Very slow to scroll geolocalized images

[StephaneP]

What steps will reproduce the problem?

1. Load an image sequence
2. Keep pressing the next image shortcut
3. Compare scrolling speed between Josm latest and tested

What is the expected result?

Scrolling speed should be the same speed for both releases

What happens instead?

With the latest release, it's very slow. Perhaps it's related to #17072 or #20659 (no gamma/sharpness applied during the test)
See the attached video: At first I was using josm tested, then Josm latest.

Please provide any additional information below. Attach a screenshot if possible.

URL:https://josm.openstreetmap.de/svn/trunk
Repository:UUID: 0c6e7542-c601-0410-84e7-c038aed88b3b
Last:Changed Date: 2021-04-23 21:32:22 +0200 (Fri, 23 Apr 2021)
Build-Date:2021-04-24 01:30:58
Revision:17814
Relative:URL: ^/trunk

Identification: JOSM/1.5 (17814 en) Windows 10 64-Bit
OS Build number: Windows 10 Pro 2009 (19042)
Memory Usage: 5566 MB / 7218 MB (2602 MB allocated, but free)
Java version: 1.8.0_191-b12, Oracle Corporation, Java HotSpot(TM) 64-Bit Server VM
Look and Feel: com.sun.java.swing.plaf.windows.WindowsLookAndFeel
Screen: \Display0 1920×1080 (scaling 1.00×1.00)
Maximum Screen Size: 1920×1080
Best cursor sizes: 16×16→32×32, 32×32→32×32
System property file.encoding: Cp1252
System property sun.jnu.encoding: Cp1252

Plugins:
+ Mapillary (1.5.37.6)
+ OpeningHoursEditor (35640)
+ PicLayer (2a9aa7a)
+ apache-commons (35524)
+ apache-http (35589)
+ areaselector (368)
+ austriaaddresshelper (1597341117)
+ cadastre-fr (35727)
+ continuosDownload (91)
+ contourmerge (v0.1.6)
+ ejml (35458)
+ geotools (35458)
+ javafx-windows (35655)
+ jaxb (35543)
+ jna (35662)
+ jts (35458)
+ log4j (35458)
+ opendata (35640)
+ photo_geotagging (35715)
+ photoadjust (35640)
+ reverter (35732)
+ tageditor (35640)
+ todo (30306)
+ turnlanes-tagging (288)
+ turnrestrictions (35640)
+ undelete (35640)
+ utilsplugin2 (35691)

Tagging presets:
+ %UserProfile%\Documents\Gitlab\preset-josm-collectivites\Presets-collectivités.xml

Map paint styles:
- https://josm.openstreetmap.de/josmfile?page=Styles/Enhanced_Lane_and_Road_Attributes&zip=1
- https://josm.openstreetmap.de/josmfile?page=Styles/Lane_and_Road_Attributes&zip=1
- https://raw.githubusercontent.com/species/josm-preset-traffic_sign_direction/master/direction.mapcss
- %UserProfile%\Documents\GitHub\MapCSS-JOSM-Bicycle\cycleway.mapcss

Last errors/warnings:
- 00016.563 E: Failed to locate image 'object--wire-group'
- 00016.672 E: Failed to locate image 'regulatory--dual-lanes-cyclists-and-pedestrians--g1'
- 00017.107 E: Failed to locate image 'regulatory--texts--g1'
- 00017.108 E: Failed to locate image 'regulatory--texts--g2'
- 00017.217 E: Failed to locate image 'void--car-mount'
- 00017.218 E: Failed to locate image 'void--dynamic'
- 00017.219 E: Failed to locate image 'void--ego-vehicle'
- 00017.220 E: Failed to locate image 'void--ground'
- 00017.221 E: Failed to locate image 'void--static'
- 00017.371 E: Failed to locate image 'warning--kangaroo-crossing--g1'

[StephaneP]

If I insist to scroll the images, Josm-latest start to freeze, and it ends up with a crash.

[skyper]

Replying to StephaneP:

Memory Usage: 5566 MB / 7218 MB (2602 MB allocated, but free)

Looks like lots of memory is used or at least already allocated. Doe you have enough physical memory on the machine to safely allow JOSM to use 7GB

Replying to StephaneP:

If I insist to scroll the images, Josm-latest start to freeze, and it ends up with a crash.

Might be a "Out of memory" problem or using too much virtual memory. A thread dump might help to tell more about the problem.

[StephaneP]

Yes, I have 32GB on this machine.
With latest, I see these messages (josm launched with -Xmx8g) :
2021-04-25 15:04:48.282 SEVERE: Handled by bug report queue: java.lang.OutOfMemoryError: Java heap space java.lang.OutOfMemoryError: Java heap space
and
2021-04-25 15:04:48.354 WARNING: Too many errors. Dropping ReportedException [thread=Thread[org.openstreetmap.josm.gui.layer.geoimage.ImageDisplay$LoadImageRunnable,6,main], exception=java.lang.OutOfMemoryError: Java heap space, methodWarningFrom=BugReportExceptionHandler#handleException]

With Josm-tested I can reproduce the same problem, but I need to scroll thru much more images before Josm complains.

With an older release (17056), I can't reproduce the problem, memory used grow much slower, and from time to time it decrease. This release don't crash, and the "scrolling" is smoother than tested or latest.

BTW : The Josm image viewer is still very very slow when you compare it with ​FEH or ​Jpegview

[simon04]

Image loading has changed via r17548 (from Toolkit.getDefaultToolkit().createImage(...) to ImageIO.read(...)). Can you narrow down the regression to a specific commit?

[StephaneP]

Replying to simon04:

Image loading has changed via r17548 (from Toolkit.getDefaultToolkit().createImage(...) to ImageIO.read(...)). Can you narrow down the regression to a specific commit?

I don't have the Java skills to do this. But:
r17544 -> smooth, no crash
r17548 -> scrolling the images is slow, memory usage raise up to 8G, but i failed to crash Josm
r17678 -> same as previous
r17686 -> same as previous
r17698 -> same as previous
r17699 -> same as previous
r17701 -> same as previous, OOM (sometime). I'm not sure memory usage is really different than r17548
r17779 -> same as previous, OOM (sometime). I'm not sure memory usage is really different than r17548
r17788 -> scrolling is very very slow, OOM in less than one minute.

[StephaneP]

Hi!

I just tested r17878.
Scrolling thru the images is smooth like r17544, perhaps even smoother. Thanks for that!

But a problem remains: If I scroll thru 500 images (around 2MB each) I have to wait for about 2mn20s to see the last image.

[StephaneP]

If I "scroll" thru a lot of images, and remove the image layer, Josm still try to load all the images. In the meantime, I can't download osm data and can't do a lot of other things.

In the attached video, I've removed the image layer 10 minutes ago.

[Don-vip]

In 18150/josm:

fix #20799 - fix performance/memory issues while scrolling across many geolocated images

[StephaneP]

I confirm it's ok now! Thanks!!