Unable to View Custom Maps with Neatline

Please help if you can see where I am going wrong..
Many hours have now been spent trying to create a Neatline exhibit with a custom map.
My system is under Omeka 2.1-rc1 and the Neatline 2.0.2. Geoserver is version 2.3.4, and my server is LAMP running Ubuntu 12.

I have a geoserver workspace named:Test1617 , with a geotiff layer named: 16-17_modified (referenced with WGS84)

The desired geotiff displays properly the following link:
http://industrial-history.org/test-geoserver.html ****THIS WORKS OUT OF MY GEOSERVER ***
(which points to my geoserver at http://industrial-history.org:8080/geoserver/wms)

My attempted Neatline exhibit is at http://industrial-history.org/archive/neatline/show/second-try

Pink map tiles appear seemingly whenever I provide a bad geoserver address, or a missing or bad layer.
I am guessing that nothing appears if geoserver address and layer is correctly specified. (Why does http://localhost:8080/geoserver/wms seem to be a bad address?)
If no WMS address or WMS layer is specified, OpenStreetMap displays.

The only documentation I can find on how to do this is the 3 part tutorial by David McClure, but this relates to Omeka version 1 and ArcMap is commercial software (all I have is Quantum GIS).

I have also tried to add image layers using WMS and my geoserver(within the exhibit), without success.Text, lines and shapes work fine.

Please comment. What am I not reading or doing correctly? Is there a setup example on this, for the most recent versions of Omeka and Neatline?
I am also very unclear as to how the custom WMS service gets mapped or associated with items (records) in Omeka/Neatline 2.x..
Chris

****

Settings case1

Exhibit setup
Title:second try
Widgets: none
Active Map Layers: none
Default Map Layer*:OpenStreetMap
Image Layer: none
WMS Address:http://localhost:8080/geoserver/wms
WMS Layers: Test1617:16-17_modified
Spatial Querying:none

Results: A pink set of tiles are displayed, which seem to behave normally, as if it were a map or image.

****

Settings case 2
Exhibit setup
Title:second try
Widgets: none
Active Map Layers: none
Default Map Layer*:OpenStreetMap
Image Layer: none
WMS Address:http://industrial-history.org:8080/geoserver/wms
WMS Layers: Test1617:16-17_modified
Spatial Querying:none

Results: A normal page is displayed, but without any tiles appearing in the map window.

****

Settings case 3
Exhibit setup
Title:second try
Widgets: none
Active Map Layers: none
Default Map Layer*:OpenStreetMap
Image Layer: none
WMS Address:none
WMS Layers: Test1617:16-17_modified
Spatial Querying:none

Results: A normal page is displayed, showing the southern coast of West Africa.

Hi Chris,

Try this, in the Geoserver admin for the layer, change the coordinate reference system for the layer from EPSG:4326 to EPSG:900913 (Layers -> Test1617) and then select "Force declared" from the SRS handling menu. After you've done that, click on the "Compute from native bounds" option under Bounding Boxes, which should update the coordinate data. Save and then refresh your Neatline page.

Let me know how it goes...

Wayne

Thanks wspraph, for the reply.
Your suggestion changes the projection a bit (as you can see at http://industrial-history.org/test-geoserver.html), but does not affect the problem when trying to include this map in Neatline. I have also tried to use some of the other default raster examples, which comes with geoserver, and still have the same problem.

I now have some records added to box in the default location (below West Africa), and a drawn set of boxes in my target area (Western Massachusetts, USA). The drawn records show over the default map if I don't define my custom base layer, but I get a completely blank map if I specify a layer and geoserver address, as I listed in Case #2, in my first post.
I know it is dangerous to presume, but I don't think this is a mapping/projection issue. I suspect there is a problem in the way I try to address my geoserver in the WMS Address field. It is frustrating how I can easily preview my layers, directly out of geoserver (which you can see in the form of the test-geoserver page), and the attempts to address the same layer through Neatline produces a blank map, which won't even display the added, drawn vector objects.
I thought of trying 'Neatline Maps', but it doesn't seem to be available for Omeka 2.x yet.

Hi Chris,

This is odd; do you have a tiff I can download to take a look at? I looked at the header from geoserver, but am just getting a white box when I push it on to my local geoserver...

We're not planning on migrating NeatlineMaps for Omeka 2. There are a lot of different issues that people will encounter moving these files around using PHP (like file size limits, double bandwith, etc.). However, we are working on a parallelized system that uploads geotiffs to Geoserver (as well as automate removing black no-data fields as part of our GIS workflow). It's not quite ready for prime time yet, but you can take a look at the source at https://github.com/waynegraham/loader.

