At A Glance Main Projects Tutorials Resume

Messenger


There currently is no messenger. Manually, you can contact me via:
AIM: TecGumbo

Google Talk: palen1c at gmail.com

Twitter: @Technogumbo



Quick Agile Setup With Docker And JIRA

Sun, 15 Apr 2018 21:53:51 EST

Everyone is crazy about the Agile process. When used in real world projects with real teams does it live up to the hype? I have enjoyed using JIRA for Agile project management each time I have used it. As someone who has managed and used Agile via JIRA I attest that as long as Agile is utilized in the correct scenario, it works well. Keeping team focus on the product and not the process helps.

The quickest way I have come across to setup JIRA is to use a docker image of it. Instead of the involved process of creating your own image, someone has already done the hard work.

You can find the docker images for JIRA at: https://hub.docker.com/r/cptactionhank/atlassian-jira/

Comments:0

Quick No Hassle PlpgSQL Stored Procedure Debugging

Thur, 15 Mar 2018 21:52:51 EST

Of all of the open source databases; PostgreSQL has always been my favorite. I have used it on several projects, and it has always proven reliable. Recently, I have been working on increasingly complex stored procedures. During development, it can be difficult to determine error states. A quick way to print output that will be visible is to utilize the raise notice statement.
It works very similar to print statements in other programming languages. For example:

raise notice '% %', declared_variable_1, declared_variable_2;

Will print the values of declared_variable_1 and declared_variable_2.

Comments:0

How to Achieve Greater Than 4K Resolution Playback With HAP

Fri, 13 Oct 2017 21:51:51 EST

When working with public displays you often have to pull off non standard video playback. Examples include video walls, multi-monitor setups, and multi-projector displays. The challenges when working with these displays often revolves around flawless video playback at large resolutions and sync.

Flawless video playback starts to become very difficult (in 2017) at resolutions greater than 4K. Traditionally, the larger native resolution of a video, the more computing resources it takes to achieve playback. Many consumer playback codecs also fall short at larger resolutions. The most popular codec/format, H.264 does not currently support resolutions greater than 4096x2160. If you attempt to run resolutions larger than this on modern operating systems, you may see no playback at all.

The sync problem has to do with multiple devices playing back one video source in perfect synchronization. This is not a trivial task. When you reach the maximum amount of monitors/projectors one device can support, you have to utilize multiple independent devices to accomplish full playback. In 2017 If you also want to play back a video source larger than 4K, you often have to utilize more than one hardware device to do it due to the limitations outlined above.

A six by one array diagram.
Commercial off the shelf solutions often handle this by utilizing one device as a master and others pull time code or similar information from the master over a network to synchronize the other hardware playback devices. I feel for the engineers who had to program these solutions. They are very difficult. You have to throw out frames on some sources to stay in sync and guess as to actual synchronization. It is a very difficult balancing act. Commercial solutions I have worked with that utilize techniques similar to this are the popular Brightsign hardware playback devices, and solutions using Watchout. Both of these solutions can be very expensive depending on how big of a display you are creating.

I didn't even mention the complexity of producing video for multi-display setups. Often editors will produce one massive video, but then have to split it up into multiple regions for playback on the different devices.

Thinking outside the box regarding sync, resolution limits, and playback you can avoid the difficult synchronization problem if you can play back ultra high resolution on one device. In a recent project, we were able to pull off 11520px x 1080px playback on one computer. This was a 1x6 orientation for a video wall; but using the same technique you could pull off: 5760 x 3240 in a 3x3 orientation. There are a few modern video cards that support up to 6 displays. You can also use techniques that chain cards together to get multiple outputs on one computer. We had to utilize a computer because the application functioned as ultra high resolution playback as well as an interactive multi-touch wall display.

A three by three array diagram.
The true secret behind that massive resolution is the Vidvox HAP codec. HAP was designed for video DJ activities. When working with HAP you end up with monstrously huge files. Our 4 minute 11520 x 1080 file was around 30 GB in size. When working with HAP you often need a SSD. HAP utilizes very low CPU resources but requires massive disk access speed. Even big players in the space like Dataton's Watchout now support HAP.

I was previously utilizing Adobe Media Encoder to create HAP compliant videos. Adobe Media Encoder takes too long to encode HAP. Thankfully as of September 2017 HAP encoding in FFMPEG Has been fixed as well. If you have a build that supports it, you can utilize a command like this to transcode an input video to HAP with AAC audio:

ffmpeg -i InputVideo.mov -vcodec hap -format hap -chunks 4 -acodec aac OutputHap.mov


"chunks" is an important HAP specific command. The default is 1. Chunks are not supposed to exceed the number of cores the playback computer has. I noticed with a chunk of 1, I was experiencing large memory utilization on a quad to eight core machine. A chunk setting of four really smoothed things out in my use case. My custom application for the multi-touch display and super resolution video playback was written in Unity3D.

AV Integrator's often use off the shelf components such as Brightsigns or Watchout, but with the combination of HAP + SSD savvy installers may want to take a second look at rolling their own solutions. They may end up with significant cost savings if they have the programming expertise to pull it off.

Comments:0
Techno Gumbo RSS Feed Michigan Livestock Data Services The Research Computing and Engineering Podcast