Profile

Cover photo
Frank Broniewski
33 followers|46,687 views
AboutPostsReviews

Stream

Frank Broniewski

commented on a video on YouTube.
Shared publicly  - 
 
Der Scout-NPC hilft einem afaik auch beim Geheimtüren-Finden
 ·  Translate
1
Add a comment...

Frank Broniewski

Shared publicly  - 
 
In addition to my rather wordy post yesterday (https://plus.google.com/110325835149755177811/posts/AWN2dMrNadC) I created something more visual today: Stick a ruler to your screen and hold an iPod next to your browser. Both screens show the same image ( http://c.geometrico.lu/basemaps/tiles/1.0.0/basemap-lb/mercator/16/33887/22332.png ) but now you can clearly see that the image in the desktop browser has another dimension than the one on the Ipod.
1
Add a comment...

Frank Broniewski

Shared publicly  - 
 
Before the weekend starts: #pretty_hillshade  hillshade rendering script in #Python  
1
Add a comment...

Frank Broniewski

Shared publicly  - 
 
6  beers and a potato! :-)
1
Add a comment...

Frank Broniewski

Shared publicly  - 
 
Found our old animated, interactive map for the Luxembourg #INGMarathon  from 2010. Much better than PDF. Press the play button and see how the different runner groups perform on the course ...

Of course it's based on +OpenStreetMap data and was actually much more recent than the  #VDL  map in that time.  And it had more info in it too ;-)
Marathon. Refreshing Point: Ravitaillement / Erfrischungspunkt / Refreshing point / verfrissings punt; Changing point: Point de changement / Wechselpunkt / Changing point / wisselen punt; Fastest runner group / schnellste Läufergruppe / le plus vite coureur / snelste loper groep ...
1
Add a comment...

Frank Broniewski

Shared publicly  - 
 
My village in the morning sun - yes sun!
1
Add a comment...
Have him in circles
33 people
Matthijs Melissen's profile photo
Air TV, l'autre télé's profile photo
Luc Brentener's profile photo
referencement airtv's profile photo
Paul Colbach's profile photo
Vicky Veritas's profile photo
Patrick Jackson's profile photo
Guillaume Rischard's profile photo
Gretchen Peterson's profile photo

Frank Broniewski

Shared publicly  - 
 
A small table that helps to calculate meters per pixel to a scale value. Probably trivial but maybe useful one day :-)
1
Add a comment...

Frank Broniewski

Shared publicly  - 
 
A post on the #Mapproxy mailing list (http://osgeo-org.1560.x6.nabble.com/MapProxy-Seeding-in-a-certain-scale-td5071082.html) came up concerning scales in raster maps. I think this concept is heavily broken:

TL;DR
The scale isn't the scale and a screen doesn't have dots so it doesn't have DPI. Don't do scale dependant calculations on fictional units.

This whole scale and DPI thing related to scale calculations is something that bothers me for quite some time already. It's actually a difficult topic and in most of the cases I think we're making the wrong assumptions all the time.

The DPI setting is something that appears in almost all mapping software for years already. I know Mapserver, Mapnik, Mapproxy, OpenLayers and Leaflet have settings for this or assume a certain default value. Mostly this value ranges from 72 to 91 to 96 and expresses also an evolution in screen resolution from CRT to TFT and even further to those nifty highres devices we carry all around with us - or maybe not all of us ;-).

I don't know what you are doing with your maps regarding the output format (I'm speaking of screen vs. paper), but I think it is safe to assume that at least users of Mapproxy and OpenLayers are targeting the screen and not the paper. And that's the crux of the matter - a screen doesn't have any dots. It doesn't have a DPI value at all! A screen has pixels, that's why there is a PPI unit after all [1]. So it's not OK to assume that the screen as an output format has an arbitrary DPI value at all - because it hasn't.

I know everbody needs a good scale for a map and that is how we work with maps in relation to the real world - we use the scale. But when we are using maps on the screen things start to get complicated because screens have so different attributes regarding the number of pixels and their size. This makes them so different from paper output formats and relate them to the real world units we need for scale computations. And another thing: pixels aren't always square [2], so it's not safe to assume that you have the same scale in east and north direction of your map.

Let's talk about those nice tablets and smart phones out there. Their usage is rising enormously [8] and they take the PPI value to new levels [3]:

Iphone: 326 PPI
Ipad: 264 PPI
MacBook Pro 13" Retina: 227 PPI
Kindle Fire HD 7": 216 PPI
Kinlde Fire HD 8.9": 254 PPI
Samsung Galaxy S4: 441 PPI
Samsung Nexus 10: 300 PPI

Have a look at the list and see for yourself. The always assumed 72 DPI is just plain wrong and its even getting wronger with every new generation of devices. This leads not only to problems concerning the content of the map you are serving [4] but also to problems with the scale you have in mind - because it isn't anymore the scale you think it is.

