These are chat archives for openseadragon/openseadragon

26th
Aug 2016
John Susi
@johnsusi
Aug 26 2016 10:56
I have a question about tile loading concurrency: When i use addTiledImage the tiles are loaded one at a time. Would it be possible to improve on that making them all load at once or is there someway to query openseadragon about what tiles are visible (but not loaded)
Ian Gilman
@iangilman
Aug 26 2016 16:21
@johnsusi Check out openseadragon/openseadragon#1013 ...sounds like what you're talking about
Ian Gilman
@iangilman
Aug 26 2016 17:01
@johnsusi The fix has been merged... can you try the latest OSD master to see if it fixes your issue as well?
John Susi
@johnsusi
Aug 26 2016 17:47
wow that is so weird. I was searching the issues for something similar that i knew i had seen before but couldnt find it. I'll give it a try
@iangilman much better I'd say but also a bit different. Before OSD only loaded two tiles for me (level 11) on the initial load, now it loads 11, (from different levels) making it look progressive. I assume that how it is intended to work
Ian Gilman
@iangilman
Aug 26 2016 17:59
@johnsusi Yes, it's supposed to be progressive, but it's odd that that fix would have such a dramatic change!
What version were you using before?
John Susi
@johnsusi
Aug 26 2016 18:00
master, but i had the same behaviour with the latest release 2.2.1
Ian Gilman
@iangilman
Aug 26 2016 18:00
Also, ideally it should load the highest level that is covered by a single tile first and then move up from there. If two tiles is enough to cover level 11 in your scenario, then it should load level 10 first and then the two tiles for level 11
John Susi
@johnsusi
Aug 26 2016 18:00
let me play around a bit with this just to confirm its not something from my code
Ian Gilman
@iangilman
Aug 26 2016 18:00
Do you have any unusual options?
Cool, sounds good... if you can share a repro, that could be good as well
Ian Gilman
@iangilman
Aug 26 2016 18:06
Well, I'm getting something a little similar... before the change I'm getting 18 tiles and afterwards I'm getting 25 (with the image from http://openseadragon.github.io/ )
I think it's a matter of the fix uncovering a flaw in our logic, not that the fix itself is bad
John Susi
@johnsusi
Aug 26 2016 18:12
I see that we have immediateRender = true at least,
Ian Gilman
@iangilman
Aug 26 2016 18:13
That would explain why you're only loading level 11 before.
I'm writing up an issue for it... would you be up for investigating?
John Susi
@johnsusi
Aug 26 2016 18:14
sure, I'll dig into it
VoidVolker
@VoidVolker
Aug 26 2016 18:19
sadly, my wrapper want to drop :(
@iangilman did you have ths #1015 bug?
I can't reproduce
Ian Gilman
@iangilman
Aug 26 2016 18:23
@johnsusi Excellent! I've finished the ticket and tagged you in it
@VoidVolker Sure enough I can reproduce it with http://localhost:8000/test/demo/iiif.html
@VoidVolker Maybe you can figure out how to fix it without breaking your original patch?
VoidVolker
@VoidVolker
Aug 26 2016 18:29
also, I fixed similar bug
ok, I'll check it later
Ian Gilman
@iangilman
Aug 26 2016 18:29
Cool, thank you! I figure you know that area well
John Susi
@johnsusi
Aug 26 2016 18:35
@iangilman Actually it might be working as intended. Before the patch it loaded level 11 (two tiles), but looking at it now it appears to be loading level 10 before and then 11, which gives it a "progressive" look but level 10 is actually the first level that covers the entire area in one tile
Ian Gilman
@iangilman
Aug 26 2016 18:37
@johnsusi Yes, but it shouldn't be doing that with immediateRender on
...and now it looks like it's also going all the way back down the chain and loading levels 9, 8, 7, etc.
...which it definitely shouldn't be doing regardless of immediateRender
Did you read my theory in the issue? Does it make sense?
(fwiw, I prefer the non-immediateRender look)
John Susi
@johnsusi
Aug 26 2016 18:42
yes you are right, after loading the current level it goes all the way down (up?) to 0
Ian Gilman
@iangilman
Aug 26 2016 18:42
:D I guess physical pyramids usually have the tip at the top, but I think of dzi pyramids as having the tip at the bottom
John Susi
@johnsusi
Aug 26 2016 18:43
haha
Ian Gilman
@iangilman
Aug 26 2016 18:43
Cool, so we're seeing the same thing. I think that needs fixing
Thanks for spotting it by the way!
John Susi
@johnsusi
Aug 26 2016 18:43
thanks for fixing the initial problem ;-)
Ian Gilman
@iangilman
Aug 26 2016 18:44
That was @captaincodeman... I'm just facilitating :)
John Susi
@johnsusi
Aug 26 2016 18:44
I'll do some more digging, I need to learn more about the inner workings of openseadragon anyway
Ian Gilman
@iangilman
Aug 26 2016 18:44
Sounds good... let me know if you need any guidance along the way!
John Susi
@johnsusi
Aug 26 2016 18:45
will do
Ian Gilman
@iangilman
Aug 26 2016 18:45
:)
John Susi
@johnsusi
Aug 26 2016 18:47
another unrelated issue. I had some trouble when switching between tall and wide images in the same viewer.
I solved it using fitBounds: new OpenSeadragon.Rect(0,0,1,1) (when calling addTiledImage)
but is it the "right way"? :)
Ian Gilman
@iangilman
Aug 26 2016 18:49
You're doing multi-image?
If so, then yes, doing a fitBounds after adding a new image makes sense
If not, you could use open which automatically does a fitBounds
(actually it does viewer.viewport.goHome, which calls fitBounds)
Keep in mind that if you don't specify the width/height of your image it defaults to a width of 1... that means for tall images they might have a height of more than 1 (and wide images have a height of less than 1)
You can always do fitBounds(image.getBounds())
John Susi
@johnsusi
Aug 26 2016 18:52
not specifying width and height could be the source of my problems
Ian Gilman
@iangilman
Aug 26 2016 18:53
It doesn't really matter if you just have a single image, but with multiple images you might care
John Susi
@johnsusi
Aug 26 2016 18:53
yeah and I need to preserve where in the image the user is looking even if the sizes change
Ian Gilman
@iangilman
Aug 26 2016 18:54
Cool... you'll definitely want to pay attention to the image bounds then