[Mike B]

Optimization can only go so far; eventually, if you want the best experience from a game, you need better hardware. Only you can determine when you've reached that point, but the software developers are always trying to get the most out of the higher-end systems.

In my case, I'm still using an old system - closing in on 10 years, really. But it was new to me about 5 years ago, and was high-end for its time (got the motherboard and Core2 Extreme processor from a friend who was upgrading, and built a new computer around it for about $300 reusing some parts from my older one. The video card I got from him was pretty basic, and didn't run ORTS well; eventually, I found $100 for a newer nVidia card which made a huge difference - ORTS is very playable, usually at the refresh rate (75 Hz CRT monitor) or better. So that was a case where the hardware upgrade was not HORRIBLY expensive and made a real difference.

Key point: my main computer is a "tower" type desktop. That makes it fairly simple to do incremental upgrades without having to budget for a total replacement periodically (as is necessary with laptops). Laptops might be attractive because you can carry them around, and they come complete and ready to go (no connections, no fiddling) - but in the long run you pay for that. My previous computer saw 2 complete motherboard replacements in the same mid-tower case, and in this one (got a new case for cooling reasons) I've upgraded other parts occasionally, too. The biggest improvement after the graphics card replacement, for instance, was getting a cheap (I use the term advisedly) SSD for a boot and MSTS/ORTS drive (about $100 for that, too). And despite what it sounds like, $100 is something I have to budget and save for - it's not chump change on a mostly-fixed retirement income.

Bottom line: all software needs optimization, but if the software you're using is actively maintained and updated then at some point you have to throw hardware at it too, or else accept a lesser experience.

[Lindsayts]

Kazareh, on 10 December 2016 - 05:10 PM, said:

Vince, do you care to pay for a new computer for me..? If not, don't bring that up. Anyone can 'throw money at a problem till it goes away', however not all of us have the money to DO that.

Also Optimization isn't dumbing the system down, either. Proper games with good optimization run better than others and still have great things in the Graphics department. They can also be adjusted with settings as well for example, adjusting strength of AntiAliasing, Texture Quality, etc. There really isn't much a reason ORTS Can't do the same, and just get the program to run smoother and better optimized on the graphics front. I mean when a game with as old as a graphics engine as World of Tanks can produce an image as such? I don't think ORTS has much arguing room here. That screencap when taken was running at 60FPS due to frame-limiting, where as if unlimited would have ran at 142FPS. Yet sometimes, even when my Radeon HD 6950 was working? ORTS struggles to break 40FPS.

Edited for clarification of one point and additional screencap

The limitation in OR's graphics is because it can currently only use a single thread for graphics, this I believe is a limitation with the tool kit used and updating to a better tool kit is going to be VERY difficult and time consuming and this will very likely be well beyond the current resources availible. I really high spec system is not required though any 4 core CPU mother board with say 8gig of ram and something like Radeon 5870 does most routes well. In Australia such systems are starting to be junked so a second hand system should not be to expensive. Note: I do still run OR on such a system, it now being over 5 years old.

Lindsay

[Lindsayts]

conductorchris, on 10 December 2016 - 05:34 PM, said:

I agree that an activity editor is important. Fewer and fewer people can even run MSTS.

Another course would be to continue work on the timetable format. Introducing switching into the timetable format and you'd no longer have so much need for an activity editor. If switching could be randomized in the manner of skyline computing's activity generator, well that would be even more cool.

Another thing Open Rails does not have which the other sims do is a marketing department. This influences how many people know of it and thus the amount of potential people who might work on it and improve it. If only I had more time I'd do some work in this area myself, do alas I don't.

Christopher

The lack of a marketing department is a good point, this though is quite common in open source, so the program/s MUST be able to stand up out there on its own and be seen to be good, Openrails itself is excellent, I believe its the lack of some tools that is letting the side down. We are well on the way to a reasonable route editor as well as a consist editor, I believe there is a path editor. An acitivity editor that can modify and existing actiivity as well as making new ones would be real nice to have.

Lindsay

[Goku]

Kazareh, on 10 December 2016 - 05:10 PM, said:

Vince, do you care to pay for a new computer for me..? If not, don't bring that up. Anyone can 'throw money at a problem till it goes away', however not all of us have the money to DO that.

But better code might be a lot more expensive than better computer. Do you really have no 40$ for decent gpu for open rails?

Kazareh, on 10 December 2016 - 05:10 PM, said:

Also Optimization isn't dumbing the system down, either. Proper games with good optimization run better than others and still have great things in the Graphics department. They can also be adjusted with settings as well for example, adjusting strength of AntiAliasing, Texture Quality, etc. There really isn't much a reason ORTS Can't do the same

No. There is a one really BIG reason. Fast game engine requires not only code optimization but most important: content optimization. Content has to be made with specific requirements. Our content for MSTS was not made by proffesional game designers, but by ordinary users. They did not know how to prepare good quality game content and MSTS tools did not provide tools that modern game engines use for it's content. For example all current game engines use deferred rendering that provides amazing performance, but content must be prepared for that kind of rendering.

