Review: Epiphan DVI2USB 3.0 Frame Grabber

Syncing the Slides

The more genteel approach is to acquire the speaker’s PowerPoint/Keynote, export the slides as individual image files, and use the screen recording as a reference video, replacing the video with still images of the slides at the points at which the speaker moved through the deck. This process can be labor-intensive; for me it typically takes at least an hour to do this for a 45-minute presentation, and with a sizeable backlog of sessions to edit, no editing help, and multiple conferences in succession in the fall and spring, sometimes I find it tough to build this step into my workflow.

Red Giant offers a nifty plug-in called Presto (Figure 3, below) that automates this process (provided you have the reference video of the screen so the application knows when to sync the slides), and produces lovely results. Unfortunately, its NLE compatibility is limited, and not recently updated, and it doesn’t work with my NLE of choice--Adobe Premiere Pro for Windows, and thus doesn’t leverage GPU acceleration. I’ve found it extremely slow—more time-consuming, if less labor-intensive, than syncing the slides myself.

Figure 3. Magic Bullet Presto

The other limitation of the slide-syncing approach is that it sacrifices some of the nuances of the presentation. For example, if the presenter summoned bullet points one at a time on some slides to coincide with the moments at which she introduced specific points in her talk, those elements will be lost in translation if you simply sync the static slides. And if the presenter made last-minute changes to the slide deck, those changes may not be reflected in the version made available to you. Finally, if the presenter showed any videos are animated graphics, you’ll either have to request original files or revert to the “shoot-the-screen” version to include that component, resulting in a potentially jarring loss of image quality.

Enter the Epiphan

A commenter on an article I wrote last winter about streaming with the Livestream Broadcaster clued me in to a company called Epiphan that’s developed a family of relatively low-cost PC/Mac-based products designed specifically to provide full-motion video screen captures of external devices. Their offerings range from the low-end VGA2USB (MSRP $299) to the high-end VGA2Ethernet ($1,600)--both external devices--and DVI2PCE Pro ($2,200), an internal capture card.

Figure 4. Epiphan's DVI2USB 3.0 frame grabber. Click the image to see it at full size.

I really wanted something I could manage and control entirely from my riser, if necessary working with the laptop I typically have on hand, without requiring anything from the presenter, the presenter's laptop, or the A/V company except a split feed and a VGA drop. Given my limited budget and the fact that my specific needs require capturing feeds directed to VGA projectors (and thus pulling feeds from VGA video distribution amplifiers/splitters), the question was whether to purchase the VGA2USB or the $699 DVI2USB 3.0 (Figure 4, above). Even though the DVI2USB 3.0 was quite a bit more expensive and meant I’d need an additional VGA (female)-to-DVI (male) adapter, it was a fairly easy decision. The key limitation of the VGA2USB is that its maximum capture frame rate declines as its capture resolution increases. So while it can capture 640x480 at a respectable 28fps, it’s limited to 3.1fps at 1920x1080. Although I may never need to capture full 30fps video at 1920x1080, I didn’t want to purchase a device that would lock me into something lower, so I went with the DVI2USB 3.0, which promised full-frame-rate 1080p30 capture, after tracking down a used unit on eBay for $450. Figure 5 (below) shows the Epiphan capturing a PowerPoint from Streaming Media East 2014.

Figure 5. Capturing a PPT with the Epiphan Capture Tool.

Capturing a Viable Video Stream

The Epiphan Capture Tool produces AVI files that Adobe Premiere Pro can't recognize or import. So the next workflow challenge, if you’re capturing video to edit in post, as I generally am, is to choose a “DirectShow-compatible” screen capture application to actually record the feed in real-time in an editable format. After striking out with a few other options, I ended up going with Windows Media Encoder, which allows me to choose the Epiphan as a capture device, capture just the Epiphan Capture Tool window (and thus run other windows on top of it as needed--see Figure 6, below), and set encoding parameters. For the most part I’ve been recording 24p video at 10Mbps at whatever resolution I get from the source feed (which varies depending on what I get from the presenter’s PowerPoint or keynote--see parameter settings in Figure 7, below Figure 6). It will even capture live audio of the session with an internal or attached mic, which won't sound very good, but will be very helpful for syncing the captured presentation stream with the video of the speaker in post.

Figure 6. Choose this setting to record only the Epiphan Capture Tool window (or any specific window from a running application).

Figure 7. Setting compression parameters in WME by importing a profile.

Here’s an example of what the Epiphan delivered at the DBTA Data Summit conference in New York City last month. While the stage was by far the darkest I’ve ever shot, and the video of the speakers suffered accordingly, the full-motion, full-quality screen capture went a long way to compensate for the deficiencies of the video of the speakers.

This workflow may sound a bit convoluted--and it took some trial and error to make it work, a challenge compounded by the fact that I couldn't completely simulate the conditions of recording a conference session in my home office, and thus essentially had to start testing it live while filming the opening keynote at the Computers in Libraries 2014 conference in April. But however convoluted it may sound, once I sorted out the initial difficulties, I found this live production workflow very manageable as a single operator running two cameras (one fixed wide shot, one MCU following the speaker) and managing the Epiphan/WME capture (really just a matter of starting and stopping a recording session for each presentation, using preset parameters) simultaneously for numerous sessions each day throughout two days of filming at Data Summit 2014 in May in New York.