Maps.me rendering engine

[Alexander]

Maps.me was open-sourced recently: https://github.com/mapsme/omim

Their rendering engine vastly superior to osmand one, much faster and overall rendering quality, colors also good

Any plans to adapt or use some parts of MAPS.ME rendering engine?

[Yevheniy Pyatnychko]

Rendering on maps.me is really great! ...Osmand would be really great, if maps were faster

[Aceman444]

I for example do not like the rendering in MAPS.ME. It has less details, has less colors and appears to me it is getting slower with each release.
Maybe OsmAnd can be as fast if anybody creates a rendering style with these less features (maybe like the LightRS style).

Dňa streda, 30. septembra 2015 16:44:16 UTC+2 Alexander napísal(-a):

[Alexander]

The style is ok i think (e.g. http://i.imgur.com/IsJKkry.png) also, i'm talking about rendering engine mostly, colors and detail level could be adjusted

[Poutnik for NNTP]

The point is,
comparable map source has to be used
( as processing speed may differ for different map formats )
and comparable rendered details have to be used,
to compare rendering engines.

[xmd5a]

Complexity of Rendering style in OsmAnd does not affect rendering speed too much (+-10%)

среда, 30 сентября 2015 г., 17:44:16 UTC+3 пользователь Alexander написал:

[V S]

OsmAnd indeed uses different rendering technique than Maps.me engine. OsmAnd uses CPU rendering, maps.me GPU. So some features may be missing and that could be critical. So in theory it is hard to compare speed of engines if they are using different acceleration technique, what it looks like GPU is winning trend, on the other hand if OsmAnd starts using multi-core CPU, the situation can change. 

Best Regards,

Victor

[Alexander]

Thanks for details. Well, on my 4-core phone difference in speed is huge. Both are usable but maps.me wins by large margin. Simple low-detail rendering style doesn't change much.

[Victor Shcherb]

Right now OsmAnd uses only 1 core and we are thinking how to change that, because it will give some boost definitely.

--
You received this message because you are subscribed to a topic in the Google Groups "Osmand" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/osmand/7MAbPRjoNmk/unsubscribe.
To unsubscribe from this group and all its topics, send an email to osmand+un...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Mirco Zorzo]

I agree today only few device are at least dual core the more are quad core or better. This will ne a huge improvement.

Mirco

[Aceman444]

Is the OsmAnd OpenGL renderer (that is in the works) a dead end?

Dňa sobota, 3. októbra 2015 17:35:54 UTC+2 V S napísal(-a):

[Harry van der Wolf]

2015-10-04 13:37 GMT+02:00 Aceman444 <acesha...@gmail.com>:

Is the OsmAnd OpenGL renderer (that is in the works) a dead end?

I don't think so as it is the renderer engine on IOS. I assume you want to go to one standard on multiple platforms.

 

[Harry van der Wolf]

Sorry, wrong answer.

You put me on the wrong track and I didn't think any further.

The renderer in the works is the Qt renderer, the one also being used on the iPhone.

Both the Qt and the current renderer implementation use OpenGL to address the hardware.

Harry

[Victor Shcherb]

Yes, Qt-renderer uses different technique working with textures (generating texture to upload a tile) and it uses vector layer of symbols. Mainly it was developed and used for iOS. Technically it still renders each tile on CPU (except the symbols), it has own advantages and disadvantages but it still very different than rendering polygons directly on GPU.

We don't observe too much improvement of Qt-rendering, but we see quite a lot work to polish it for all devices. 

Best Regards,

Victor

[Alexander]

Any insights how maps.me rendering is different and why it so fast? As far as i understand it's because of offloading most of work to GPU. Adopting similar scheme could benefit the project a lot

[JeCh]

Osmand is now almost perfect, but for me the rendering is one of 3 weakest points. The other 2 are route calculation (mainly it's speed) and lack of online services (for example collecting data from drivers and using them for routing).

Regarding routing, it is not just the speed which is bad also the lack of any 3D visualisation. For example look how good maps look in this navigation: https://play.google.com/store/apps/details?id=com.generalmagic.magicearth or buildings rendering here: http://osmbuildings.org/

Switching to Qt is the right choice IMO. They even have ready to use map solution (at least as an inspiration). Some sources for information are here: http://blog.qt.io/blog/2017/03/09/provisioning-openstreetmap-providers-in-qtlocation/ and http://blog.qt.io/blog/2016/10/04/customizable-vector-maps-with-the-mapbox-qt-sdk/

Another advantage I can see is the fact, that the developers won't have to care which 3D rendering engine must be used. Qt will automatically pick OpenGL, Vulkan or whatever else is available.

[nickjoh...@gmail.com]

The developers are working on an OpenGL rendering engine which will support 3D/perspective mode for navigation. It's taking a long time but it's a lot of work. IIRC there are some nightly builds which use the new rendering engine here:
http://download.osmand.net/latest-night-build/

As for routing speed, whenever I test Maps.me, it is much slower for walking and cycling routes, so I wouldn't be showing Maps.me as a great example of fast routing.

Nick

[JeCh]

I think nobody wanted to give MAPS.ME as a good example of routing. It was given as an example of fast rendering.

I heard about the work on OpenGL renderer maybe more then 4 years ago. But since then Qt announced their own accelerated map renderer with OpenStreetMap support. I'm not sure if it has all the features Osmand needs. But I just thought to point it out so the developers won't have to reinvent the wheel. I know that they are probably aware of it, but just in case. :-)

Vladimir

[Jan van Bekkum]

What about MapFactor Navigator? I am always impressed with the speed with which it calculates 3 alternative routes, also for long distances. I use OsmAnd for my daily navigation, but I use Navigator for planning.

[Poutnik]

I suppose MFN uses some variant of contracted hierarchy algorithm, which  is much faster then usually used variants of A-star algorithms. It is used by many online routing services and relies on very heavy routing data pre-processing. Also its initial data are simplified. But it about routing, not rendering.

Dne 21/06/2017 v 06:46 Jan van Bekkum napsal(a):

-- 
Poutnik ( The Wanderer )

My Brouter profiles 
https://github.com/poutnikl/Brouter-profiles/wiki