TrackSim can work with an image up to 4000x4000 pixels, but the current release dated "19Jul2" is 1088 x 1024 (tall x wide). The left half (512 pixel columns) is the same as the release last month. The top 256 pixel rows of that are essentially the same as the original release in 2018. Additional image fragments were added primarily to support the first three F-1/10 race tracks (U.Penn, Porto, and Turin). This coordinate list is divided into two sections, the original release, and then the F1/10 race track additions.
The current release contains (mostly in the right half) images to add verisimilitude to the "MSB" tracks, two areas where the 2019 summer program expects to be testing our development car in a variety of credible race configurations.
For more information on the Tagged Image File Format see the Aldus/Microsoft
Technical Memorandum dated 8/8/88.
Top 256 Pixel Rows
F-1/10 Additions
MSB Additions
Back to front of TrakSim
1 2 3 3 3 4 5 6 7 8 9 10
2. The dark traffic light can be used for off-center views, where the color of the light is directed and not visible except to vehicles on the street directly in front of it.
3. You can specify a traffic light shorter than 255 by 6 pixels or more, and the shadow will not appear on the street below it.
4. These coordinates include a faint gray outline around the white paint.
5. These coordinates exclude the faint gray outline around the white paint. You can use less width and height for a shorter or narrower line on the street.
6. There are eight pixels of line width starting row 240, but you can specify its position more precisely by including up to 12 additional (blank) pixels above (or below) the image in the file. You can also make the line narrower by starting on a later row or specifying less height.
7. These coordinates include a faint gray outline around the white lamp globe. You can make your luminary smaller by starting later (row and/or column) and/or making the height and width less, but your reduced rectangle will not be displayed as oval. Distant luminaries tend to lose pixels and show as non-oval anyway.
8. Used as paint in high-resolution (32ppm, option +4) mode, these numerals can be used to number key coordinates on the close-up map display. They could also be used in low-resolution and display much larger.
9. I scraped images for the front, rear, and two sides from the internet, then wrote a program to merge these into the four 45o views, with some manual fixups where the perspective didn't come out right. They look a little clunky, but workable.
10. There is a mode in TrakSim which lets you calibrate the servo range
for steering and speed. I added this warning to let users know that their
self-driving code had been overridden. If you don't use that mode, you
could recycle this space in the image file for other purposes. Actually,
you can do that with any of the image fragments in the file that you are
not using.
11 12 12 13 12 13 13 13 14 14 14 15 16 17 18 18 19 20 21 21
Compared to nearby standing people in the video, the Turin track seems to have used 6" (15cm) tubing, whereas the U.Penn and Porto tracks used 4" (10cm) tubing. This is easily achieved by adjusting the scale factor, the larger scale factor representing the smaller tubing.
12. You can spread the tube-wall end over several grids, and adjust the width of the image fragment used for it to match, because repeating the image fragment doesn't make sense when one end has a fold-over. Left ends have a well-defined starting point, but right ends may take some fiddling with the width and horizontal image coordinate to get it to come out right. The (minimum) widths shown are at the 32ppm resolution to cover one grid cell (2m = 64 pixels); for other wall heights (like 24ppm) you would reduce or increase the width proportionately. Note that the angled views are also specified for the 45o diagonal, which is 2 meters times the square root of 2.
13. In order to mitigate the bloated look of track boundary walls two (park) meters (10" = 25cm actual) wide when seen from the end, I replaced some of that width with empty space, which you can select by choosing a different starting column on the same image fragment. If you were watching it as your car moved from one side to the other, you'd see the end cap jump in the opposite direction, but TrakSim cars cannot look sideways while moving forward, so the effect is never actually seen. In any case these end-caps are only used for hair-pin turns, which so far have been limited to the Turin track.
14. The Porto track set up the finish line as a horizontal bar across the track, held up by two posts placed just in front of the tubing wall. I simulated this effect by making a one-grid stretch of the wall taller to include the post, and then making the crossbar into a luminary that doesn't actually give off any light, and that is visible only straight-on from the track. The spacing isn't perfect, but close enough to serve as a landmark for a driving program memorizing the track.
15. The Porto race was held in a very ornate European building with windows and balconies overlooking what appeared to be a covered patio. From the videos I made up a composite frame of a wall consisting of one lower alcove with the balcony and windows in that vertical segment, then repeated it all around the track area. Other scenery was then constructed as walls in front of that.
16. In addition to the tubing walls, the Turin track used as track boundaries, either singly or in a run, cards about 2 meters long by a half-meter tall, pre-printed with commercial-art houses along a street over a white background. I approximated the drawing and added a slight off-white cast to the sky so it would not be transparent. Because TrakSim tries to put a shorter wall in front of a taller when they both render at the same raster base, I also did some overlapping tube-walls from the Porto goal-post segment (but omitting the post) so it would be (transparently) taller than the cardboard, and therefore render behind the cardboard, as appears in the video.
17. The room used for the Turin track had some tables set up at one end, apparently for refreshments. Different videos showed the tables in slightly different configurations, so I picked one. They were always behind the houses cards at that end, so my screen capture got only the upper strip of view, which I manually extended downward to halfway, then left the bottom half blank.
18. I made a separate wall to represent the baseboard using a square swatch about the right color, then adjusting the scale factor to get the wall height correct. I also included blank space above the swatch to make the wall computer as taller, so it would come behind the houses card where they overlap. The F-1/10 rules webinar suggested flattened out cardboard as an alternate wall material, so I included a strip of uncolored cardboard that can be sized and repeated for that purpose.
19. Low-resolution brown stripes on floor as in Turin.
20. The image is only 44 pixels tall in the file, but by including different amounts of blank space below it, I could make the same fire extinguisher image display waist-high on one side of the Turin room (as an artifact), and shoulder-high on the other (as a wall segment with different views).
21. Sometimes I needed to verify the scale factor on an artifact, so
I created this precisely measured meter stick to put next to it in the
scene. When I was testing hardware mobility, I bought a 12" traffic cone
(that would be 2.5 park meters) to give an unmistakable landmark for the
software to look for. Then I needed to make sure the simulator worked the
same, so here's its image.
22 22 22 22 22 22 22 22
Tom Pittman
Rev. 2019 July 4