Missouri Lake Maps

[cagney]

Originally Posted by boat_alexandra

The great lakes charts use this projection and the have always worked.
..........

Those charts projection is Polyconic, not Lamberts conformal conical.

[.Paul.]

Sean,

Originally Posted by boat_alexandra

The great lakes charts use this projection and the have always worked.

With the recent projections improvement, you can even quilt them with charts of different projections.

If it doesn't work it should be fixed...

The Great Lakes charts from NOAA use either Polyconic projection or Mercator projection depending on the chart. These work in OpenCPN.

Ok... there should be an automatic way to convert them to kap. After all, kap is already tif.

Manually?? I thought it could be automatic.

There are some open source software tools that might be used to convert GeoTIFF to KAP. Google found some discussions concerning converting KAP to GeoTIFF. I only planned on converting one "historical" topo map into a chart. So, I did it manually because that was faster than searching for a program to speed up the process. The GeoTiff I downloaded to make the KAP came from database of historical maps at USGS. The current maps are downloaded as GeoPDF's, and they are much larger because they include many layers of information that is probably not going to be used for a navigation chart.

I don't see what the problem is from your log.

OpenCPN should be able to display these charts. I think the problem is you need the polynomial coefficients. Did you include them?

The Lambert Conformal Conic charts are downloaded from NOAA using the Chart Downloader plugin. The polynomial coefficients are in the KAP header, and the REF points are also there.

Sure you can also convert them to mercator, but this will either lose data, or you must use an output image dimensions much larger than the original image and in this case, the resulting file will be much larger. Furthermore, because of the simplistic compression scheme used by kap, once you convert projections, the compression will be terrible, so I imagine the resulting file 3-4 times larger for the same quality. It would be better to keep them in the native projection.

Here is the "other story" I mentioned in an earlier posting:

I use the NOAA Chart Reprojector to convert the Lambert Conformal Conic charts into Mercator charts. However, the NOAA Reprojector was written for BSB_2. It reads BSB_3, and it simply copies BSB_3 information into the new chart. The Reprojector does not calculate new polynomial coefficients for the reprojected chart it just copies them without changes. The polynomials were added with BSB_3. So, it is necessary to delete the polynomial coefficients from the charts that are output by the Reprojector. I delete them before feeding them into the Reprojector. I also delete the Notice to Mariners correction notes from the KAP headers. The Reprojector also deletes all of the color tables except RGB. So, my routine gets a copy of the color tables before running the Reprojector, and it puts them back into the new KAP chart afterwards.

The Reprojector does a good job of reprojecting the chart image without changing the color codes of the pixels. The chart images can get bigger--especially when it is used to remove skew to make a north-up chart. If the original chart is north-up, the reprojected chart is not much bigger.

Paul

[seandepagnier]

Originally Posted by cagney

Those charts projection is Polyconic, not Lamberts conformal conical.

You got me there... But I don't even think the great lakes charts are polyconic, at least the formula we are using isn't correct. Instead we rely completely on the coefficients in the chart.

I think you can change conformal conic to polyconic and it will work correctly. You need something like 20 reference points. Try changing it to polyconic and run bsbfix.

Otherwise, yes we could be even smarter and support conformal conic directly which would only require 2 reference points per chart, but if all of these charts already have enough reference points there wouldn't be any reason to.

I think we might like to reject all conic charts without sufficient reference points as we can not be sure of the latitudes used in the projection. The current polyconic fallback is probably never correct.

Originally Posted by .Paul.

Sean,
There are some open source software tools that might be used to convert GeoTIFF to KAP. Google found some discussions concerning converting KAP to GeoTIFF. I only planned on converting one "historical" topo map into a chart. So, I did it manually because that was faster than searching for a program to speed up the process. The GeoTiff I downloaded to make the KAP came from database of historical maps at USGS. The current maps are downloaded as GeoPDF's, and they are much larger because they include many layers of information that is probably not going to be used for a navigation chart.

Also might be interesting to convert geopdf to kap selecting which layer.

The Lambert Conformal Conic charts are downloaded from NOAA using the Chart Downloader plugin. The polynomial coefficients are in the KAP header, and the REF points are also there.

Which charts? I will try the downloader but cannot download all the charts in the catalog. Again, I think maybe they will work if you change the projection to polyconic provided they have poly coefficients or enough reference points (maybe need to run bsbfix)

I think this highlights a minor problem... opencpn should _not_ quilt charts other than mercator without opengl. The way it works currently is incorrect because each chart's projection is unique the cannot go together.

