Discussion in 'General Discussion' started by LokiD, May 7, 2020.
Why are you going off-topic in your own thread ?
I think your right.
I should step away from here and not for just a short time. I mean permanently. My frustration at rf2 is getting the better of me, and its clearly not going to be the offline experience/direction that I'm seeking.
Yes there's other sims out there, but there is something about rf2 when it does it right its amazing. Its just its a mess of a title and for me that one single moment of whoa doesn't count anymore.
So maybe yes its time for me to properly step away so good luck devs in the future of rf2 hope you achieve your goals.
Again? How many times is that now? (rhetorical questions) I lost count years ago.
(stupidly clicked "show ignored content" - lesson learnt!)
I to find Rfactor to be one of the most frustraing programs to deal with that I currently own, but its maybe because i love it so much! Its partly to do with how damn complicated it is, I find if you wanna set up a really good offline race in RF2, you've gotta plan a day around it, finding a car and track that the A.I. work well on for you, and at your skill level, then there is the fact its different everytime you drive because of track conditions, time of day and what not. But the hours racing you spend all that time over, its just phenominal!
Offline i do tend to spend most of the day with one car and one track, if im gunna set up a long race, its that indepth i really need to get a lot of practice in, its amazing how the tire model reacts to wear and the differing grip on the track (as well as that feeling on the edge of grip) as the race goe's on. Not too long ago a set up a 24hr race at Le Mans with the GT3 cars and set the time cycle to do it all in 1 hour, with changing weather and rain at a certain point, it was just fantastic, the way the grip changed ever so slightly from lap to lap, feeling the grip go as the track got wet, waiting till the very last minute to pit for wets (becasue up to a point slicks on a very slighty wet track are about equal to wets on a dry track, until it gets wetter and then you want the wets), then after it started to dry as the sun came up feeling the grip gradually come back, musch more so on the appearing dry line, staying away from the still very wet patches, just brilliant!
On the subject of Rfactor 2 being frustrating:
Ive been running an online server recently called: POTATO Thats for quick and easy pickup racing with only official content, and a lovely fella who's been following it all offered to donate me a load of parts to build a proper server PC. Anyway when I spoke to him on the phone we had a long chat about Rfactor and we both agreed that owning Rfactor is a bit like the stories you hear from Old Alfa Romeo drivers or certain enthusiasts cars. Most of the the time you'll have your head under the bonnet, sometimes it won't even start, but the few times a year it works perfectly, holly hell is it one joyous occasion and so worth it!
Another sim which I still love for its driving feeling that really feels like a car to me is the legendary Richard Burns Rally, I don't know if its all down to the tyre model with that one, or just good physics in general, but back in the day, that an RF1 were my favorite driving feeling, still got em both installed!
Hope these new updates coming sort out alot of stuff and get you back in the game mate!
There's a lot more shades than black and white.
You can't run the tyre model rF2 does in realtime on current PCs. People can argue about the output (usually without bothering to take any measurements or compare against some reference, but still) and that's their right, same as they can do with any model, but to do what it does and end up doing it 'live' they have to precalculate a whole host of scenarios and then apply the most correct one (or perhaps an interpolation of the most correct ones - I don't know the details) to suit the current situation.
To label that physical or semi-physical is sort of missing the point; if you had a fast enough PC to do exactly what it does (overall) in realtime, you might be tempted to call it physical despite it having exactly the same outcome just because it's all being done in realtime - but labelling anything calculated 'physical' is really a bit of a misnomer. All these solutions are just aiming for the best results, none of them are or will ever be perfect.
@LokiD I posted at work and was probably more abrupt than I could have been. But, we've been on this merry-go-round before. I don't know how many times we have to talk about the same things before you don't come back and make more insinuations that completely ignore the previous discussions. You surely know I'm not a fanboy, or you really are trolling. I call a spade a spade, whether it's issues with rF2 or your posts. But it would probably do us all some good to step away occasionally.
What do you mean by canned physics effects?
They improved chassis model by adding flex calculations to it.
Same as rf2 did.
Now it is part of physics engine.
He probably refers to effects like rumble strips, curbs hitting, and tire lock vibrations.
This is one of the most elitist comments, to find about something, probably don't even understood correctly (and most of us don't), because if you would understand it, then you would've not assumed, that this is something "canned". Pretty amusing.
A proper tyre and chassis model is developed after many steps. The older ISImotor versions didn't have some of the features, rF2 has. So it's also all just added on top of it?^^
Btw. you guys are talking about the "5 point tyre model", but it's actually the contact patch, that has been changed, to give the actual tyre model more information/numbers to work with. (To finally eliminate the horrible curbs of death and sharp edges problem)
Perhaps I didn't clarify enough. An effect such as 'chassis flex' would not need to be 'added' into a physics model such as BeamNG's, because the chassis flex is an emergent property due to the entire car being created from virtual beams and nodes which have physical properties applied to them. On account of said construction, the soft-body flex is already present, without needing to be 'added' as an additional update. It's the same as the emergent properties you get from the physically modeled tyres - for better or worse. Some of those emergent properties create unpredictable or undesirable results until they're balanced with more variables, or more accurate data. Other emergent properties create a situation such as rF 2's, where curb behavior isn't really a problem because the fidelity of the tyre model insofar as contact patch is concerned is far greater than that of AC. These are the pros and cons to empirical vs physical models that Niels was referring to in his video.
