|
|
Post by chadgolf on Mar 2, 2019 21:37:01 GMT -5
Did you see anything about reprojecting data? That can cause these slight mismatches too. Mixing data outside of a single "contracted flight plan" will have a lot of small issues. For instance, if you had 2 West Virigina plots and one Ohio plot. Or Two 2010 lidar file and once 2015 lidar file.
But then again, we're at the mercy of the accuracy of the lidar companies, the governments, and the sat companies. This is all extremely complicated stuff and getting it right is a good career for many people.
|
|
|
|
Post by B.Smooth13 on Mar 4, 2019 7:34:39 GMT -5
I've been thinking about this one for a few days as I've tinkered, and I always come back to 1 thing that I think is a bit "off" with the terrain. I've never been to Pasa, obviously, so I have no first hand experience at the course. All I know is what I've read, and when I created the plot I found a hole-by-hole write up from someone that I used as reference to see how close the generated course was to the real thing. It was really spot on, but the thing that seemed off was scale, especially on the NorthEast corner of the plot (back 9).
The 11th hole is the one that stood out the most. In the description I read, the author noted that the approach shot was uphill by something around 15-20ft. In game, however, it's an uphill shot of around 30-40ft, so it's basically double what I would have expected. The most obvious reason that I can think of for this ties back into to fact that I had to force scale on 1 of the 3 LIDAR data sources that I used. I used 3 in total, 2 of which appeared to be identical with the .034etc scale, but the third was .9999. Since .034 was used by 2 of the 3, I used that to force the scale of the 3rd. chadgolf could that be an answer as to why the terrain in that portion of the plot seems especially exaggerated? Even beyond that 1 section, some of the elevation variation just seems overdone in areas. I haven't gone into GE to check their elevation estimates to see if that is actually the case, and I'd love to go back and tinker with the settings in Chad's program to see if anything makes a difference, but since I've already started tweaking the layout in-game, I don't want to regenerate the course file and lose that work.
I'm not overly concerned with having the elevation exaggerated, if that is the case, since it'll probably make it more fun to play in a video game, but it's something that has always felt a bit "off."
|
|
|
|
Post by chadgolf on Mar 4, 2019 11:23:03 GMT -5
I've been thinking about this one for a few days as I've tinkered, and I always come back to 1 thing that I think is a bit "off" with the terrain. I've never been to Pasa, obviously, so I have no first hand experience at the course. All I know is what I've read, and when I created the plot I found a hole-by-hole write up from someone that I used as reference to see how close the generated course was to the real thing. It was really spot on, but the thing that seemed off was scale, especially on the NorthEast corner of the plot (back 9).
The 11th hole is the one that stood out the most. In the description I read, the author noted that the approach shot was uphill by something around 15-20ft. In game, however, it's an uphill shot of around 30-40ft, so it's basically double what I would have expected. The most obvious reason that I can think of for this ties back into to fact that I had to force scale on 1 of the 3 LIDAR data sources that I used. I used 3 in total, 2 of which appeared to be identical with the .034etc scale, but the third was .9999. Since .034 was used by 2 of the 3, I used that to force the scale of the 3rd. chadgolf could that be an answer as to why the terrain in that portion of the plot seems especially exaggerated? Even beyond that 1 section, some of the elevation variation just seems overdone in areas. I haven't gone into GE to check their elevation estimates to see if that is actually the case, and I'd love to go back and tinker with the settings in Chad's program to see if anything makes a difference, but since I've already started tweaking the layout in-game, I don't want to regenerate the course file and lose that work.
I'm not overly concerned with having the elevation exaggerated, if that is the case, since it'll probably make it more fun to play in a video game, but it's something that has always felt a bit "off."
