Messages

Could Stressed be implemented so to:
1) delete and duplicate vertexes in the same way (right click) you can delete and duplicate polygons?
2) select the polygons even by clicking with the mouse over each of them?

Mouse picking is implemented in SVN, it makes skinning a breeze. I'll make new Win32 build next time I boot into Windows. The vertex list is a fixed structure whose size depends on the selected primitive type. I guess I could make it more flexible, but you guys should really use a real modelling tool and use stressed only to finalize the shape and paint-jobs.

A question for dstien: upgrading the number of polygons related at shapes car1 and car2 (used during the race) can affect the fluidness of the game?

I haven't done any heavy testing, but it should only depend on the amount of available CPU cycles. If the framerate is ok when racing against an opponent with two official vehicles on the screen, there shouldn't be any difference using one vehicle with twice the amount of details. Another factor in addition to the number of polygons is the culling parameters that we haven't figured out yet...

Glad you asked that question, Mark, as it affects directly my experiments on the GT3 - the regular Corvette car1 shape has 53 polygons, while the downscaled car0 I'm adjusting to replace it has 201 (with a couple additional ones to cover the underside of the car). Unfortunately, quadruplying the polygon count seems to be a bit too much: With two equal cars on screen, the frame rate starts to suffer at about 16000 cycles, a value far too high for many pipsqueaks. Even with a single car the limit seems to be as high as 11000-12000. Since I always race at 20k cycles, I would never notice the difference by myself. That's a major blow to my current design approach...  

When you export the unknown culling values are discarded. When importing all bits are set. Maybe it helps to set these correctly? I think I've figured out roughly how the unknowns works, and stressed should recalculate these values automatically once implemented.

What about the the rear lamps when braking?

My (unproven) theory to explain the rear lamps is that they are painted with specific materials (mostly #45, but also #12 and #14 for GTO) so that the game knows it needs to lit them up (that is, change to another shade of red, likely #47) under braking. Maybe we should paint the car body as a whole with #45 and check whether it glows when braking... Anyway, the instruction to make the lamps glow is, in all likelihood, hard coded.

That's correct.

When I try to use it in the game, several polygons are missing from the bodywork, getting visible only when I turn the camera angle (but then other polygons disappear).

Nice work on the Ladas, CTG! The problem you're having is most likely that you have drawn the polygons inside-out. Enable "back-face culling" in the 3d view to display only the front faces.

3, save the file as stlm00.3sh? Or shall I totally rename it?

Use stlm00.3sh and move the old file out of the way. If not, Stunts will just load the original file. If you rename it entirely you'll have to rename the other files used by the car as well + editing the tuning file.

Strange... I modified a single polygon's color, saved the file and Stunts was frozen when I tried to use the car.

Stressed doesn't compress saved files. Did you rename it accordingly with the .3SH extension instead of .P3S?

I really like this idea. I'm more than willing to help out if there are any technical issues. Tuning is next on my list for resource type support in stressed. Can CarBlaster edit dashboard settings?

I do have some plans for making my own cars, maybe I have one ready for the 2012 season.

Nice work Mark and Duplode! The renewed interest in Stunts modding motivated me to continue introducing bugs. I'll put up a new binary "soon".
For Duplode and others following SVN, Google now has Atom feeds for project updates.

Can you change the shape of the steering wheel?

The steering wheel is just a bitmap image that you can replace with anything you want, but I think the blue indicator will still move along a fixed arc.

- running red lights at the front of the car (if the mills can "move", maybe some car elements also can - if not, there's still the possibility to put a thick red line)

Cars can't have animations except for the hardcoded wheel transformations.

A question for dstien: upgrading the number of polygons related at shapes car1 and car2 (used during the race) can affect the fluidness of the game?

I haven't done any heavy testing, but it should only depend on the amount of available CPU cycles. If the framerate is ok when racing against an opponent with two official vehicles on the screen, there shouldn't be any difference using one vehicle with twice the amount of details. Another factor in addition to the number of polygons is the culling parameters that we haven't figured out yet...

I don't know which polygon is what on the car.

A shade of red is added to selected polygons, but this simple highlighting fails on small details or surfaces that are already coloured red. Drag the mouse combined with right/left/middle/right+left buttons to rotate and translate the 3d view on different axes.

