I cringe every time I see it.

[Jeremy150]

I never claimed that optimizations needed to be made to drawing models? I was just pointing out that these models might not be as bloated as you think they are.

To your point of LODs, again, I can't disagree that every game uses them and are 100% necessary. However as said, we have noticed performance issues from using LODs. First example that comes to mind is Washougal. Originally the trees all had LODs but we decided to take them out because they were causing more lag than without them. Maybe it was a model problem? The only other thing I can think of is, maybe if too many LODs are popping it creates a lag spike?

[jlv]

I never claimed that optimizations needed to be made to drawing models? I was just pointing out that these models might not be as bloated as you think they are.

To your point of LODs, again, I can't disagree that every game uses them and are 100% necessary. However as said, we have noticed performance issues from using LODs. First example that comes to mind is Washougal. Originally the trees all had LODs but we decided to take them out because they were causing more lag than without them. Maybe it was a model problem? The only other thing I can think of is, maybe if too many LODs are popping it creates a lag spike?

When I said "bloated models" I was answering this:

My bad, I should have been more specific, when I say "better graphics" I don't mean better textures, higher resolution and more geometry in the models. Honestly for a game that came out in 2007 you can make this game look pretty good. I just thought since he was working on flagger character models, this might lead to an update where we can use more 3D objects with less aggressive LOD's and leave the 2D billboards for very distant details and smaller grasses. I know what it would take to overhaul the game and I don't expect that out of JLV.

Basically he was asking for more speed with less aggressive LODs and I was telling him that's not going to happen.

The highest level of detail doesn't really matter as far as frames/sec goes. 10,000 vs 11,000 doesn't matter much. The problem is the lowest level of detail. There it's 10,000 vs 500. That's a killer. It leaves the game engine with no quick way to draw the model.

Do you still have the Washougal trees? Did it lag going up in detail or down?

[Jeremy150]

(triangle count)
They are the trees Wilson released here

From memory it lagged as going up to the highest LOD.