|
26-03-2017, 21:16
|
#1
|
Registered User
Join Date: Aug 2009
Location: oriental
Boat: crowther trimaran 33
Posts: 4,414
|
improved grib overlays
This is a really annoying feature to test because there are many possible grib files.
In any case, I have improved the plugin to hopefully support overlays in all possible cases (with opengl) as well as some other improvements. Please test with lots of different gribs.
Also, we need some kind of progress bar when loading very large grib files it takes a while.
https://github.com/OpenCPN/OpenCPN/pull/818
|
|
|
27-03-2017, 16:23
|
#2
|
Registered User
Join Date: Jun 2015
Posts: 379
|
Re: improved grib overlays
Quote:
Originally Posted by boat_alexandra
This is a really annoying feature to test because there are many possible grib files.
In any case, I have improved the plugin to hopefully support overlays in all possible cases (with opengl) as well as some other improvements. Please test with lots of different gribs.
Also, we need some kind of progress bar when loading very large grib files it takes a while.
https://github.com/OpenCPN/OpenCPN/pull/818
|
Hi, there's gribs with more than 1024 I or J points, without downsampling there's no overlay, it's a regression.
Big files have other issues with memory usage, there's no obvious way for closing a grib, you have to open a small one.
BTW weather_routing should use some kind of refcounting rather than cloning every grib again and again (in batch mode).
|
|
|
27-03-2017, 16:48
|
#3
|
Registered User
Join Date: Aug 2009
Location: between the devil and the deep blue sea
Boat: a sailing boat
Posts: 20,420
|
Re: improved grib overlays
How difficult would it be to develop the feature so that optionally (toggle on-off) only the arrows over the selected route are displayed and then so that the arrow from the specific time-point for the route is dispalyed?
I have seen it in another software. I think it is a HUGE feature and sadly missing in packages like OpenCPN, qtVlm, etc.
What do you say?
+hugs&best regards,
barnakiel,
|
|
|
30-03-2017, 14:08
|
#4
|
Registered User
Join Date: Aug 2009
Location: oriental
Boat: crowther trimaran 33
Posts: 4,414
|
Re: improved grib overlays
Quote:
Originally Posted by did-g
Hi, there's gribs with more than 1024 I or J points, without downsampling there's no overlay, it's a regression.
|
I can add down sampling, but these large files take longer to load already?
Quote:
Big files have other issues with memory usage, there's no obvious way for closing a grib, you have to open a small one.
BTW weather_routing should use some kind of refcounting rather than cloning every grib again and again (in batch mode).
|
Weather routing has to interpolate on a lot of time slices though, so I don't know if it would help.
|
|
|
30-03-2017, 14:10
|
#5
|
Registered User
Join Date: Aug 2009
Location: oriental
Boat: crowther trimaran 33
Posts: 4,414
|
Re: improved grib overlays
Quote:
Originally Posted by barnakiel
How difficult would it be to develop the feature so that optionally (toggle on-off) only the arrows over the selected route are displayed and then so that the arrow from the specific time-point for the route is dispalyed?
|
This exists in weather routing for computed routes, but you suggest it for manual routes as well? I think it wouldn't be difficult, but you also have to know what speed you are going.
|
|
|
31-03-2017, 11:00
|
#6
|
Registered User
Join Date: Jun 2015
Posts: 379
|
Re: improved grib overlays
Quote:
Originally Posted by boat_alexandra
I can add down sampling, but these large files take longer to load already?
|
IMO it's user call. Anyway some files may not be that big, a lot of oceanic stuff don't use regular grids but they are shoehorned in one with a lot of undef value.
Quote:
Originally Posted by boat_alexandra
Weather routing has to interpolate on a lot of time slices though, so I don't know if it would help.
|
IIRC in batch mode for each run it's requesting and cloning every grib records even if they overlapped a previous run.
|
|
|
31-03-2017, 17:18
|
#7
|
Registered User
Join Date: Aug 2009
Location: oriental
Boat: crowther trimaran 33
Posts: 4,414
|
Re: improved grib overlays
Quote:
Originally Posted by did-g
IMO it's user call. Anyway some files may not be that big, a lot of oceanic stuff don't use regular grids but they are shoehorned in one with a lot of undef value.
|
I think now the actually texture upload is fast enough now (20 milliseconds for 512x512) and the real bottleneck is someplace else. Maybe I should just knock it up to support a bit larger textures for now?
