Opacity of TMS layer not applied in loading stage

[stoecker]

I tested an TMS layer with opacity of 60% and when you zoom in and out it always happens, that some tiles are displayed with 100% and only later switch to 60%. I'd assume that happens together with loading/caching.

The visual effect is a ugly.

OTOH There are also situations where it does not happen. Maybe when loading tiles from cache?

The Layer should always honor this setting and not display tiles without correct opacity.

[Don-vip]

I tried to take a look and didn't understand why this problem occurs. @Wiktor do you have any clue?

[stoecker]

I'd really appreciate if someone could have a look. That effect is very disturbing in a project where I use josm.

[stoecker]

There is also a probably related effect that for first display newly displayed tiles are a bit shifted and jump later into correct position.

[stoecker]

A better way to reproduce:

1) In trunk/src/org/openstreetmap/josm/data/cache/JCSCachedTileLoaderJob.java#L365​ add in line 365 before cacheData = createCacheEntry(raw) a try{Thread.sleep(5000);}catch(Exception e){}. This delays tile loading.
2) Now clear tile cache with rm.
3) start josm with a saved session displaying TMS and let it load.
4) move the screen, so that at least a new full tile is visible.

You'll see a new tile which has no opacity and is made of data from lower zoom levels. As soon as the 5 seconds are over it get's replaced with the correct level and opacity.

Thought I have still no idea why the lower stretched zoomlevels don't have some opacity. As far as I see they use the same drawing functions.

[stoecker]

It's a bit clearer now. The higher zoom levels have lighter colors, that's why it looks different.

I'm thinking maybe the display of rescaled images can be delayed a bit (i.e. 200ms), so that it happens less often.

[stoecker]

No. That causes dark instead of light disturbing.

[stoecker]

In 16669/josm:

fix #18193 - reduce flickering when paning map by loading a little bit more tiles than strictly necessary

[stoecker]

In 16670/josm:

fix #18193 - typo

[GerdP]

see #20207 for a possible regression