NOAA ENC - Panning in 5.6.2 vs 5.2.4 w/quilting

[flgulfstar]

Hello,

I've noticed a substantial difference between OpenCPN 5.2.4 and 5.6.2 (tried on both Win10 and Ubuntu 22.04) in chart panning performance using NOAA ENC vector charts for my home region when quilting is enabled .

On 5.6.2 , chart panning with quilting enabled is often "choppy" near the chart boundaries - sometimes a click and pan takes a second to register, and sometimes the viewport shifts again or snaps back. On 5.2.4 panning was always completely smooth and fluid across the same area and charts with quilting enabled. RNC charts on either version are not affected.

I've experimented with different settings: OpenGL on/off, Smooth Panning / Zooming checked or unchecked, chart detail level reduced. I've tried removing and re-downloading updated charts, Force Full Database Rebuild, etc. I also tried reducing the number of charts in my database to just a few adjacent charts in my local region. Finally, I am testing this on a PC I use for route planning which has no GPS source configured, although I tried manually moving My Boat in and out of the current chart region.

Unfortunately the only fix I've found so far is either to disable quilting in 5.6.2, or to downgrade to 5.2.4. Has anyone else observed this issue?

Thank you!

[bdbcat]

flgulfstar...
I know it may be inconvenient, but a video of the effect would sure be helpful, especially with comparisons to previous versions.
On Ubuntu, I use "SimpleScreenRecorder"



Thanks
Dave

[flgulfstar]

Dave,

Thank you for the quick response. Please find two short videos I recorded using OBS on my Windows PC, temporarily linked below (I don't think it's possible to upload them directly here?).

https://filebin.net/n4tzote1e0evp7qx

Both show panning around the same area (with chart quilting turned on) starting on chart US5FL53M (ST JOHNS RIVER JACKSONVILLE TO RACY POINT).

In the video 562.mkv, you will see a few movements early where the panning motion moves in a stuttering step, followed by some smooth panning, followed by some fairly long pauses around 0:20. The longest pause is about a second from the time of clicking, dragging, and releasing the mouse until the screen pans. Also, in 5.6.2 it often goes in two steps, a small move quickly, and then a large move after a noticeable delay.

In the second video 524.mkv you can see a series of steps panning all over the same area and the click/drag/release operation is registered almost instantly, with the screen updating very fluidly.

I hope this captures the scenario. I am glad to make additional videos if it is not apparent (a bit hard to capture the feeling of the latency). This issue is by no means a showstopper, just a pronounced difference I noticed from 5.2.4 that makes panning across large areas a little more difficult, especially when reviewing and drawing routes.

Thanks!

[flgulfstar]

Just an update: The period of choppy panning seems to coincide with when the chart US1GC09M.000 (GULF OF MEXICO) is also in scope (second or fourth highlighted green bar in the chart bar).

[flgulfstar]

If I select Hide This Chart from the chart bar for the larger-scale charts, panning performance returns to normal for the local chart that is currently active.

It seems something has changed since the earlier version which affects these "area" charts -- I am not sure how they are officially designated in the NOAA catalog but for example in my case the GULF OF MEXICO or the KEY WEST TO MISSISSIPPI RIVER charts, which I guess act as generic containers for an entire region which are then superseded by a local, detailed chart?

[bdbcat]

flgulfstar...
Thanks for the videos.
Will you try a test for me?
Please try the performance of O524 vs O562 with Depth Soundings turned OFF. Hamburger->Show Depths UNCHECKED.
Please use exactly the same scenario/charts sets as shown in the videos.
Thanks
Dave

[flgulfstar]

That works! If I turn off Vector Charts > Show Depths, the panning performance on 5.6.2 is very smooth and equivalent to 5.2.4. I used the same charts and panning scenarios, and I did not hide the larger-scale charts in this test. Performance seems consistently good regardless of other detail settings or zoom level.

When I watch the CPU usage of opencpn via top, with Depths off it rarely goes over 20% while panning, whereas with Depths selected it frequently spikes to 90-100% when moving across the areas with the larger-scale charts in scope.

[bdbcat]

flgulfstar...


Thanks for the confirmation.
So this is a regression bug on O562. I will investigate further.


Thanks
Dave

[flgulfstar]

Dave,

I appreciate the quick response. And thank you to you and the entire OpenCPN team for this amazing software, your commitment to open source, and your support for multiple platforms.

I still keep my Garmin plotter around on the flybridge because it can get wet and is readable in daylight, but we use OpenCPN for everything else. Our lower helm is a Raspberry Pi with an inexpensive USB-powered travel monitor, and we routinely go back and forth Windows/Linux/Mac for trip planning at home. Couldn't do it without OCPN!