| Hi, my name is Michael and I'm the founder of TyphonRT Media, Inc. It's just me and I have been bootstrapping various game
and performance middleware for Java for over 10 years. My story is one of a consultancy attempting to turn into a product
oriented company. I however chose a difficult category, new media tech (music-tech, initially, creative engine tech now), to
innovate in and one that requires hardcore engineering. Last year I was in the right place at the right time with the right
tools to start to build a next-gen video engine from scratch on Android. Android 4.3, last August, introduced the ability for
the MediaCodec API (low level hardware accelerated encoding / decoding of audio & video) to be combined with OpenGL ES.
Android being Android there were many challenges (re: bugs) to overcome with MediaCodec and such over the past year, but I
got things stable on the majority hardware configurations. In that time I have spent $80k out of pocket (all professional
earnings in last ~1.5 years) on the video engine and roughly 2500 hours on the engineering side building the engine over the
last year. I am running a Kickstarter to fund final launch efforts / release engineering; I'm ~95% of the way to launching and can
do it by the end of Q1 '15! A video engine of this complexity takes a lot of testing and I'll be staggering the launch initially on well tested hardware
configurations (Snapdragon & Tegra K1 SoCs). I am attempting to raise $20k (28% funded presently). The crowdfunding attempt
first and foremost is to raise awareness of the tech and to reach out to early enthusiasts for new photo / video tech on
Android. If I don't raise that amount it simply means the launch will be delayed as I go out and do another 3rd party contract
to keep the lights on in the meantime. Ultimately I have funded ~80%+ of this effort out of pocket. Now "next-gen" is a bit buzz-wordy, but indeed I built the graphics capabilities of the video engine on OpenGL ES 3.0 and
will also be supporting 3.1 which is the latest standards (compute shaders!). This allows many internal engine improvements
over GLES 2.0. The engine itself supports ~150 image operations and the first app in the suite is a creative video capture
app that offers deep effects composition capabilities while still providing an easy to use and expressive GUI on phones. It's
possible to stack up to 8 different image operations and modify them in real time while recording video. A high-concept product comparison is: "Adobe After Effects in your pocket." It so happens that
many of the previous limitations (FBO limits) on mobile have been lifted over the last year with new mobile GPU tech allowing
significantly more realtime post-processing capabilities. The Snapdragon 800 / Adreno 330 GPU (Nexus 5) being the first phone
form factor SoC capable of truly deep real time effects manipulation. Of course this is continued with the Snapdragon
805 / Adreno 420 and others like the NVidia K1 all of which are currently stable running my video engine. I suppose the really cool thing is that I essentially made the missing video / audio middleware for Android that can service
many apps and am immediately interested in releasing some of my own, but also am considering licensing the engine to 3rd parties.
I am aiming to launch a suite of modern creative video apps for Android which is a category sorely lacking on Android. Mainly
this is due to the fact that Google doesn't release a high quality video engine to app developers to create media apps with. In
comparison Apple does release a fully featured video engine and quality APIs for media development and thus there are a lot of
video / media apps for iOS and compartively very few for Android. I do want to give an acknowledgement to Brad Larson and the GPUImage open source effort. While my video engine efforts don't
share any software architecture with GPUImage I found the catalog of image operations (mainly fragment shaders) very helpful
when building my tech and as a starting point this allowed me to really focus on overcoming issues with MediaCodec versus
recreating a stable of fundamental image operations in GLSL. I fixed several issues with the shaders in regard to how they
work in combination and upgraded everything to OpenGL ES 3.0 as the baseline GLSL version. I'd be glad to answer any questions on the tech or even bootstrapping while doing "hardcore engineering". It's been a long road
to launching product and I'm almost there. Join the announcement email list at www.typhonvideo.com
For a direct link to the sizzle reel video and better quality check it out on YouTube: https://www.youtube.com/watch?v=AriO_JktgjI Thanks in advance for any support from the HN community! [edit]: I post a lot about tech / graphics topics including the progress of my video engine work on my G+ page: https://plus.google.com/+MichaelLeahy/posts I'd be glad to connect and chat with anyone interested there too! |