|
|
Post by steelerswillis on Oct 19, 2020 9:27:05 GMT -5
|
|
obmar
Weekend Golfer
Posts: 117
|
Post by obmar on Oct 19, 2020 13:16:43 GMT -5
its all about the EPSG that matches the LAS/LAZ files and open street map
finding the right files/tiles is just the beginning (unless you get lucky and EPSG is in the metadata)
|
|
|
|
Post by steelerswillis on Oct 19, 2020 13:56:36 GMT -5
I’ve got all of the that stuff down dude. It’s just where I can capture it. I’ve got 7 course plots done today. All in England UK. I want to do some US, Canadian & other worldwide courses also.
|
|
|
|
Post by steelerswillis on Oct 19, 2020 16:00:42 GMT -5
Does anyone know if there is any coastal Welsh Lidar data for my Boyo’s?
|
|
obmar
Weekend Golfer
Posts: 117
|
Post by obmar on Oct 19, 2020 22:43:22 GMT -5
|
|
|
|
Post by steelerswillis on Oct 19, 2020 23:34:55 GMT -5
|
|
|
|
Post by steelerswillis on Oct 20, 2020 9:00:33 GMT -5
Does anyone know how to change a line to an area on Open Street Maps without deleting and restarting?
|
|
|
|
Post by sirish19 on Dec 30, 2020 4:29:42 GMT -5
Hi, there isn't actual LiDAR data available for Japan but there is an open source of 5M DEM data available. I have built a few Japanese courses using this data and it is alright for the most part. Since this is DEM data, it is also not classified so I use Global Mapper to classify the data and convert to LAS. The good thing about this that you don't need any EPSG code. Let me know which Japanese courses you are interested in building and I will see about getting you the data file. Cheers, Stephen |
|
mal
Amateur Golfer
Posts: 214
|
Post by mal on Dec 30, 2020 8:24:55 GMT -5
Hi, there isn't actual LiDAR data available for Japan but there is an open source of 5M DEM data available. I have built a few Japanese courses using this data and it is alright for the most part. Since this is DEM data, it is also not classified so I use Global Mapper to classify the data and convert to LAS. The good thing about this that you don't need any EPSG code. Let me know which Japanese courses you are interested in building and I will see about getting you the data file. Cheers, Stephen EPSG codes are inside the header of the LAS files and are easily found via various free tools available to download. A 5M raster data source will result in a very inaccurate heightmap that will contain lots of interpolation, and classifying ground without actual survey data can be very hit or miss. You would be better off sculpting the surfaces manually while referencing topo imagery. I played some of your Japanese courses and I applaud your effort. However, there are striations running throughout the terrain due to missing elevation data and your surfaces lack all the nuance that the course actually has. |
|
mal
Amateur Golfer
Posts: 214
|
Post by mal on Dec 30, 2020 8:47:23 GMT -5
Does anyone know how to change a line to an area on Open Street Maps without deleting and restarting? Literally tag the line as "area" "yes" |
|
|
|
Post by sirish19 on Dec 30, 2020 11:31:29 GMT -5
Hi, there isn't actual LiDAR data available for Japan but there is an open source of 5M DEM data available. I have built a few Japanese courses using this data and it is alright for the most part. Since this is DEM data, it is also not classified so I use Global Mapper to classify the data and convert to LAS. The good thing about this that you don't need any EPSG code. Let me know which Japanese courses you are interested in building and I will see about getting you the data file. Cheers, Stephen EPSG codes are inside the header of the LAS files and are easily found via various free tools available to download. A 5M raster data source will result in a very inaccurate heightmap that will contain lots of interpolation, and classifying ground without actual survey data can be very hit or miss. You would be better off sculpting the surfaces manually while referencing topo imagery. I played some of your Japanese courses and I applaud your effort. However, there are striations running throughout the terrain due to missing elevation data and your surfaces lack all the nuance that the course actually has. mal Thanks for your feedback. Can you show some examples of the striations and lack of nuance in my courses that you have played? I am still a noob and have been designing for a relatively short time so still trying to learn and pick up advice anywhere I can get it. I am mostly designing for the purpose of playing my local courses on the simulator. I have played all of these courses IRL and while 5M data is not nearly as accurate as 5 cm LiDAR data, I have found most of them to be fairly accurate. I am still working on my sculpting technique but don't think I will ever get to the skill level where I would be able sculpt a course manually by referencing topographic imagery (especially a lot of mountainous courses here in Japan where we have significant elevation changes) so think the 5M data is at least a good starting point. Any specific advice you could give would be greatly appreciated. |
|
mal
Amateur Golfer
Posts: 214
|
Post by mal on Dec 30, 2020 12:36:38 GMT -5