For example, without re projecting (no opengl) 2 transverse mercator charts unless they have the same meridian cannot quilt, and 2 polyconic charts unless using the exact same parallels and meridian could not quilt. Please correct me if I'm confused.

Since 4.2 with opengl, even elliptical mercator or any other type of projection can be correctly quilted in several supported output projections. Most raster charts are elliptical mercator, but vector charts are displayed using spherical mercator for performance reasons. Quilts mixing these two also is currently incorrect without opengl.

Often the "incorrect" factor is less than 1 pixel.... so maybe we should detect this case and allow it... but improving the non-opengl code seems a waste of time.

Here is the "other story" I mentioned in an earlier posting:

I use the NOAA Chart Reprojector to convert the Lambert Conformal Conic charts into Mercator charts. However, the NOAA Reprojector was written for BSB_2. It reads BSB_3, and it simply copies BSB_3 information into the new chart. The Reprojector does not calculate new polynomial coefficients for the reprojected chart it just copies them without changes. The polynomials were added with BSB_3. So, it is necessary to delete the polynomial coefficients from the charts that are output by the Reprojector. I delete them before feeding them into the Reprojector. I also delete the Notice to Mariners correction notes from the KAP headers. The Reprojector also deletes all of the color tables except RGB. So, my routine gets a copy of the color tables before running the Reprojector, and it puts them back into the new KAP chart afterwards.

The Reprojector does a good job of reprojecting the chart image without changing the color codes of the pixels. The chart images can get bigger--especially when it is used to remove skew to make a north-up chart. If the original chart is north-up, the reprojected chart is not much bigger.

Paul

I have used this reprojector. It was very slow. Also charts with text, or charts covering wide geographical areas (making bigger distortions) require a much larger output image dimension to avoid loss of resolution.

My experience with 14500 was it increased in size by 8x to get the same quality once remapped to mercator, but it is an extreme case.

[cummings66]

[QUOTE=boat_alexandra;2323582]

Actually, many sailors now have complete KAP navionics charts covering entire world except denmark. It's only 1.5gb and cost nothing.
/QUOTE]

Where would I find this? I'm assuming it also contains Missouri.

[.Paul.]

Originally Posted by boat_alexandra

Which charts? I will try the downloader but cannot download all the charts in the catalog.

Sean,

All of the Lambert Conformal Conic charts are in the Chart Downloader's Predefined catalog:

USA - NOAA & Inland charts
    RNC
        by States
            MN - Minnesota

Here is a summary of this catalog:

RncProductCatalogMN

number     title                                                                      scale     projection

  14500    Great Lakes, Lake Champlain to Lake of the Woods                          1500000    Polyconic
  14961    Lake Superior (Mercator Projection)                                        600000    Mercator
  14966    Little Girls Point to Silver Bay, including Duluth and Apostle Islands;... 120000    Polyconic
  14967    Beaver Bay to Pigeon Point;Silver Bay Harbor;Taconite Harbor;Grand Marais  120000    Polyconic
  14968    Grand Portage Bay, Minn. to Shesbeeb Point, Ont.                           120000    Polyconic
  14975    Duluth-Superior Harbor;Upper St. Louis River                                15000    Polyconic
  14982    North Lake                                                                  42240    Lambert Conformal Conic
  14983    Northern Light Lake                                                         42240    Lambert Conformal Conic
  14984    Sea Gull Lake                                                               42240    Lambert Conformal Conic
  14985    Saganaga Lake                                                               42240    Lambert Conformal Conic
  14986    Knife Lake                                                                  42240    Lambert Conformal Conic
  14987    Basswood Lake, Eastern Part                                                 42240    Lambert Conformal Conic
  14988    Basswood Lake, Western Part                                                 42240    Lambert Conformal Conic
  14989    Crooked Lake                                                                42240    Lambert Conformal Conic
  14990    Basswood River                                                              10000    Lambert Conformal Conic
  14991    Lac la Croix                                                                42240    Lambert Conformal Conic
  14992    Sand Point Lake to Lac la Croix, including Crane Lake and Little Vermilon   42240    Lambert Conformal Conic
  14993    Namakan Lake, Eastern Part                                                  42240    Lambert Conformal Conic
  14994    Namakan Lake, Western Part and Kabetogama Lake, Eastern Part                42240    Lambert Conformal Conic
  14995    Western Kabetogama Lake                                                     42240    Lambert Conformal Conic
  14996    Rainy Lake-Big Island, Minn., to Oakpoint Island, Ont.;Kettle Falls         25000    Universal Transverse Mercator
  14997    Rainy Lake-Dryweed Island, to Big Island                                    25000    Universal Transverse Mercator
  14998    Rainy Lake-International Falls to Dryweed Island                            25000    Universal Transverse Mercator