Wayne

Thanks for staying with this!
The tiff is now also copied to:
http://industrial-history.org/tmp/16-17_modified.tif
You are welcome to offer another geotiff I might use. I would setup a new workspace/store/layer for it and try it. I must repeat that I tried to use other raster images, which installed with geoserver.
My intention for all of this is to use geoserver to display vintage maps, and even large charts and drawings, for our museum. Exact and accurate geolocation is not essential but I should learn how to do it properly.

Maybe I am grasping at straws, but my geoserver status page lists my data directory as:
/usr/share/tomcat7/webapps/geoserver/data
but most of the stuff installed in:
/usr/share/tomcat7/webapps/geoserver/data/data
This particular image we are playing with is in:
/home/geotiff
It didn't seem to be a problem in creating and setting up workspaces and layers with this data.

This is weird, I just looked at the 16-17_modified tiff with gdalinfo and there's no coordinate system (Coordinate System is `'). I ran these commands on the file:


$ gdalwarp -srcnodata 0 -dstalpha 16-17_modified.tif 16-17_modified_warped.tif
$ gdal_translate -of GTiff -a_srs EPSG:4326 16-17_modified_warped.tif 16-17_modified_translated.tif

This takes the black edges off and put a header on it. It's a pretty small map, so I'm not sure if it's showing up properly in Neatline since I'm not exactly sure where to look.

Could I get a link to the original by any chance? There may be something different in the way the QGIS and ArcGIS do things...

Wayne

This is weird, I just looked at the 16-17_modified tiff with gdalinfo and there's no coordinate system (Coordinate System is `'). I ran these commands on the file:


$ gdalwarp -srcnodata 0 -dstalpha 16-17_modified.tif 16-17_modified_warped.tif
$ gdal_translate -of GTiff -a_srs EPSG:4326 16-17_modified_warped.tif 16-17_modified_translated.tif

This takes the black edges off and put a header on it. It's a pretty small map, so I'm not sure if it's showing up properly in Neatline since I'm not exactly sure where to look.

Could I get a link to the original by any chance? There may be something different in the way the QGIS and ArcGIS do things...

Wayne

Solved! This all has to do with my custom map! My error was in not setting up the coordinate system properly. I guess when the link to the WMS service on my server is correct, nothing appears because the geoserver basically has nothing to serve(or, maybe it is very small and not where it was expected in the view space). The pink color tiles do seem to be an indication of a bad address to the geoserver, or bad layer name.
Question: Do I always need to a point to the geoserver as a "http://domain-name.xxx/geoserver/wms", or is there a local reference (localhost?) reference of some sort which doesn't require using the web domain name?

I need to spend more time understanding how to use Quantum GIS. I had used it's georeferencer to tack points matching a google map (Lat/Lon), but although it seemed to work, the resulting geotiff was still the cause of my problem.

A proof of the problem came when I got rid of the custom base layer (in the Neatline exhibit setup) and tried adding a map layer in the exhibit layer editor. The nurc:Img_Sample layer of North America (from the demo layers installed with geoserver)overlayed perfectly on top of the OpenStreetMap base layer (set up in the initial exhibit definition).

I hope this post helps other beginners understand this, and not lose the time I did, before you began to help. This problem would be obvious to an experienced user, but you saved me from much more time and frustration!

The custom map is my own scan from an 1884 atlas of Greenfield, Massachusetts. It was 6 pages from a bound atlas, and I have scanned and assembled the entire town into a single image (about 1GB, even with compression). The preview layer can be seen here:http://industrial-history.org/greenfield-1884.html
The original image scans are palette limited tiffs (LZH compression). They are impractically large for web FTP, so the "original" isn't easy to offer. In any case, the problem has been explained! All I need now to do is learn how to do this all correctly, with the tools I have available.

Many of us poor folk rely heavily on Open Source tools, such as Gimp, Blender, etc. It was frustrating to find the only tutorial using ArcGIS, as this tool needs to be purchased.

I am sure Quantum GIS will do this correctly. I just need to learn how, and verify the work. Are there other Open Source tools you might suggest?
Thanks for the help!
Chris

Hi Chris,

The way the system works is that the calls are made from the client machine (the web browser) to the Geoserver instance. In this case, both localhost and 127.0.01 would point at the user's computer and not the intended remote server.

I had started to write up how to do the georeferencimg with QGIS a few months ago until I realized that that plugin was missing from the OS X package I had. Any chance you'd be willing to work up a tutorial we could add to the site to help others? I'd love to be able to put together a full open source workflow stack for working with Neatline.

Wayne

I might like that, and there is more to discuss.
Please email "Chris" at the museum's email address at info@industrialhistory.org, so we can take the conversation off of this thread.