EPSG codes are inside the header of the LAS files and are easily found via various free tools available to download. A 5M raster data source will result in a very inaccurate heightmap that will contain lots of interpolation, and classifying ground without actual survey data can be very hit or miss. You would be better off sculpting the surfaces manually while referencing topo imagery. I played some of your Japanese courses and I applaud your effort. However, there are striations running throughout the terrain due to missing elevation data and your surfaces lack all the nuance that the course actually has. mal Thanks for your feedback. Can you show some examples of the striations and lack of nuance in my courses that you have played? I am still a noob and have been designing for a relatively short time so still trying to learn and pick up advice anywhere I can get it. I am mostly designing for the purpose of playing my local courses on the simulator. I have played all of these courses IRL and while 5M data is not nearly as accurate as 5 cm LiDAR data, I have found most of them to be fairly accurate. I am still working on my sculpting technique but don't think I will ever get to the skill level where I would be able sculpt a course manually by referencing topographic imagery (especially a lot of mountainous courses here in Japan where we have significant elevation changes) so think the 5M data is at least a good starting point. Any specific advice you could give would be greatly appreciated. If your target is only 2019 with a simulator, it doesn't matter much since the resolution is mostly lost with how the simulators transfer your ball flight/movement in/out the game anyway (especially with putting). All you really need is an accurate OSM containing proper sight lines and even 20M data would probably provide a "fairly accurate" experience. For examples of missing elevation data causing striations, go into camera mode and raise your viewpoint. You can see diagonal undulations that are artifacts from the interpolation process. This is a result of not having enough data points for reference in each square meter for the heightmap. The Japanese 5M DEM raster data is derived from 10 cm vertical point and 1 meter horizontal point LAS data that has been destructively compressed to make it suitable for online tile map imagery. Attempting to reverse this process is much like taking a 64k heavily compressed mp3 file and decompressing it and expecting it to now be identical to an original digital audio studio master. It just doesn't work. The data you are working with has 1 data point (averaged from original data) per 5 square meter while the common point LAS file has multiple vertical reference points in each square meter with resolution accuracy in the few cm range. There is no comparison, but chadtools does a pretty good job of fabricating data to fill in all the holes. The drastic elevation changes of the mountainous Japanese courses masks the problems with your data, but further exacerbates the accuracy of the fairway and putting surfaces by having large deltas to interpolate. You end up with everything kind of "smoothed out" and "guestimated." Take your Tribute course lidar data and import a few holes with no masking at all, then do the same but with the whole area completely masked. Compare the size of the files, and then look at the terrain in the game. Or another way to see this is to grab lidar data from NOAA and select FEET for horizontal and METER for vertical. Process the lidar data in chadtools with the FEET EPSG code and see how the terrain gets rendered. Anyway, my comment was more about you suggesting the DEM files as a suitable alternative to proper point LAS files. |
|
|
|
Post by sirish19 on Dec 30, 2020 13:51:34 GMT -5
If your target is only 2019 with a simulator, it doesn't matter much since the resolution is mostly lost with how the simulators transfer your ball flight/movement in/out the game anyway (especially with putting). All you really need is an accurate OSM containing proper sight lines and even 20M data would probably provide a "fairly accurate" experience. For examples of missing elevation data causing striations, go into camera mode and raise your viewpoint. You can see diagonal undulations that are artifacts from the interpolation process. This is a result of not having enough data points for reference in each square meter for the heightmap. The Japanese 5M DEM raster data is derived from 10 cm vertical point and 1 meter horizontal point LAS data that has been destructively compressed to make it suitable for online tile map imagery. Attempting to reverse this process is much like taking a 64k heavily compressed mp3 file and decompressing it and expecting it to now be identical to an original digital audio studio master. It just doesn't work. The data you are working with has 1 data point (averaged from original data) per 5 square meter while the common point LAS file has multiple vertical reference points in each square meter with resolution accuracy in the few cm range. There is no comparison, but chadtools does a pretty good job of fabricating data to fill in all the holes. The drastic elevation changes of the mountainous Japanese courses masks the problems with your data, but further exacerbates the accuracy of the fairway and putting surfaces by having large deltas