Zooming

Hello, I am currently working on a personal project that requires a zooming interface in flash. It should be very flexible, and allow adding of applications and widgets. I thought at first I would use a 3D engine such as papervision, but after thinking about it, I found there was way too many features I did not need and that I'd be better off coding from scratch something that would really fit the purpose, and not loose too many CPU cycles over 3D thingies that I would definitely not need. My problem is: I don't know where to begin. I have my own ideas, but I don't know how good they are. So I am turning to you guys for any kind of guideline. Advice, links to some similar open-source project (even if not flash-based), description of the code flow for a zooming interface, links to articles about zooming interfaces...Any bone you can toss at me will be warmly welcomed. Even just a few words of wisdom will do. I am sorry if I am out of my place, the forum does not follow a regular structure and I had a hard time figuring out if I could just post my question like this. Also, sorry for my English. I really hope you can help. I am quite lost. Thanks in advance.

There are a couple of options! For Flash I suggest OpenZoom: http://openzoom.org/ I would also recommend Seadragon Ajax (it came from the group I work with, fyi): http://livelabs.com/seadragon-ajax/ There's a Library as well as a simple Embed. I'd love to see what you come up with! -Kevin

thanks! Openzoom seems interesting...I will probably use it, or at least borrow lines of code. I'd rather re-code than use it as is for learning purposes as well as to accommodate my hungry ego. my I had heard of it but if you recommend it then I am reassured regarding its quality. Seadragon Ajax is new to me. Sadly it does not fit my purpose at all, but the idea is simply awesome. The architecture I am considering goes like this: interface (captures keyboard/mouse, overlays static graphics, passes instructions to canvas) ---> canvas (determines what is visible and performs scaling on required objects) ---> objects (icons, applications, texts, etc) So: - The interface captures a key event, let's say, (+), or mousewheel down. - The canvas would first calculate what objects are visible in the next frame, then send instructions to those objects, with their next scale as argument. - the objects would then render a relevant graphic (for example, I want titles to remain the same size at all times). According to you, does this make sense? Knowing that each object has it's own different views relative to its Z position (like I said, titles don't scale, some objects display more information about their content when they are close, etc), do you think it would be better to put them all in a layer object and manipulate that layer, then send relevant data to the objects so they render another graphic? I think it is a bad idea as it would require to scale the whole layer on each frame, even if some stuff is not visible, but it might be quicker because the objects would be re-rendered only when they cross a certain limit. May I ask your opinion on this matter? [edit] I do not know how much you know flash (please do not take it as an offense, one can never be too sure), but to the extent of my knowledge, in AS3, I have no access to graphic cards capabilities or cache or memory management, so in order to optimize my code, I can only manipulate objects.

Also, another few questions, if I may bother you more: - the app I am thinking of will hardly ever have anything to do with scaling images. It will probably scale icons and texts most of the time. Is openZoom relevant in such a case? - Knowing that my app should be open to additional applications, the most logical thing to do would be to have the coder of each application provide several icons, let's say 10, each on a different scale, and load these icons when needed. This said, knowing that the app is on the net, failing to load the icons when attaining a particular scale would leave an empty icon for a while, something that is simply not possible. The app should remain slick at all times. Facing this issue, I have three solutions, as far as I can see: 1) load all icons size from scratch. pros: they are cached and ready to be used anytime. Cons: longer loading times, and in case there are many icons, more memory required than it should (unless I make an icon cache library that I re-use for similar icons, but it might get complicated, and no guarantee that the same icon will be re-used enough to make this relevant) 2) use the scaled icon until the new icon is loaded. Pros: shorter loading times, loading only what's needed. Cons: if the user scales up/down several times very fast, on different points, I might end up with dozens of downlads at the same time. This would require an advanced loading queue (knowing that flash doesn't allow for resuming, icons that began loading and are abandoned will have to be downloaded all over again). 3) have the coder provide one high-scale resolution of his/her icon, and calculate pixel maps of resized versions for all sizes. Pros: easier time for the coder/designer, all icons are available from start. Cons: same as in option 1, lots of memory required. I might simply do this: 4) require a vector icon, in swf or svg. Although if I understand correctly, transformations on vector objects are slower than on pixels (in flash at least). Pros: easier time for the designer, only one icon to load; can be scaled to any level without pixelation; nothing to cache but the original icon. Cons: maybe slower? No freedom for the designer: he has to make a vector icon. Lots of questions.....And I have many more coming. You do not have to answer. I am posting just in case you or someone has a suggestion, or simply thoughts.