Messages

Thanks Zak, and on to the analysis then! First track sector definitions:



USC-style graphs are attached for following the analysis, as well as the spreadsheet with complete section data. The times were, for the sake of convenience, normalized to Carrera bonus percentage.

Charting initially the top 7 laps, the first striking feature is, obviously, Ayrton's lap. On the first two sectors we could say he and CTG, given the different car characteristics, were pretty much even matched. But then... we are treated to a most amazing demonstration of how one claims any point in space and calls it "road". Really lovely to watch! Anyway, the huge gap imposed by the champion means we can't see much more on this graph, so below it I placed another one covering only the really busy race from 2nd to 7th. Let's go pipsqueak-by-pipsqueak:

• CTG built the advantage needed to win the second place with a flawless initial third of lap (sectors 1 and 2). He lost a few tenths at the 180 degree turn on the third sector (trying to take advantage of a wider trajectory by cutting through the grass on the hill just before it didn't work well), but on the following sector any threats from Mark and co. were curbed with a nice and long flight through the sector 4 cut, landing directly on the right turn just before the slalom.
• Mark had a exceptionally good third sector, thanks to a very aggressive entry on the 180 degree turn. That reduced his gap to CTG to only 0.15s, but then his gains were lost in the next sector due to a slightly less efficient cut.
• As for myself, a slightly too high jump and a gear change miscalculation on the first sector loop cost me about half a second, and thus the third place and possibly the second place in overall classification as well...   At least my race was more solid from the third sector onwards. I succeeded to limit the damages by taking a wider cut on sector 4, allowing me to take the following turn at a pretty high speed (and thus having said sector done only 0.15s slower than CTG, who used a much more direct cut). Without that, I would have finished 6th at best...
• Bonzai Joe's lap progression was similar to Mark's, but a few precious tenths lost in sectors 2 and 4 kept him away from the podium. In sector 4, he performed a bug flight similar to CTG's but shorter, which forced him to brake too hard before the following right turn. Remakably, BJ set the fastest final sector, overtaking CTG in the process.
• CTG and Zak had strong races as well. The pair (alongside with BJ) stayed very close to each other during most of the race, exchanging positions often. They only drifted a bit more from the podiumers at sector 4 (which cut was, as it seems, a key part of the race for everyone), and only settled into their final positions at the final sector.

Last, but not least, there is interesting analysis to be done on the following section of the scoreboard, positions 8 to 12. There are two graphs as well for them, one including Pavekiller and the other not, for clearer visualization. SuperBrian dominated this group with solid auto-gears racing. AbuRaf followed him closely, to the extent the low top speed of LM002 allowed him (the 8th place battle was mainly decided on the "fast" sectors 3 and 4). Krys and Dstien were close to the fight above them during the first half of the race. Dstien lap in particular was quite interesting to see - he beat me on the first sector, and then almost surprised Krys by recovering more than two thirds of his gap on the final sector... Finally, Pavekiller only missed the sliding jumps in the final sectors in order to be involved on the 10th place fight, or maybe even above.

All in all, Z88 was an extremely interesting race, and we almost didn't notice it while racing... on a closing note, it is very positive to see close racing by several drivers just below the podium battles. I feel having solid competition at the midfield will provide positive feedback to the number of newcomers by providing a stepping stone and intermediate goals for up-and-coming pipsqueaks.

While I was watching Z88 replays for performing the analysis, I identified a few issues with tgm's replay, a BB 1.0 lap. First of all, it was not finished, but rather stopped in the middle of the final cork l/r. The additional ~2.8s would not affect the final scoreboard order, and it probably wasn't done in bad faith - the probable scenario was tgm realized having sent an invalid lap, sent a proper replay one hour later (as shown by the stats) but, either thinking the system would deal automatically with such problems or just driven by general newbie shyness, did not warn Zak of what had happened. Anyway, such issues may well happen again with other newbies on the future, and thus the "20 0x00" check appears to be more important after this. There is a second, more subtle problem: as mentioned, his lap was done with BB 1.0 . Looking at the scoreboard or with RPLInfo, the returned length of the replay is 1:55.30; however the actual replay tape stops at 1:56.40... the 0.1s difference appears to come from the different .RPL headers from versions BB1.0 and BB1.1, which mean laps having the same length will result in files two bytes longer if they are done in BB1.1 . That is another issue to be checked if support for versions other than BB1.1 is to be kept.

uh, whats a good number of polygons to have and is better to use quads instead of triangles, will polygons with more sides get better preformance?

