Re: [CODE4LIB] very large image display?
On Jul 25, 2014, at 11:36 AM, Jonathan Rochkind wrote: > Does anyone have a good solution to recommend for display of very large > images on the web? I'm thinking of something that supports pan and scan, as > well as loading only certain tiles for the current view to avoid loading an > entire giant image. > > A URL to more info to learn about things would be another way of answering > this question, especially if it involves special server-side software. I'm > not sure where to begin. Googling around I can't find any clearly good > solutions. > > Has anyone done this before and been happy with a solution? If you store the images in JPEG2000, you can pull tiles or different resolutions out via JPIP (JPEG 2000 Interactive Protocol) Unfortunately, most web browsers don't support JPIP directly, so you have to set up a proxy for it. For an example, see Helioviewer: http://helioviewer.org/ Documentation and links to their JPIP server are available at: http://wiki.helioviewer.org/wiki/JPIP_Server -Joe
Re: [CODE4LIB] very large image display?
Jonathan Rochkind wrote: > For this project, there is really only a handful of big images, and > simplicity of server-side is a priority -- so I think it's actually okay to > pre-render all the tiles in advance, and avoid an actual image server -- to > the extent tools can work with this. to make static tiles: for deepzoom vips[1] is the faster. as version 7.30[2] you can run $ vips dzsave huge.tif my_dz_dir deepzoom.py[3] is another option, but slower. if you want to consider zoomify (usable with a leaflet plugin) there is zoomifyimage[4]. inside the package in contrib/netpbm_shell_script you'll find a bash script[5] wrapping netpbm -- raffaele, @atomotic [1] http://www.vips.ecs.soton.ac.uk/ [2] http://www.vips.ecs.soton.ac.uk/index.php?title=What%27s_New_in_7.30 [3] https://github.com/openzoom/deepzoom.py [4] http://sourceforge.net/projects/zoomifyimage [5] https://gist.github.com/anonymous/0f8e623561b7ae57ff88
Re: [CODE4LIB] very large image display?
Hi, Version 2 of the iipmooviewer also works with static dzi/deepzoom see under protocols in the readme): https://github.com/ruven/iipmooviewer. It also has some support for annotations & "image blending" which may cover your layering use-case. Eoghan On 26 Jul 2014 07:58, "Toke Eskildsen" wrote: > Code for Libraries [CODE4LIB@LISTSERV.ND.EDU] On Behalf Of Jonathan > Rochkind [rochk...@jhu.edu] wrote: > > > Then I eventually found OpenSeadragon, which a couple other people in > this thread > > suggested, which looks like a pretty good fit. It looks like it possibly > can work with > > entirely pre-rendered tiles served statically with no image server, > using the "DZI" format. > > (http://openseadragon.github.io/examples/tilesource-dzi/). I haven't > actually gotten to > > a proof of concept here, but I think it'll work. > > That is what we use for a tool for ad-hoc QA of scanned newspapers: > https://github.com/tokee/quack (does not require ALTO-files, just > images). It works very well and fairly simple for a few thousands of > images, but scaling to millions would probably be problematic due to the > amount of very small files on the file system making backup and similar > operations very heavy. We are planning to use OpenSeadragon in production, > where the image backend will be pyramidal TIFFs wrapped in an image server. > > We do not use annotations but do use overlays. Multiple layers is > simulated by placing a large transparent PNG as an overlay, but that is > heavy to render for the browser on 30MP+ pixel images, making zoom somewhat > choppy. I would not attempt it on 100MP+. > > - Toke Eskildsen >