Finally got around compiling Stressed for Win32. Since Windows don't offer system-wide management of third party packages I've tried to include all dependencies (hence the file size). Tell me if I missed anything and if the program actually works at all. It should be able to open and edit resource files containing plain text, bitmap images and 3d shapes.



stressed-0.1.0-rc1-win32bin.zip

Maybe if we did some more editing there would be positive feedback for developement, though.

Further development can be done in exchange for cool Stunts creations or racing technique counseling.

I'll try to whip up some Windows builds soon*.

*) Definition of "soon" is subject to change.

Ok, it's settled, "soon" means "within 1/2 year".

just found some short tips on how to disassemble dos game, some may be interested:

The biggest problem with debugging Stunts is that the code is self-modifying, it can't just be loaded into a disassembler. LOAD.EXE patches together the executable code that needs to be loaded into memory based on graphics and sound setup. It would be interesting to know if the result from LOAD.EXE can be dumped to a DOS EXE image. Since the header files are included one might think so, and we've already reversed the worst parts of the loading process; data decompression.

When I reversed the run-length decompression located inside the variable-length compressed code files I used the DOSBox Debugger to locate the code and dump the memory. By importing the raw memory into a disassembler we can initiate code parsing since we know the locations containing interesting code. I'd like to avoid this messy workflow next time I explore Stunts code...

What about Duplode, Antti, dstein?

Mine isn't very exciting, it's just an abbreviation of my real name. I used to have a unique nick with a thrilling back-story, but abandoned it as the Internet became mainstream and I found myself having to explain and spell it in awkward social situations.

Antti is a very common Finnish male name, but with all the uncertainty and doubt surrounding Stunts scoreboard entries thanks to certain community members, nobody knows if he's real.

If you cancel the replay recording before it ends by itself, the replay is considered invalid (as the time is calculated from the replay length)

Indeed. I didn't realize this, it's possible to cheat up to -1 second by simply deleting 20 bytes.

I checked a few replays and it looks like 20 0x00 bytes will be added after the finish line is crossed. The problem is that the user can press escape after finishing, in one of my tests I could cancel the recording exactly at the finish line (no 0x00 bytes written). However, by using heuristics we could guess that a replay not ending with a few 0x00 bytes does not cross the finish line and at the very least ask the user for confirmation.

Also, the track section embeded in the replay file should be checked, but you probably do this already?

It's certainly not dead, just resting. I got distracted with some other projects lately. I tried compiling Stressed for Win32 when I saw Krys' post today, but it wouldn't link dynamically and static linkage resulted in a 11 megabytes executable.

Tracks, avatars and skins are exchanged when loading online games, that may suck some bandwidth. Like all other online real-time games connection latency is the most important factor.

But be carefull, this game is highly addictive. It'll affect your Stunts research.

I did (quite) like that. I change the name of original STPMIN.P3S to STPMIN Original.P3S so I could have them back, and then paste the STPMIN.3HS but it didn't work... Maybe stunts recognized the other file despite I changed it's name?

DosBox caches files. Did you do this while DosBox was running?

Now, free users of TMNF (TrackMania Nations Forever) car race versus full game users of TMUF (TrackMania United Forever) on stadium environnement.

More importantly, they finally ditched that horrible StarForce malware.

It would be cool to plan a date and time rendez-vous and create our private server to race with our favorite tracks.

My TM skills are on a par with my Stunts pipsqeakship, so don't expect much competition.

I wonder if it's possible to make a lot more detailed car skeletons for Stunts (~5000-10000 points polygon, like NFS PU cheat cars). Or more exactly: do you think Stunts would be able to handle more difficult polygons written in the same program?

Unfortunately the shape format is limited to 255 vertices and 255 polygons. The teapot has 114 vertices and 96 polygons, so the limit isn't that bad. Remember that Stunts don't have lighting, shading is faked by preset colors. Without real-time shading high-poly details won't be visible. The low screen resolution and lack of floating point co-ordinate precision doesn't help either. Anyway, I think Dosbox would struggle when rendering such a number of polygons in an emulated software pipeline, and it would probably hit some arcane memory handling barriers as well.

Stunts' classic collision and dynamics simulation wrapped in modern, hardware-accelerated rendering is probably our holy grail. A man can dream though. A man can dream...

Added the ability to replace all references to a material within a paint-job. This makes skinning a little less tedious. Just right-click on an entry in the materials table and select "Replace" from the context menu.



No graphics testing without teapots.
Attached the 3SH file with four paint-jobs in case someone want to play with it.