to interpolate. You end up with everything kind of "smoothed out" and "guestimated." Take your Tribute course lidar data and import a few holes with no masking at all, then do the same but with the whole area completely masked. Compare the size of the files, and then look at the terrain in the game. Or another way to see this is to grab lidar data from NOAA and select FEET for horizontal and METER for vertical. Process the lidar data in chadtools with the FEET EPSG code and see how the terrain gets rendered. Anyway, my comment was more about you suggesting the DEM files as a suitable alternative to proper point LAS files. mal Thanks for the explanation although I am not sure I understood most of it, haha. I do understand that there are way fewer data points and a lot more smoothing out though making it much less accurate. I was really hoping for more of a specific example to one of the Japanese courses where it is obvious that the nuances of the course have been lost. You mentioned the Tribute course as an example, but just to confirm, the Tribute is a US course created using actual LiDAR. steelerwillis was asking if anyone knew where/how to get Japanese LiDAR and I simply informed that there is no LiDAR data available for Japan and that the only available data was 5M DEM. I even called the Geographical Survey Institute in Japan to try to get the original data but they won't give that out or even let you buy it (save for universities or other government institutions that need it for research purposes). Whether or not that is a suitable alternative to proper point LAS files as opposed to trying to manually sculpt an entire course is really subjective. I am not an engineer so don't have any expectation that I could manually sculpt a surface by referencing topographical imagery and get anywhere close to one accurate data point per 5 meters. |
|
mal
Amateur Golfer
Posts: 214
|
Post by mal on Dec 30, 2020 14:25:15 GMT -5
If your target is only 2019 with a simulator, it doesn't matter much since the resolution is mostly lost with how the simulators transfer your ball flight/movement in/out the game anyway (especially with putting). All you really need is an accurate OSM containing proper sight lines and even 20M data would probably provide a "fairly accurate" experience. For examples of missing elevation data causing striations, go into camera mode and raise your viewpoint. You can see diagonal undulations that are artifacts from the interpolation process. This is a result of not having enough data points for reference in each square meter for the heightmap. The Japanese 5M DEM raster data is derived from 10 cm vertical point and 1 meter horizontal point LAS data that has been destructively compressed to make it suitable for online tile map imagery. Attempting to reverse this process is much like taking a 64k heavily compressed mp3 file and decompressing it and expecting it to now be identical to an original digital audio studio master. It just doesn't work. The data you are working with has 1 data point (averaged from original data) per 5 square meter while the common point LAS file has multiple vertical reference points in each square meter with resolution accuracy in the few cm range. There is no comparison, but chadtools does a pretty good job of fabricating data to fill in all the holes. The drastic elevation changes of the mountainous Japanese courses masks the problems with your data, but further exacerbates the accuracy of the fairway and putting surfaces by having large deltas to interpolate. You end up with everything kind of "smoothed out" and "guestimated." Take your Tribute course lidar data and import a few holes with no masking at all, then do the same but with the whole area completely masked. Compare the size of the files, and then look at the terrain in the game. Or another way to see this is to grab lidar data from NOAA and select FEET for horizontal and METER for vertical. Process the lidar data in chadtools with the FEET EPSG code and see how the terrain gets rendered. Anyway, my comment was more about you suggesting the DEM files as a suitable alternative to proper point LAS files. mal Thanks for the explanation although I am not sure I understood most of it, haha. I do understand that there are way fewer data points and a lot more smoothing out though making it much less accurate. I was really hoping for more of a specific example to one of the Japanese courses where it is obvious that the nuances of the course have been lost. You mentioned the Tribute course as an example, but just to confirm, the Tribute is a US course created using actual LiDAR. steelerwillis was asking if anyone knew where/how to get Japanese LiDAR and I simply informed that there is no LiDAR data available for Japan and that the only available data was 5M DEM. I even called the Geographical Survey Institute in Japan to try to get the original data but they won't give that out or even let you buy it (save for universities or other government institutions that need it for research purposes). Whether or not that is a suitable alternative to proper point LAS files as opposed to trying to manually sculpt an entire course is really subjective. I am not an engineer so don't have any