Hi,

It took some time but I managed to create a new feature in EUMETCastView, namely the creation of video's of Metesat-9/-10/-11.
The 'modus operandi'  is as follows :
- set the General Vertical Perspective projection parameters to the desired view ( also the canvas size !). The size of the GPV canvas establishes the canvas size of the generated video.
- go to 'Settings movie XML file'
- add the pathlist of the received Meteosat files, for ex. Y:/received/bas/E1B-GEO-1/2022/12/01
- you can change the day and/or night spectrum, the Gshhs overlay files and the overlay colors
- fill in the satellite name (Met-11, -10 or -9) depending on your pathlist
- the threadcount depends on your computer. One thread is used for calculating one image. The more cores on your computer, the quicker the video will be generated. The best practice is to set the threadcount to the number of cores of your computer.
- for HRV images check the HRV checkbox 
- to make a single test image enter the exact 'coordinates' of that image in the form "YYYYMMDDHHmm". Always create a test image ...
- to make a video clear the 'single image' field and push the "Create video or single image' button.

The creation of the images is a different process (EUMETCastVideo) , so you can close EUMETCastView without disturbing the video creation process.
You can find the latest release at https://github.com/hvanruys/EUMETCastView/releases .
Examples of video's created : https://www.youtube.com/channel/UCXt6CBoa9X_u2s3eRsVEUhQ .
Examples of images from EUMETCastView : https://flickr.com/photos/137270544@N02/ .

Grts,

Hugo


The test images will be written to the "tempimages" directory and the images used for creating a video will be written to the "tempvideo" directory. It is your responsibility to clean up this directory.

On 02/12/2022 10:18, Hugo wrote:

Hi,
It took some time but I managed to create a new feature in EUMETCastView, namely the creation of video's of Metesat-9/-10/-11.
The 'modus operandi'  is as follows :
- set the General Vertical Perspective projection parameters to the desired view ( also the canvas size !). The size of the GPV canvas establishes the canvas size of the generated video.
- go to 'Settings movie XML file'
- add the pathlist of the received Meteosat files, for ex. Y:/received/bas/E1B-GEO-1/2022/12/01
- you can change the day and/or night spectrum, the Gshhs overlay files and the overlay colors
- fill in the satellite name (Met-11, -10 or -9) depending on your pathlist
- the threadcount depends on your computer. One thread is used for calculating one image. The more cores on your computer, the quicker the video will be generated. The best practice is to set the threadcount to the number of cores of your computer.
- for HRV images check the HRV checkbox
- to make a single test image enter the exact 'coordinates' of that image in the form "YYYYMMDDHHmm". Always create a test image ...
- to make a video clear the 'single image' field and push the "Create video or single image' button.
The creation of the images is a different process (EUMETCastVideo) , so you can close EUMETCastView without disturbing the video creation process.
You can find the latest release at https://github.com/hvanruys/EUMETCastView/releases <https://github.com/hvanruys/EUMETCastView/releases> .
Examples of video's created : https://www.youtube.com/channel/UCXt6CBoa9X_u2s3eRsVEUhQ <https://www.youtube.com/channel/UCXt6CBoa9X_u2s3eRsVEUhQ> .
Examples of images from EUMETCastView : https://flickr.com/photos/137270544@N02/ <https://flickr.com/photos/137270544@N02/> .
Grts,
Hugo
The test images will be written to the "tempimages" directory and the images used for creating a video will be written to the "tempvideo" directory. It is your responsibility to clean up this directory.

==================================================================================

Excellent. Thanks Hugo. Looks like a fun weekend ahead.
I can see other good updates as well, including inclusion of GOES 18 etc.

Regards,
Douglas.

Hugo, I meant to ask before I mess things up if I can still paste in my old POI.ini and EUMETCastView.ini., or does these major updates require the use of new ini. files.

Thanks.
Douglas.

Hi Douglas,

No problem to copy the old  POI.ini and EUMETCastView.ini .

Regards,

Hugo

On 02/12/2022 13:15, Hugo wrote:

Hi Douglas,
No problem to copy the old  POI.ini and EUMETCastView.ini .
Regards,
Hugo

========================================================================

Thank you.

Regards,
Douglas.

On 02/12/2022 13:15, Hugo wrote:

Hi Douglas,
No problem to copy the old  POI.ini and EUMETCastView.ini .
Regards,
Hugo

================================================================================

Hugo I am having a problem running the program.
I ran it from the .bat file as instructed which brought up a Dos window with lots of things flying by. However when it closed EUMETCastView did not appear. If I try and run it from the EUMETCastView.exe I get a small 'system error' window with a selection of missing .dll files which on checking seem to be in the Bin folder.

Any suggestions Hugo.

Regards,
Douglas.

Douglas,

Is it possible that your PATH environment variable also has a path to the old directory of EUMETCastView ?
This release uses version Qt 5.15.6. 

regards,

Hugo 

On 02/12/2022 14:59, Hugo wrote:

Douglas,
Is it possible that your PATH environment variable also has a path to the old directory of EUMETCastView ?
This release uses version Qt 5.15.6.
regards,
Hugo

=====================================================================================

Hugo, not really sure what that means but the old EumetcastView program has been completely removed. After the DOS window closes nothing happens (I assume at that stage EUMETCastView should launch).
Now at a complete loss as to what else to try.