The way that computing is going, at least insofar as my colleagues that own software development firms inform me (or at least insofar as my programmers inform me from also owning one) is toward the direction of emergent neural networks. I spent a while reading the brief behind BeamNG and why they feel their approach is the best way moving forward - physically modeling the entire structure of the car. And I completely agree. In the short term it might mean their game is a scattered potato with no direction, but in the long term it's the only way to get the truly emergent results that you would want from true simulation. For instance, there is no way that a colleague of mine's company could've coded their poweramp model by hand, in spite of having several PhDs behind it. They used a neural network to create an immensely complicated, truly dynamic model. It was an emergent outcome from feeding tons of real-world data into an AI learning system. This is as opposed to somebody creating a simpler static power amp model, and then 'adding' dynamic behavior as a few lines of code on top to approximate outcomes empirically.
Hopefully that helps clarify the original statement.
Sounds almost like things, the Kemper Profiler guys may work on.^^
I get your statement. It was still pretty elitist. The similar results are given, but it's "amusing" because a feature was added afterwards. The ISI engine also was not intended to utilize the tyre model, it's now running on with in rF2. The workwise of ACC doesn't even matter for the end-user, the feature palette is similar and in some occasions even more functional (pressure and temp management as an example).
It doesn't. A physics engine can be updated by assuming less parts as rigid, taking deformation of these parts into account which were first neglected. This has nothing to do with canned effects.
EDIT: its as a response to Ermz. CrimsonEminence beat me to it xD
Im reading this thread and getting the jist of but I must admit most of its going right over my soggy noodle, but one things like soft body physics and dynamic tire flex is mentioned, it does make me wonder why LFS never gets mentioned, they we're doing this impressively before either the the two platforms mentioned and with a very small team, I've often thought the top sim that could ever be produced would be an amalgamation of the best bits of all of them. Utopian idea I know.
LFS was FAR ahead of its time, but its development is awfully slow (no bashing here, i was trying LFS around 2003/2004 and was already blown away by its driving feel in this old state) so most of the simmers probably don't even really know it up to this day to talk about it that much.
Server down for a bit whilst i sort a few bits out
People have been talking about LFS a lot on the YouTube comments of late, though I must admit to my shame that I've never played it nor heard very much about it (other than when it was brand new). Is there a video or something to that effect which showcases these revolutionary things they were supposedly doing back then? I'm genuinely curious.
re: other guys. The point I'm making is that if you have a relatively complete physical model then you don't need to go back and add so many of these aspects which, in retrospect, you realize were necessary to driving all along. AC has fairly mediocre on-curb behavior to this day because of its tyre model. It also has a similarly mediocre FFB engine on account of many 'missing factors'. In ACC they could no longer get away with a model that simple, so they went back and added in the variables required to handle more extraneous circumstances. Yet for how many years have people been driving with a fundamentally flawed model prior to that?
Similarly the jump from rF 1 to rF 2 incorporated aspects of the 'RealFeel' plugin to improve FFB, along with rewriting the tyre model completely. I'm not trying to make some 'elitist' statement about rF 2 necessarily being better regarding retrospective changes. Lord knows it still needs many more of them in order to be a complete game. However, I think the conceptual approach to its tyre model is going to lead to outcomes where the driving sensation remains more natural even when a tyre is taken beyond regular measurable limits ie. almost all racing applications, and we can see that demonstrated quite clearly by the fact that driving 'on the limit' still feels far more natural and communicative in rF 2 than it does in ACC, in spite of the updates done to their empirical model over the last 2 years.
Admittedly, the only game currently going as far as physically modeling the entire vehicle (to my knowledge) is BeamNG, and that has a long way to go to come to fruition, but seeing the leaps and bounds it's making, and given that at present it's probably the most accurate drifting simulator on the market (not to mention off-roading) speaks a lot to the validity of the approach.
When is a physics model complete? Just because more details are added doesn't make it flawed prior to this.
What is modelling an entire vehicle? Do you think they don't make a simple assumption?
Yeah exactly, good point...
Q: "when is science complete ?"
Just go to the LFS website and download the demo, believe it or not they're still updating it to this day, oh and there is a massive drifting scene for it!
@Ermz You are right, we have driven several years in AC with tire problems, just like we have in rF2. Both have problems, I am aware that AC has a simpler tire model, although this has its advantages and disadvantages.
Wow! That's really impressive for such an old game.
Separate names with a comma.