It seems that the number of polygons is the limiting factor on performance issues, the number of sides does not matter much. car0 can have pretty much as many polygons as you wish, since it isn't used for racing anyway. Most car0 shapes from original cars have around 150-180. car1 shapes usually have ~70 polygons (if I remember correctly, the Carrera has the most polygons in car1 - 83). I ran your Ranger under DOSBox and there was some visible frame skip at 9000 cycles. It would probably be a good idea to merge most triangles into quads - the process would be somewhat annoying, but from looking at the shape I'd say it should be possible to reduce to ~100 polygons at least without losing detail. Another point is that, should you desire to have a car0 with extra details in relation to car1, you'd most likely need merging the polygons (as the upper limit for number of polygons is 255).

(For the record, there's another factor that may be involved with performance: the culling mechanism. Dstien is currently studying how to implement auto-generation of culling data, so we'll probably be in a better position to evaluate that in the near future)

Oh, and BTW I saw the wheels, and it seem to be a problem of coordinate orientations (which points are ahead/behind/above/below the wheel center). I'll try fixing it in a while and post back the .3SH .

Yes, and you'll have to add us as friends in order to see our "ladder". But nothing stops you from creating a "ghost" account only for that purpose I guess...

Strange things happened on this race - no hidden replays at all during quiet days ( ! - can anybody tell if that had ever happened before?) and a very tight scoreboard from 2nd to 7th despite relatively low activity during the final days. And then Ayrton bookends his title run by pulling off a shocking, even by his standards, eleventh-hour jump to impose his biggest victory in the whole season... all of that seems to indicate a section analysis would be very interesting, and, after performing it, one sees that's the case indeed. I only won't publish it now since I don't have BJ's replay - Zak accidentally put his 1:01.05 with Jaguar on the new track zip instead of the final Carrera lap. Once either BJ or Zak post said replay I'll update my spreadsheet and publish the analysis with further commentary.

He will have an experienced team around him to help, like Bilardo (Argentina's coach in World Cups of '86 and '90) and some other local coaches with good ideas and willing to work (Many of them took part in '86 team).

So it will be sort of a "management by committee", at least in the initial months... let's see how it goes.

I would have preferred to Bianchi, anyway Maradona can be good as coach. I think that the selection of a coach is not so important as the removal method, but yes this time, because we are in the limit of the classification.

Maradona at least has charisma and personality - one wonders how exactly that would convert into leadership, though, and how would the relationships within the management team would work. Anyway, more than the qualification itself I (from my unprivileged point of view) suppose the biggest fear should be that your NT goes down the same route Brazil went - playing sucky football in nine out of ten games. And having both Brazil and Argentina sucking at the same time would be bad for everyone, for football as a whole...

Looking at the bonuses it seems this will be a Audi/Vette race mostly, on the lines of Z83 - which feels like a good sign, as Z83 was my strongest ZakStunts race this year. And the layout is cute 

I ran the game without any other applications running other than AV and Firewall... Suitcase shuffle works great, I did an all correct 37 suitcases there, over 10000, but then the rest doesn't help, I get distracted or relaxed or something... And still can't get more than around 25 questions in Map Mayhem. How did you do to get 35 Dup?

For me all games run at similar speeds, even when running it on FF3 from a pendrive. I find Map Mayhem to be easier than the others because it requires shorter/simpler mouse movements and the visual hints from the maps are, on average, recognized faster. Comparing the three games, City Blitz is the harder for me by some margin...

What I feared... Damn!