Presumably everything needed to run the program is included.

Regards,
Douglas.

Hugo, by adding a 'Pause' instruction into the batch file I have been able to copy the entire contents of the opening DOS window.
I suspect it may tell you something (beyond me). I could send it directly to you as a .txt file attachment if that would help.

Regards,
Douglas.

Hugo, solved the problem.
I had removed the old program but after the new program had been installed so I deleted the new EUMETCastView and completely re-installed it.
Now it has loaded successfully. May well have been a path thing.

Many thanks.
Douglas.

On Fri, Dec 2, 2022 at 02:18 AM, Hugo wrote:

Examples of video's created : https://www.youtube.com/channel/UCXt6CBoa9X_u2s3eRsVEUhQ 

Hugo,

your youtube videos look really impressive! What pixel size do you normally choose for frame image production?

Best regards,
Ernst

P.S. As you might remember I only have FullHD resolution monitors here :-(.

Thank you Ernst,

The original canvas size of those video's are all 1920x1080 pixels when I upload them to YouTube.
EUMETCastVideo renders every frame as a GVP projection and you can choose with HRV or not. For video's with the complete globe I don't use HRV.
For closeups (Europe) it is nicer to take the HRV images (more details). Of course it takes longer to render with HRV...

Regards,
Hugo
 

On Sat, Dec 3, 2022 at 05:09 AM, Hugo wrote:

The original canvas size of those video's are all 1920x1080 pixels when I upload them to YouTube.

Hugo,

I noted a one day MSG3 RSS movie is 29 seconds. This evaluates to a frame rate of 10. But AFAIK youtube seems to ask for  frame rates 24+.
Do you upload videos with 10 FPS and youtube makes some sort of frame interpolation? Or do you interpolate frames with your software?

My videos are made with ffmpeg and have only 6 FPS (hard coded in my script). But ffmpeg is rather mighty (and complex). I only scratched
the surface. I seems e.g. that ffmpeg can interpolate frames as well. There is yet a lot of room for users of my 4.x scripts to experiment :-).

https://blog.programster.org/ffmpeg-create-smooth-videos-with-frame-interpolation

Cheers,
Ernst

Yes, I do use the interpolation possibilities of ffmpeg. 
In main.cpp of EUMETCastVideo :
process.setArguments({"-r", "10", "-f", "image2", "-i", inputimagename, "-filter", "minterpolate='mi_mode=blend'", "-c:v", "libx264", "-pix_fmt", "yuv420p", "-y", outputvideoname});
Cheers,

Hugo

Hugo just to confirm all is working fine now including animations.
My problem turned out to be caused by importing my old'POI'. So I just re-built it from scratch and all works well.

Just one small note. The gshhs overlay file reference in the 'Preferences' should refer to version 2_3_7 (which is what the program folder is called). Once changing it to that I was able to get overlays.

Once again Hugo many thanks for all your hard work with the program.

Kind regards,
Douglas.

On Sat, Dec 3, 2022 at 06:36 AM, Hugo wrote:

Yes, I do use the interpolation possibilities of ffmpeg. 

Hugo,

I haven't downloaded your program or source code but my eyes made me suspect something like that :-). I currently rebuild 22 movies every hour on a GNU/Linux receiver with BAS / HVS-1 / HVS-2 that does produce all the frames, prepare MSLP overlays  and generates a couple of LEO pictures as well. It might not be something an average user really needs. I just wanted to know how far I can go without causing packet loss. I will certainly do some experiments with frame interpolation but that again adds more CPU usage. On my Windows 10 system  PyTROLL/Satpy scripts need up to twice the execution time. So this is probably rather something for interactive use to produce videos that can also be shown on youtube.

Cheers,
Ernst

Hugo,

I made a test with my eruption video of Mauna Loa 20221128. Unfortunately there are some images missing in the EUMETCast received
GOES18 data (you will see some jumps in both videos). I found these encoding times on my receiver Kallisto (Xeon 3.3GHz, 24GB RAM):

No   blending, 1 day = 144 (input) images,   6 FPS : encoding time 0min 17sec
With blending, 1 day = 144 (input) images, 11 FPS : encoding time 2min 32sec

Videos *.webm can be downloaded from wetransfer here:  https://we.tl/t-uOLBt2mFVz

Interestingly the size of the videos does not differ much. On the blended video
the time in the legend is harder to read because the numbers are blended as well.

Maybe I add that as a switch in my future movie scripts (Graham will use it ;-) !)

Cheers,
Ernst

Hi Ernst

You're not wrong - you know me too well !!

Regards

Graham

Ernst,

Very interesting test ... strange that the encoding is almost 10 times longer .
Is there any advantage for using webm video's ?

Cheers,

Hugo

Hugo,

webm is a truly *free* codec that plays directly in any modern browser and can easily be integrated into HTML5 pages
where you can even play different videos concurrently. It has also been developed for the WEB. Most other codecs have
been developed for CD/DVD players. There is also webp as picture format that can be used as a replacement for jpeg.
webp images can further be concatenated for animations similar to animated gifs (with similar problems of final file size).

Cheers,
Ernst