Live compositing with a video switcher

How would you handle a request to produce 8 hours of picture-in-picture video? And deliver it by end-of-day, tomorrow?

Innovation is in the HP DNA. We have renewed our historic focus on products and services, and are committed to getting them to the right customers.
 — Chandrakant Patel, HP Senior Fellow and Chief Engineer of HP Labs

Patel says: “At HP, product development involves a number of stages, including research, brainstorming, team identification, development and commercialization. Through internal events like HP Software’s InnoBlast, HP brings together the best and brightest to ensure that they are making products that matter and bringing them to market in a timely fashion.”

I was asked to record one of the first all-day InnoBlast events and to embed the talking-head presenters into their associated PowerPoint decks or screen demos. There were to be a total of 23 presentations and most would accompany their ideas with the PowerPoint decks and a few would show a screen demo or video during their presentation.

As we started to brainstorm how to capture and deliver this project we first tossed around the idea of distributing each presentation with Adobe Presenter. In Presenter, the talking head video would appear in one column of the interface and the deck or demo would fill the rest. There are many advantages to Presenter if you are using an LMS system but there are disadvantages if you’re just looking to deliver a video. The specific drawback here was that you would need a single video for each PowerPoint slide. We’d end up having to track time codes for slide changes during the presentations and then slice the talking head videos to match the time codes of reach slide. That’s a lot of work!

I wanted a way to eliminate as much post production as I could and to capture as much of the presentations “in camera” as it were. I threw out the idea of compositing the talking head video picture-in-picture style with the presentation and recording this live-to-disk. An additional advantage was that this could be streamed behind the firewall whereas the Presenter PDFs could be distributed once downloaded. That was a real issue since the presentations are proprietary.

We were now on agreement as to what the deliverable would look like. I just needed to decide on how to execute the quickest and most cost-effective workflow. I knew that the Blackmagic Design ATEM 1 M/E had a DVE feature I had yet to use and that would allow me to crop and shrink the talking head video and embed it in the video picture-in-picture style. As asked for all the decks to be standardized in 4×3 format to allow an additional column in my 16×9 deliverable that could hold the inset. I was told building a 16×9 template with a designated area for the talking head would not work for this project.

The big challenge I had at the time was getting the video out of the laptop and into the switcher. Today, I would use HDMI out or a DisplayPort to HDMI converter. At the time of the project I had to work with a VGA out. This would require a scan converter to convert the signal to something usable. We had a Matrox Convert DVI Plus on-hand from a previous series of projects that would accomplish this. This particular model had HD-SDI out which turned out to be a benefit — the Convert DVI Plus was easily over 50′ from the editing station. An HDMI solution would not have worked because of the distance.

At the end of the day, the biggest problem this project faced was getting the feed out of the ConvertDVI. We had a variety of problems that surfaced at different times throughout the day. At one point, the laptop kept re-installing the drivers. While this did not affect the recording you did see the message appear above the windows task bar every few minutes. We had tested the setup in advance of the event and this problem did not happen! Knowing in advance that the ConvertDVI was finicky we had two laptops on-hand and pre-tested for the event. Both had different problems the day we were live.

ConvertDVI Plus issues to the side, the recording itself was textbook simple. There was no real switching going on since the ATEM was only combining the talking head with the laptop feed. We recorded the Program feed directly to H.264 with the Blackmagic H.264 Pro Recorder (don’t try this today. There are audio sync issues with this box). For safety, we recorded camera ISOs and ran TechSmith Camtasia on the presenter’s laptop. A trick here is to enable audio recording in Camtasia so that you can sync the Camtasia to video later in Apple Final Cut Pro or Adobe Premier.

The next day all we had to do was take the 23 QuickTime files and add opening / closing slates to the files and deliver the project.