Such anticipation... All I'm sure of is that I won't release a semi-finished version, though - Originally I was to fix release date at 30/11 as well, but with some luck it may be earlier than that.

Grazie, and I was thinking about a divulgation video as well - YouTube is there to spread the news...

Well, I could be wrong, but if this is not a real very crypted message, crossed with JacStunts... 

That was mostly subconscious, but still...

Great news! Will we be able to test it before the contest or are you keeping it top secret?

Well, I would prefer to avoid having to do multiple releases, be in the form of a "preview" intemediate release or as corrections needed on an early "final" release. As I tend to get obsessive with double-checking the smaller details while finishing a project, the final release will be, in all likelihood, quite close to the November 30th deadline. So you can say it's more on the "top secret" side   

and as for lag, the car 1 and car 0 are the same models, wich is 238 triangles permodel(not including wheels) i dont have dos box, so i ran it on a real dos computer using an intel 386 cpu @ 33mhz, not much lag with 2 cars loaded.

Wow, DOSBox is such a resource hog...

anyone here wana take a shot at the wheels?

OK, post it and I'll have a look at them later on...

whats the max amout of paint jobs stunts can handle?

The limit is 128, so in theory you can use (almost) as many as you wish (I have not tested implementing more than seven, though, so you'd better check if the car selection screen works normally in that case).

and why do my wheels glitch out when i turn my car to the left or right

Wheel rendering has a number of specificities that need to be taken care of. If you want to avoid trouble with them, first of all I suggest you to move the wheels to the first four positions on the polygon list and order them like this: front-left, front-right, back-right, back-left. Vertex order within the wheels matters as well, so the safest route is to clone exactly the coordinate values from some other car and then just displace them to the correct position.

Two extra tips that might be of help: First, be sure you're using the two-sided flag for any polygons which inner surface is to be exposed (as most likely is the case in the rear part of the truck body). Also, I noticed from your screen shots that your model is all composed by triangles. In case the absence of n-gons means there is a lot more than about 90 polygons on car1, it would be a good idea to double-check if there is no risk of frame rate issues (I had problems when trying to use a downscaled version of Corvette's car0, with 200 polygons). Should you feel the need to test that, the best route would be launching DOSBox with two cars loaded and lower the number of cycles progressively. If the frame rate remains normal up to ~8000 cycles in all likelihood there will be no problems. Else, it could be useful to merge a few triangles into n-gons (a number of .OBJ editors can't do that properly, but Anim8or can - see the first post on page 16 of the "Dissecting" thread should you need it).

Good find - with slightly different proportions it could well pass for a real kanji for the naïve... But Japanese (or Chinese) ideograms can be a bit shocking at times. For instance, I was quite perplexed to find in an online dictionary that this kanji, , which is just the one for "woman" () written three times, is used in the verb "to commit adultery", as well in words for a number of perversions... 

looks great, cant wait to drive it. mabey we can make a video of all the new mod cars...

Grazie, and I was thinking about a divulgation video as well - YouTube is there to spread the news...

I think it's more important to complete these cars first: dashboards, "tuning" with CarBlaster...

Yes, that's for sure - our machines need to be really well polished for the big show starting on November 30th, especially if it is going to be broadcast all over the Internet  . I'm somewhat fortunate in that respect, as most of the CarBlaster tuning for the Skyline was already done by early March, so other than adjusting the dashboard parameters in CAR*.RES, only some extra testing and possible last-minute corrections will be necessary. As for the dashboard, I managed to perform the limiting step, namely decomposing the NSX "dash" bitmap into its constituting elements, with less suffering and better results than expected, so I'm pretty optimistic about the progress of this part of the design.

We should open a new section for custom cars with the Stunts files and OBJ graphics...

Indeed, that's something which is pending for a while. Probably a "Game Modification"/"Modding" main category with file format descriptions as well as the pages for Stressed and CarBlaster would be a good start. It could be created be a subcategory of "Game information" as well, but then it might be a bit too hidden from the public eye.

Speaking of the Wiki, yesterday I noticed the SQL issue mentioned on my previous post also compromises the "Search" function...