Since the Chart Downloader does not list chart numbers for the NOAA RNC's, you will have to select the individual charts by title. The chart titles can be seen in the Code-box, above. If you scroll the Code-box to the right, you should see the scale and projection for each chart.



Paul

[.Paul.]

Originally Posted by boat_alexandra

I have used this reprojector. It was very slow. Also charts with text, or charts covering wide geographical areas (making bigger distortions) require a much larger output image dimension to avoid loss of resolution.

My experience with 14500 was it increased in size by 8x to get the same quality once remapped to mercator, but it is an extreme case.

I agree that the NOAA Chart Reprojector is slow, and it does increase the size of the KAP a little. Here are the file sizes reported by the DIR command in a CMD prompt window (Windows 10) for NOAA chart 14500:

 Directory of c:\Charts\RNC\us\14500

01/27/2017  04:34         5,868,960 14500_1.KAP



 Directory of c:\Charts\RNC\us_reprojected

01/27/2017  06:39         6,010,095 14500_1_M.KAP

The original 14500_1.KAP file is 5,868,960 bytes.
The reprojected 14500_1_M.KAP is 6,010,095 bytes.
The size only increased by about 2.5 percent.

I prefer your reprojections in O with OpenGL enabled.

Paul

[seandepagnier]

Originally Posted by .Paul.

I agree that the NOAA Chart Reprojector is slow, and it does increase the size of the KAP a little. Here are the file sizes reported by the DIR command in a CMD prompt window (Windows 10) for NOAA chart 14500:

 Directory of c:\Charts\RNC\us\14500

01/27/2017  04:34         5,868,960 14500_1.KAP



 Directory of c:\Charts\RNC\us_reprojected

01/27/2017  06:39         6,010,095 14500_1_M.KAP

The original 14500_1.KAP file is 5,868,960 bytes.
The reprojected 14500_1_M.KAP is 6,010,095 bytes.
The size only increased by about 2.5 percent.

Yes, but is the quality actually the same when zoomed in so you can see the pixels? Compare the two charts, look at text etc...

Originally Posted by .Paul.

Sean,

All of the Lambert Conformal Conic charts are in the Chart Downloader's Predefined catalog:
DE]Since the Chart Downloader does not list chart numbers for the NOAA RNC's, you will have to select the individual charts by title. The chart titles can be seen in the Code-box, above. If you scroll the Code-box to the right, you should see the scale and projection for each chart.



Paul

First time I used the chart downloader... hmm! I think it's very annoying because I can only see 2 charts in the lower list at a time and no way to expand it. Why I can't select all?

Anyway, it's a bit broken because in the free wifi connection it tends to randomly inject ads into things which corrupted the charts. The plugin doesn't seem to notice this, but the charts don't work. So I do it all over again.


I got the charts!!

diff --git a/src/chartimg.cpp b/src/chartimg.cpp
index 824de7b..579516f 100644
--- a/src/chartimg.cpp
+++ b/src/chartimg.cpp
@@ -1064,7 +1064,7 @@ InitReturn ChartKAP::Init( const wxString& name, ChartInitFlag init_flags )
                                     bp_set = true;
                               }
 
-                              if(stru.Matches(_T("*POLYCONIC*")))
+                              if(stru.Matches(_T("*CONIC*")))
                               {
                                     m_projection = PROJECTION_POLYCONIC;
                                     bp_set = true;

They work now.


Anyway, I think it would be better to define PROJECTION_CONIC instead here. Then we should reject any kind of conic charts that do not provide sufficient reference points or polynomial coefficients.

Finally, with opengl, it is reverted so that in single chart mode, it's trying to use native coordinates.... I think it shouldn't, instead should just use mercator, because it actually is stretching the chart (pinning at the reference points) which is more accurate than simply sticking it on the right place and size because mercator is also more optimized