expectation that I could manually sculpt a surface by referencing topographical imagery and get anywhere close to one accurate data point per 5 meters. I know the Tribute is a US course and am very familiar with it, and I only suggested this since it is a course you have done with decent lidar data. I was pointing out a method that you can simulate what happens when a conversion is done from lidar to DEM files like you are using for the Japanese courses and you can actually then see why you have lost much of the terrain nuance in your Japanese courses. It is obvious to someone who works with lidar data that the greens have been averaged out and smoothed compared to their real-life counterpart. If you are not able to see (and feel) the striations in the elevation changes on your courses then I suppose it doesn't really matter, and I stand by the comment that it would be better to build the course with a topographic slope file and google earth pro where the in-game tools let you get down to roughly 1inch vertical resolution over a space of less than a foot. You are working with something that represents a single data point contained within about 55 square feet and some of your greens would be only about 5 of these data points, there isn't much data to build an accurate putting surface with, hopefully that makes sense to you. edit: I want to make sure you understand that I do think you have done an admirable job with the Japanese courses and they can certainly supply a fun and interesting round in the game. |
|
|
|
Post by sirish19 on Dec 30, 2020 22:35:19 GMT -5
I know the Tribute is a US course and am very familiar with it, and I only suggested this since it is a course you have done with decent lidar data. I was pointing out a method that you can simulate what happens when a conversion is done from lidar to DEM files like you are using for the Japanese courses and you can actually then see why you have lost much of the terrain nuance in your Japanese courses. It is obvious to someone who works with lidar data that the greens have been averaged out and smoothed compared to their real-life counterpart. If you are not able to see (and feel) the striations in the elevation changes on your courses then I suppose it doesn't really matter, and I stand by the comment that it would be better to build the course with a topographic slope file and google earth pro where the in-game tools let you get down to roughly 1inch vertical resolution over a space of less than a foot. You are working with something that represents a single data point contained within about 55 square feet and some of your greens would be only about 5 of these data points, there isn't much data to build an accurate putting surface with, hopefully that makes sense to you. edit: I want to make sure you understand that I do think you have done an admirable job with the Japanese courses and they can certainly supply a fun and interesting round in the game. mal Hi, thanks for the feedback. Certainly no offense taken in any of this and I really do appreciate the feedback and dialogue, this is the best way for me, and anyone else reading these forums to learn. I generally post pics here and on the simulator forum when I publish a course and do ask for feedback but generally don't get any feedback, at least not enough to take and improve on the next course, so any insight is helpful. You mentioned above building a course with a topographic slope file and google earth pro. What do you mean by topographic slope file and where would you get something like that for Japan for example? When building my courses, I start with the DEM file for the basic plot and do use Google Earth, photos and videos of the course (I generally take my gopro out for a couple of rounds on these courses prior to building), and also use course yardage books and green books where available to further sculpt the greens (see below). So from those, and from what I can tell from playing the courses in real life and on the sim in 2019, I feel like I am getting a fairly accurate representation of the course. On my more recent courses, thanks to some feedback here, I have started work on sculpting the bunkers more so hopefully that is improving. I am just wondering if you are actually seeing some artifacts in the greens or in these courses that make it obvious to someone who has never seen the course IRL that nuances are lost or if you are just assuming that they are since DEM data is being used rather than proper LiDAR. If you could point something specific out in a screen shot and suggest how that could be improved, that would be very helpful. I am still not sure how it could be better to start with a flat course and then sculpt manually the entire course based on images in Google Earth than to start with 5M data and then sculpt from there while referring to images. Again, especially when you are building a mountain course and you have dramatic elevation changes throughout the entire plot. There are a number of courses here that I have wanted to build but there is actually no data available at all, or there are sections of missing data throughout the course. For those, I have given up on trying to build them as it really would mean building from scratch.    |
|