Do you have a source of such large gridded gribs I can experiment with? Downsampling is really a shame.
Quote:
IIRC in batch mode for each run it's requesting and cloning every grib records even if they overlapped a previous run.
|
Yes, this is for sure, but I wonder if there would be a significant speed up? Is it worth it? The grib stuff is usually performed in a different thread, so it shouldn't slow things down on multicore systems even if it is doing extra work.
Weather routing algorithm could be rewritten to be much faster (10x faster in most cases) and I think this is a more important effort...
|
|
|
31-03-2017, 18:45
|
#8
|
Registered User
Join Date: Jun 2015
Posts: 379
|
Re: improved grib overlays
Quote:
Originally Posted by boat_alexandra
I think now the actually texture upload is fast enough now (20 milliseconds for 512x512) and the real bottleneck is someplace else. Maybe I should just knock it up to support a bit larger textures for now?
Do you have a source of such large gridded gribs I can experiment with? Downsampling is really a shame.
|
discontinued but hires files have 2500x2500 points
ftp://ftpprd.ncep.noaa.gov/pub/data/nccf/com/ofs/prod/
or a 2800x1800 points
http://195.154.231.142/AROME/france.grb2
it's a 600MB file
Quote:
Originally Posted by boat_alexandra
Yes, this is for sure, but I wonder if there would be a significant speed up? Is it worth it? The grib stuff is usually performed in a different thread, so it shouldn't slow things down on multicore systems even if it is doing extra work.
Weather routing algorithm could be rewritten to be much faster (10x faster in most cases) and I think this is a more important effort...
|
The issue is not speed but memory.
- On a 32 bits system you can reach the 2 or 3 GB limit and kill OCPN.
- If you reach OCPN cache memory limit (default 1GB) it's game over, OCPN can't free cache memory and it closes and re opens charts all the time .
|
|
|
01-04-2017, 13:56
|
#9
|
Registered User
Join Date: Jun 2015
Posts: 379
|
Re: improved grib overlays
Hi,
a bug
loading is ok but after paning around I have :
|
|
|
02-04-2017, 05:16
|
#10
|
Registered User
Join Date: Aug 2009
Location: between the devil and the deep blue sea
Boat: a sailing boat
Posts: 20,420
|
Re: improved grib overlays
Quote:
Originally Posted by boat_alexandra
This exists in weather routing for computed routes, but you suggest it for manual routes as well? I think it wouldn't be difficult, but you also have to know what speed you are going.
|
I will try to grab a screen or make a draft and repost here.
I have seen it during recent VG - the excess wind arrows are deleted from the screen so that only the arrows under the planned / calcutted route remain. HOWEVER - these remaining arrows are drafted for the time point of the grib file that represents the time when the boat will be in this specific point. So all one can see is the track and what the wind is doing over the track when we get there.
As you said, speed must be either assumed or else picked up from the weather routing calculation.
I will be back with some kind of graphical input.
barnakiel
|
|
|
02-04-2017, 12:01
|
#11
|
Registered User
Join Date: Aug 2009
Location: between the devil and the deep blue sea
Boat: a sailing boat
Posts: 20,420
|
Re: improved grib overlays
Linking to two images from two different solutions.
I think something like image 1 would be great but add the toss option to remove the excess arrows as seen in image 2.
Image 1:
http://www.squid-sailing.com/img/cms/squid-routage.png
Image2:
http://www.islersailing.com/nick_routing_sample.JPG
So that under the track what one can see the arrows as the wind would be when we are in any specific route point/time.
What do you think?
barnakiel
|
|
|
|
Thread Tools |
Search this Thread |
|
|
Display Modes |
Rate This Thread |
Linear Mode
|
|
Posting Rules
|
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts
HTML code is Off
|
|
|
|
Advertise Here
Recent Discussions |
|
|
|
|
|
|
|
|
|
|
|
|
Vendor Spotlight |
|
|
|