-
Notifications
You must be signed in to change notification settings - Fork 151
Development log
Welp, haven't updated this in a while! Checked in a pull request this morning and made a few tweaks to the styling, deployed version 0.4.2.
Added a few more functions for displaying photos, options for hiding the controls and using a click-and-drag motion instead of the standard mouse following motion. Bumped version up to 0.2.0, felt good.
Created a Todo list in the wiki of features I'd like to implement at some point, in no particular order, but mostly with the easy ones at the top. Listened to this old Alan Parsons Project record I have and hung out on the couch writing on my ipad. Programming has certainly become a cozier endeavor in my lifetime.
Also started adding demo.html into the zip build, along with additional examples. Branched off onto a dev branch and began adding support for still photos. Kind of hack at the moment, but totally works. Need to find a nice Mars or Moon panorama or something to use.
Many updates over the weekend. Cleaned up the repository and grunt tasks for deployment and build, created a new website to host code and examples, added onscreen controls for fullscreen play and mute with font awesome icons, api calls for updating the video file. Have to cleanup the zip file download to include a working demo and see if I can make it work with images as well.
Having the ability to split a very large image texture into pieces to put it all on the gpu would let this be a fairly good panorama image viewer as well.
Updated readme with more detailed explanations. Added scripts to grunt for copying demo files to the github demo hosting repo. Deployed an actual demo of the code there and provided links. Implemented play and pause functions, and added lon/lat settings as passable options.
Did a whole bunch of research on canvas fallbacks and mobile rendering. Long story short: WebGL on mobile isn't up to this particular task yet. Given what I know, it wouldn't be hard to write an app that would do it, but getting people to install an app is another story.
If I could get the video texture to work, we'd at least have a good chance of support on all new Android phones going forward out of the box, but it's hard to say if that'll work easily. Going to post a bug report in mrdoob's repo and see what they say.
Finished a basic jQuery plugin implementation, cleaned up the demo files so they didn't have Valiant-specific references (though the video file still isn't in the repo). I have to export a really short high res clip to show, probably the thumbs up or reno clip. Gotta clean up the documentation next, and update build scripts to export a ready-to-use minified version that includes jQuery and ThreeJS. Also I need to wire up the play, pause and maybe seek functions.
Began migrating to a standard jQuery plugin, probably another 4 hours or so before I have something basic working with a more standardized interface.
Picked the project back up tonight and began officially licensing it with the standard MIT license. There've been a few new cool 360 degree video projects coming out, so seems like people might actually be interested in this code I've had hacked together for a while. I'll try to get my server-side stitching code together as well.
Today I began rearranging files to be less specific to my road trip footage, and planned out the beginnings of a roadmap for the project. It seems like it would be useful to more people now that 360 degree videos are becoming common.
Got git push->deploy working (git push live master), and synced files up to the server. Then did research on different camera modules once I started noticing artifacts in the video. Looked into various modules that might someday build an error-free 360 degree rig with real frame syncing and everything. Realized how much work hardware is and put it on the back burner, trying to nip avenues of thought like that in the bud before I end up doing them and blowing a year of my life.
Also copied all 65k extant rendered panorama frames. I sort of think that the video in this one is good, it's a solid proof of concept, but to push it further will really require a different approach. Did some light color correction on a few segments and dumped to full res files, they are enormous and pretty at 3576x1589. Nearly 4k, though at the top resolution it's not exactly usable. Still, it's beautiful to look at when it's at full res, though I can only imagine what you could get if you actually had good enough hardware.