The OGC falls into the same pit, even if they think they're more clever. They define that a pixel is 0.28 mm*0.28 mm [5] which correlates to 90.7 DPI. A pixel on the iPhone 5 for example is actually 0.0779 mm * 0.0779 mm [6].

A raster map with 256px (your typical tile) has a OGC dimension of ~ 72 mm and a iPhone 5 dimension of ~ 20 mm - that's less than a third of the OGC dimension!

So when you say you deliver e.g. a map at a scale of 1:20.000 to your users you probably don't!

So what's the actual impact on your applications / maps with this wrong scale? I guess it would be OK if we would accept a certain DPI/PPI value as the default and take the scale as a ratio of real world units towards this default value. But not as the "scale" we're used to. Because then everyone is referring to the same ratio. But this isn't the case [5 again]. So what's happening is that your carefully selected and scale dependant cartography styles don't work anymore. You promise a map at a scale that you don't deliver.

So what to do then? Unless you don't want to deliver all your maps as a PDF document ;-) [7] I guess the resolutions are the way to go. Mapproxy has them, OpenLayers and Leaflet have them. Relate the pixel (your native output dimension on the scree) to ground units. Let the client do the scale calculation because he may know more about the real world dimensions of the device he's running on than your server does. But I think it would be best if we would avoid the scale at all in our maps when we output raster data - or we need to put a constraint on the map that relates another dimension to your map to make a scale valid. Because otherwise it isn't.

I'm sorry for this long post, but this is something that bothers me already for quite some time. Thanks for reading :-)

Frank


[1] https://en.wikipedia.org/wiki/Pixels_per_inch
[2] http://documentation.apple.com/en/motion/usermanual/index.html#chapter=B%26section=5%26tasks=true
[3] https://en.wikipedia.org/wiki/List_of_displays_by_pixel_density
[4] http://appleinsider.com/articles/12/03/13/applecom_upgrading_to_high_resolution_images_ahead_of_retina_ipad_launch/
[5] http://hub.qgis.org/issues/6430
[6] http://stackoverflow.com/questions/610193/calculating-pixel-size-on-an-iphone
[7] http://mapbrief.com/2013/05/14/pdf-sharing-is-not-data-sharing-a-public-service-announcement/
[8] http://www.brandwatch.com/2013/06/smartphone-and-tablet-use-continues-to-rise-the-stats/
1
Add a comment...

Frank Broniewski

Shared publicly  - 
 
I'm playing around with hillshading from #SRTM  scenes. I'm writing a small python script that uses standard geoprocessing and image tools in order to generate the hillshade. Here's a scene from N42W001 in the pyrenees overlayed with #CorineLandCover  data from that area. I'm quite happy with the results so far ...
1
Add a comment...

Frank Broniewski

Shared publicly  - 
 
Some impressions of my route to where I work 
1
Add a comment...

Frank Broniewski

Shared publicly  - 
 
Tour de Luxembourg 2013

The #skodatour "Tour de Luxembourg" passes our office this year .Therefore I looked at the official site to find out how the stage passes by. But the image they have there is a bit hard to read ... at best.
So I downloaded their PDF docs and made a map from it, well for the first stage anyways. If I find the time, I'll do the others too.

Don't let the #stupid Google+ preview fool you. The map's not about green energy 

http://www.geometrico.lu/examples/tour-de-luxembourg-2013.html
1
Add a comment...
 
Tour de Luxembourg 2013

Does anyone have a good map of this years tour? They're passing our office on the 12 june and I could'nt find any, just stupid PDF and a bad resolution image. I thought I'd send them a interactive map for future enhancement but its impossible to do without a decent route template map ....
1
Matthijs Melissen's profile photoFrank Broniewski's profile photo
8 comments
 
I guess the new dataset isn't simply just put online yet. But they used it already for the handbook. Who knows how their internal decision making works ...

When we did the map in 2010 for them we used  a custom OSM as a basemap and were more recent than the VDL online map of the marathon. So their online map being out of date is nothing new :-)
Add a comment...
People
Have him in circles
33 people
Matthijs Melissen's profile photo
Air TV, l'autre télé's profile photo
Luc Brentener's profile photo
referencement airtv's profile photo
Paul Colbach's profile photo
Vicky Veritas's profile photo
Patrick Jackson's profile photo
Guillaume Rischard's profile photo
Gretchen Peterson's profile photo
Work
Skills
GIS, Geoarchäologie, Mapserver, Python, PostGIS & Postgresql, Mapnik, Javascript, Leaflet
Story
Tagline
Mapamathics
Basic Information
Gender
Male
Looking for
Friends, Networking
Ich habe dort sehr gut gegessen und eine nette Bedienung erlebt. Die Stühle und die Einrichtung könnten etwas gemütlicher sein, aber insgesamt war es ein sehr netter Abend dort.
Atmosphere: GoodDecor: GoodService: Very Good
Public - 2 years ago
reviewed 2 years ago
1 review
Map
Map
Map