I believe (stand to be corrected) that the vids highlight the differing repercussions for the player over the Ai in similar situations..i.e. first one the player is forced off by the impact from the Ai coming up the inside, yet when he did the same thing to the Ai it stayed on the racing line as if nothing happened. In short...a nuclear bomb going off wouldn't force the Ai offline! I also fully realise that this has been happening since time began and will continue untill PCs are powerfull enough to give Ai the same physics as the player. Unless some physics/ai genius can code a solution that is!
That is simply untrue. Why on earth would it require full physics modeling? Just as an example, my "other" sim, 12 year old Papyrys NR2003, works just fine on the collisions using simplified AI physics. This behaviour can be even adjusted through configuration files and you can modify them pretty much like you want them, everything between like on tracks or like on slippery ice. It all comes to how track surfaces, grip and mass is modeled to the physics. It by no means require full calculations for every tire, suspension or even downforce like it is for players car. Also Game Stock Car Extreme, which uses practically rFactor1 engine, has quite nicely working collisions. I haven't played much else than the CART mod with it, though, but I have no problems with collisions there. With R3E even the small collisions are in most cases disastrous for the human player, like many have said.
If you are not mistaken there! Assetto Corsa does right this: it calculates each and every AI car'S behavior by right the same and full physics algorithm set that the player's car gets calculated by. Yet the problem complained about in this thread is >FAR< more evident in AC, than in R3E, as I earlier described. People in better knowledge than I am said that probably no other racing sim for PC until today operates with such a complex and powerful set of physics, than Ac does - not just for player's car, but for all cars there are, including all AI cars. It is claimed that simplified physics algorithms for AI cars simply do not get used. AI, driving physics and collision physics are three different things. They interact with each other to varying degrees, but they are not the same things. This is no "monolithic" chess engine.
I don't know how AC works...anyway, it's a resounding failure. Regarding AI behaviour and collision physics, R3E is way behind Rfactor 2 or GSCE.
i remember playing a lot of GT legend and Gtr2 .. and i mean a LOT ... Never had that issue with Ai. One thing i did noticed too ( and mentioned it before as well ) try adac 2014 and i dont seem to have this issue. Maybe it was just luck? I dont know but i did a championship a while back in Adac 14 and after a couple races i was surprise to find very few odd things with the AI. Who ever is having constant issue with the "spinoramafornoreason" try Adac14 and let me know if i am right or wrong. I would love your opinion and more testing.
Hohoho... Its the benchmark to beat in several categories, especially driving physics and interior and exterior car modelling, and obviously a very huge number of people disagree with such a coarse statement like yours. And currently it is way more successful, as a business and in user numbers, than R3E, rF2 or GSCE. Its online activity level dwarfens R3E. - "Resounding failures" look very differently. Feel free to not like it. But leave it to saying right that - instead of blowing your opinion out of proportion. AC performs extremely successfully since its launch, and much more successful than Kunos ever expected. Not to mention the top reviews it gets from testers and magazines. Some non-systematically collected numbers from a free survey (n=around 300) from some time ago. AC performs second best in online race choices, right behind iR. Not bad for a sim that time and again gets attcdked for its MP mode (which works solid and reliable, btw, just is not offering all feazures people would like to get). LINK And this is from my personal, subjective comparison ACvsR3E in the R3E sales thread in Kunos' own forums (yes, they tolerate that) LINK, #25 That is the essence of why I use both sims parallel, and prefer both to any other.
If so, my reply belongs into the category "misunderstanding" then, no offence meant. In fact, I even agree. AC's AI is lacking. But 1.2 update is in the starting block, and Stefano said in a stream he solved the issue of why the AI hesitates to overtake. He said the breakthrough came with switching off certain collision detection algorithms if the car in front is not braking, and that the AI now is more aggressive and seeks its chance to start a fight. We will see. The update also brings Zandvoort, laserscanned, and for free, plus several cars, also free. These are good times for sim racing.
But with that said i do agree with you, Ac is a success and i could not be happier for them. Ac and Dirt Rally are currently my to go too sim after R3e. We can only wish more and more success to all of them because in the end it means more fan of our favorite hobby
That is the general understanding. Whether 1.2 indeed fixes all AI, or is more an improvement, remains to be seen.
The problem is, we dont even know when will the patch 1.2 be released as AC has some slow production issues where they DLC and patch takes extremely long time to release to the public with months of complete drought of any content. I highly doubt we will see Zandavoort by the end of this month. AC's team is way too small for the work they are doing. They are ambitious but I think they really need to step back to consider their manpower in delivering their content. Especially now they have console build to worry for their small team where I rather see R3E head to console than AC because R3E is a much more developed game than AC. Earlier of this year, they promised that we will get 50 new cars for 2015 where so far, we only received 10 new cars via their first DLC while another 5 is coming soon. At the same time, they also promised that we will get 5 new tracks this year, where we only have one extremely lengthy track so far. Kunos can promise anything, it remains to be seen whether they can deliver it. I used to be enthusiastic with AC up to early this year due to many disappointments with their PR, their lack of communications other than teasers on their personal facebook and their ability to keep their promises. And also their toxic community that will defend the game despite there are a lot of glaring faults with the game then trash other games that does things differently to AC. I play mostly R3E, pCARS and Dirt Rally these days, because these 3 developers constantly deliver content to enrich the game along with decent communication with the fans. And those 3 games do compliment each other well, like pCARS does mostly fulfill my GT and road cars needs with their decent GT3, GT4 and LMP line up, R3E has a lot of good touring content in the form of DTM, Group A and WTCC and Dirt Rally does offroad amazingly with cars from the 60s to modern. All 3 games has a good amount of high quality tracks to race on too. R3E just need to release the GTR3 pack hopefully with Ginetta GT3 and Spa, then, R3E will be my GT3 racing game of my choice.
The ADAC Gt masters 2013 and 2014 allready have lots of GT3 cars, you dont have to wait for a GTR3 pack.
I prefer all of them to be in one class. Not jumping from ADAC 14 to ADAC 13 when I want to drive a Nissan GTR GT3. Or change to GTR3 for Alpina B6. Not to mention, there are a lot of good looking liveries that had just been updated for GTR3 that looked way better than ADAC liveries. GTR3 pack is very much meant to put all the GT3 in the same grid, instead of jumping from one category to another.
The same here, yesterday I had a great online battle ruined by those strange behaviour. In my opinion, to animate players to drive safely. The way is not tricking the physics, is implementing a safety rating; as AC is doing for the 1.2 patch.
What I mean is the fact that it seems to me that every sim I have ever played has the Ai weighing about 10 tonnes and the player cars like a feather....so if I tap the Ai in front I will always spin out, yet in every real race I see the car in front being hit spins in reality NOT the car behind. As I say in every game I've ever played that seems to be a constant (mainly Simbin stuff, so could be a factor). Never played NR2003 and don't play RF1, 2 or GSCE.
All I know is that I have not played a sim where I believe that the Ai car weighs the same as the one I'm driving (within rules of course).
I will add my request to the others that AI collision physics be looked into. Very minor contact and when the player rear-ends AI will often send one into an uncontrollable spin. That said, it does seem to vary somewhat between cars and tracks.