Can you post the names and/or source for the lidar data? Force lidar unit should only be used in most dire of circumstances. I bet some things are off because the EPSG code is US Surveyor Foot but the data got projected in something else like International Foot or Meter. US surveyor foot is 0.3048006 where International Foot is 0.3048, makes a big difference over a full course. Force EPSG is fine on its own, and already says what unit the data is in. I haven’t found a true need for Force Lidar Unit yet so maybe it needs to be removed. The last remote chance is that the ground distances are in feet while the vertical elevation is in meters, that could exaggerate heights. I consider this the worst result possible because it’s totally inaccurate and is WHY WE CRASH MARS ROVERS INTO THE SURFACE. As far as I know unless requested from NOAA incorrectly, I haven’t found this in the wild. Let me know and I’ll take a look and I can send you updated EPSG/or fix the raw lidar data so you don’t crash into Mars. |
|
|
|
Post by B.Smooth13 on Mar 4, 2019 11:37:10 GMT -5
Can you post the names and/or source for the lidar data? Force lidar unit should only be used in most dire of circumstances. I bet some things are off because the EPSG code is US Surveyor Foot but the data got projected in something else like International Foot or Meter. US surveyor foot is 0.3048006 where International Foot is 0.3048, makes a big difference over a full course. Force EPSG is fine on its own, and already says what unit the data is in. I haven’t found a true need for Force Lidar Unit yet so maybe it needs to be removed. The last remote chance is that the ground distances are in feet while the vertical elevation is in meters, that could exaggerate heights. I consider this the worst result possible because it’s totally inaccurate and is WHY WE CRASH MARS ROVERS INTO THE SURFACE. As far as I know unless requested from NOAA incorrectly, I haven’t found this in the wild. Let me know and I’ll take a look and I can send you updated EPSG/or fix the raw lidar data so you don’t crash into Mars. I'll go back today and grab the text output that I saw upon initial loading in of the LIDAR data. The first time I did it, I saw the different scale numbers, and the program actually froze entirely, like it couldn't compute with 2 different scales (and/or my CPU simply couldn't handle it). I'll let you know what I find. |
|
|
|
Post by B.Smooth13 on Mar 4, 2019 12:10:18 GMT -5
It got further into the process this time, but ultimately shows a memory error, making me think it's a limitation on my end via CPU ability? Anyway, you can see the different scales, and as predicted, different EPSG values, too. What would you recommend? I'd like to get the base terrain as "normal" as possible, obviously. Appreciate the help. |
|
|
|
Post by B.Smooth13 on Mar 4, 2019 18:30:47 GMT -5
FYI if anyone wants to see a raw import, search “pasLIDraw” in game.
|
|
|
|
Post by chadgolf on Mar 4, 2019 19:54:51 GMT -5
It got further into the process this time, but ultimately shows a memory error, making me think it's a limitation on my end via CPU ability? Anyway, you can see the different scales, and as predicted, different EPSG values, too. What would you recommend? I'd like to get the base terrain as "normal" as possible, obviously. Appreciate the help. Only in a round about way. Since the units are off, it's trying to draw a golf course for you that's as big as 1/4 of the continental United States. I'll look into this soon. |
|
|
|
Post by gamesdecent on Mar 4, 2019 21:03:20 GMT -5
Well, they don’t call it the US Open for nothing...
|
|
|
|
Post by chadgolf on Mar 5, 2019 1:27:37 GMT -5
The most obvious reason that I can think of for this ties back into to fact that I had to force scale on 1 of the 3 LIDAR data sources that I used. I used 3 in total, 2 of which appeared to be identical with the .034etc scale, but the third was .9999. Since .034 was used by 2 of the 3, I used that to force the scale of the 3rd. chadgolf could that be an answer as to why the terrain in that portion of the plot seems especially exaggerated?" I've looked at this, and the data.... isn't great but looks salvagable. I used "Force Lidar EPSG Projection" 2227. Leave Force Lidar Units BLANK I used 4 files: Processing: ARRA-CA_CentralCoast-Z3_2010_000026.laz Processing: ARRA-CA_CentralCoast-Z3_2010_000075.laz Processing: ARRA-CA_CentralCoast-Z3_2010_000083.laz Processing: ARRA-CA_CentralCoast-Z3_2010_000171.laz Maybe I have some uneeded data, but I usually grab extra. There is INCORRECT METADATA encoded in the files, but from what I can tell they are all consistent to each other. The EPSG 2227 will make them all consistent, and I don't see any of the misalignment anymore. The 11th hole is the one that stood out the most. In the description I read, the author noted that the approach shot was uphill by something around 15-20ft. In game, however, it's an uphill shot of around 30-40ft, so it's basically double what I would have expected.