[ATW]

I have a total of 4GB of GPU RAM, 24GB of CPU RAM on a laptop. It's a great beast laptop but ORTS needs new options especially for the future to detail of new routes an trains an not always the need to buy new high end computers even though my laptop stutters FPS on some occasions due to spawning, distance levels as well as if I increase the detail in my GPU settings to high quality.

I recently experimented heavy tests with ORTS Experimental tab for "Extend Object Viewing Distance To Horizon" which I like since it maximizes the viewing distance of objects set by the viewing distance radius in video tab which doesn't pay attention to a objects LOD very well. Unlike MSTS.... ORTS doesn't have a separate TAB Slider for Train Objects vs Route Objects as I am looking to maximize scenery object viewing distance on some route runs without having to lose a big amount of FPS because my long train or AI maximizes it's viewing distance too with max LOD's regardless of the shapes max view distance. There needs to also be the Texture quality control for both object type sliders.

Before anyone tells me of the Experimental slider TAB "Level of Detail Bias" I have used that an seen some behavior changes in LOD's for my train an have mine set to the half negative ranges to eliminate unnecessary distance details of LOD's which increases FPS. Can there be a option slider for "Level of Texture Quality Bias" being something similar to DXT compression but in game?

So what I am suggesting in performance optimizing is something similar to MSTS which is separate sliders for different objects where current ORTS tabs I speak of can be used as templates to start them out 1 by one testing:

World
"Level of World Detail Bias"
"Level of World Texture Quality Bias"
"Extend World Object Viewing Distance To Horizon"
"Max World Object Viewing Distance"



Scenery/Terrain
"Level of Terrain Detail Bias"
"Level of Terrain Texture Quality Bias"
"Extend Terrain Texture Viewing Distance To Horizon"



Train
"Level of Train Detail Bias"
"Level of Train Texture Quality Bias"
"Extend Train Object Viewing Distance To Horizon"
"Max Train Viewing Distance

[disc]

Goku, on 10 December 2016 - 10:31 PM, said:

No. There is a one really BIG reason. Fast game engine requires not only code optimization but most important: content optimization.

That's true, most of the MSTS contents cause too many draw calls, as the objects are not attached, and using many texture files instead of one big texture file which is faster.

[Goku]

Worst thing is that in modern 3d models, transparent parts has to be separated from other parts, because deferred rendering is unable to render transparent parts, but in most MSTS addons whole shapes are configured as blend shaders. It makes using deferred rendering pointless, because you have to render these all blend shapes using classic forward rendering. :(
Today deferred rendering is only reasonable choice, because you are rendering geometry only once and all other effect are pure gpu. In classic rendering each effect doubles cpu usage.

Example how deferred rendering looks like: http://www.adriancou...graphics-study/
You can see how transparend parts are rendered at the end of rendering.

[Lindsayts]

On high detailed graphics...............

Its folly to judge ANY game, particularly a train simulator JUST on graphic detail alone, such a train sim WILL NOT have any longevity at all.

Most routes and models for MSTS are done by normal people with a high interest in railways, the "game engine" MUST be able to handle work done by such people with little to no knowledge on how graphics are drawn on a PC.

As I have already stated any updating of this "game engine" will require a new basic tool kit with with the required facilities, this will require a major rewrite of the graphics sub system, this will very likley be beyond the current development team numbers. Even then it STILL has cater for my second point.

There is little point in comparing a route with models and everything else done by true train enthusiasts to game graphics developed in house to suit a particular "game engine" such graphics will almost certainly be better but such graphics will always be VERY limited by the necessity for commercial success, a model or route done by the true enthusiast will not have that limitation.

I will repeat an important point rewriting Openrails for a new graphic subsystem will be major effort that will very likely take a lot of time and effort and require specilaised programming knowledge, this will not likely to be done any time soon. The point of this thread is to try and make Openrails overall easier to develop new material in a relatively easy fashion, so more users AND therefore developers of all types will be attracted to Openrails.

Lindsay

[disc]

Lindsayts, on 11 December 2016 - 11:10 AM, said:

the "game engine" MUST be able to handle work done by such people with little to no knowledge on how graphics are drawn on a PC.

It's handled. Just the unoptimized content is slow, just like in every game engines, except those that are using directx 12 or vulcan, or mantle apis.

[cjakeman]

Lindsayts, on 11 December 2016 - 11:10 AM, said:

The point of this thread is to try and make Openrails overall easier to develop new material in a relatively easy fashion, so more users AND therefore developers of all types will be attracted to Openrails.

Can we focus on Lindsay's intention as the Original Poster - How can we more easily develop new content?

Rob's timetables have been a terrific step forward.
Stefan Paitoni started an Activity Editor - is anyone familiar with that project?
I wonder whether Goku has had time to plan for procedural generation of fences, hedges, powerlines etc. in his Route Editor.