As for the heights, I can't really confirm because Google Earth matches what the game produces. BUUUUT they may also have been fooled by bad data. I looked and looked and can't find any independent references for the altitude change. 11th green is unplayable, but could be flattened a little bit or just the game physics. I can't tell if that's because it's higher or not. One thing I've learned is that the golf course descriptions are almost always wrong. Just like yardage markers and score card yardages sometimes don't agree with rangefinders, so it seems elevation is unclear. On the HB forums, someone was adamant that because one of the hole as Muirfield had the description "plays gently downhill" that it couldn't have it's actual 'U' shape. If you can find a source, I can scale the elevations for you to make up for this data. But if it's fooled us, it's fooled everyone. |
|
|
|
Post by B.Smooth13 on Mar 5, 2019 7:12:37 GMT -5
chadgolf thanks for the help, I'll grab that 4th data source and use the force EPS 227 format. As for the height/elevation, I'm not overly concerned with it. I know the greens are meant to have massive undulations anyway, and I've got no problem going through and making them playable in accordance with whatever resources I can find that map out contours, or at least describe them. This is a course I'd considered tackling before LIDAR was possible, so I already feel like I've skipped ahead a month or two by not having to figure out the elevation by hand.
|
|
|
|
Post by B.Smooth13 on Mar 23, 2019 11:11:55 GMT -5
After doing the Nicklaus course and getting familiar with what I'm looking at, and going back here to start making meaningful progress, I've been spinning my wheels. There's something off with the alignment of heightmap to textures - specifically bunkers - that's driving me crazy. I've remapped the first 3 holes 4 or 5 times now, using different map sources to try and figure out which one matched up with the LiDAR data, and they're all just seemingly wrong in different ways lol. Guess I'll learn how to do the plot shift, because there's a chance the course/textures/etc. just need to move a bit North to match the terrain better, but that's just more learning that I haven't had to do yet. Frustrating, as it feels like I've wasted a few hours trying to figure this out and didn't get any closer to a solution that I was before I started...ugh
|
|
|
|
Post by B.Smooth13 on Mar 27, 2019 6:29:09 GMT -5
chadgolf talk to me about terrain scaling...
I'm asking in this thread, because it's the only course where something has just felt "off." I spent the better part of an hour yesterday doing tons of terrain shifting (and playing with spline settings) to try and line up bunkers with the obvious terrain they fit in. I originally thought it was due to me mapping in OSM with the wrong background image source, but that wasn't it (although some are off by a few feet, but always in the same direction which wasn't the problem). Then I thought all that needed done was some terrain shifting, which helped, but still isn't really as accurate as I'd like. The confounding thing is, depending upon which bunker on which hole on which side of the property you look, the edges of bunkers can be off from where the terrain shows they should be by a few feet, but not all of them are in the same direction. I haven't spend tons of hours confirming this, mostly because I'm not entirely sure how I would do it, but it seems like there is a general scaling snag somewhere that's throwing things off - i.e., from the middle of the plot out to the edges is something like 2ft/1px, but the terrain is 1.5ft/1px. I have no clue if that's the system of scaling you have included in the tool as an option, but it's how I'm thinking about the problem, so hopefully it makes sense.
So...other than just messing around with the scaling value, do you have any guidance to offer that would help me line things up better? If nothing else works, I can always manually touch up all the bunkers to ensure they look natural, but obviously it would be preferred to have the LiDAR data do that for me. Thoughts?
|
|
|
|
Post by chadgolf on Mar 27, 2019 8:37:34 GMT -5
I have this data downloaded and will take another look for you.
It’s very possible something else is still off like bad height units or they used the wrong foot.
I may write to them and confirm the details of the data/try to track down the data from an original source.
|
|
|
|
Post by B.Smooth13 on Mar 27, 2019 10:57:42 GMT -5
I have this data downloaded and will take another look for you. It’s very possible something else is still off like bad height units or they used the wrong foot. I may write to them and confirm the details of the data/try to track down the data from an original source. That would be fantastic, thanks very much for the help. I have held of on editing in-game since I haven't gotten the alignment down perfectly and didn't want to lose that work, so I'll hold off a bit longer until I hear back. Thanks again! |
|
|
|
Post by chadgolf on Mar 29, 2019 1:23:44 GMT -5
Haven't forgotten about this, but it may take me a while to hear back/look into details more. I think we both have others things to do, so we're not stuck, but we'd both like to see this get made.
|
|
International Record: 7-0-0