Streaming Media

Streaming Media on Facebook Streaming Media on Twitter Streaming Media on LinkedIn

How to Create Custom Picture-by-Picture Looks for Remote Streaming Production

If you're looking for a way to elevate your livestreaming productions beyond what clients can do themselves, then you will want to take over more control in your productions, starting with the PxP and PiP looks, using external hardware and/or software. The benefit of this approach is that your recordings will generate a standard HD signal for later editing and on-demand viewing on different platforms.

Software Solutions

While hardware solutions have the advantage of presets and ease of use, software solutions remove many of the limitations of hardware switchers at the expense of ease of use. Pricing can range from free, with open source software like OBS, to $350 for vMix HD. Vimeo Premium account holders might be surprised that their subscription includes free use of Livestream Studio 6, which I understand to be a white-label version of Telestream’s Wirecast Studio. Definitely check your hardware and OS against the published minimum requirements for best results. And then, before you go live, test everything.

Software video switchers may have presets for creating PxP or PiP looks, and these serve as a good starting point (Figure 3, below). But one of the reasons you may be considering a software solution is that you have more control over all aspects of each of your input video signals in your composite output.

Figure 3. vMix software offers several presets for handling multiple video signals in a single display that can be further customized.

The first step is to add your computer presentation and video camera inputs. Since a software solution can internally emulate a webcam UVC output to your video conference solution, you aren’t limited to just using a webcam. So if you want to connect a professional video camera using a USB or PCIe capture card, you can. In vMix, the external output emulates a webcam signal that your video communication SaaS can see.

Then, you need to open the computer presentations that you want to share or add the photos or video files to the software. A big pro tip is that this whole process will be much easier if you connect two or more computer monitors to your computer or laptop and get familiar with controls for resizing multiple windows on a single computer monitor.

You then want to add the video and computer input to a blank canvas (or a background template if you have created one). Initially, your inputs will layer one on top of the other, so you will need to resize and position these inputs so they are next to each other (PxP) or so that one smaller one is on top of the other (PiP).

Software solutions have many options to add a computer presentation. These include an external video input with a capture card, screen capture, or NDI connection  You can also enter the URL of a cloud presentation, such as PowerPoint, online. I have used all of these options as some point during the past year.

Creating a 50/50 PxP Look

Let’s assume you don’t have presets available to you, and you need to build each look from scratch. You will have to scale each input to 50% if you want to fit two 16:9 inputs next to each other. Then, you will move one left on the horizontal X-axis by -0.5 and the other one right by 0.5. This 50/50 look gives equal weight to both the video and the computer input. It assumes that both signals are the same 1920x1080 HD resolution and that this is what your project is set to. The same principles apply if you are working with a 1280x720 project. If you have an external video camera or an HD video camera with a full HD 1920x1080 output, then you’re in good shape. But many webcams operate at different resolutions, and you may need to make adjustments depending on what you are trying to accomplish.

If your webcam video signal isn’t the same size as your computer signal and you want it to be, adjust the scaling on the video input until the vertical size matches. Then, adjust the X-axis if needed so that the video fills its 50% share of the screen. If the video signal has a 16:9 aspect ratio, then you might be done here. But if the video signal has a nonstandard aspect ratio, which is typical on many webcams, you have a bit more work to do.

If I see that we have matched the vertical size of the video input, I will assume that if there is an issue, it is with the horizontal size of the video input, which will have to be increased. Because you obviously don’t want to stretch the video, you will have to overscale it so that the horizontal size matches the computer signal. Adjust the X-axis if needed to center it on its 50/50 share of the screen, and then crop the video signal as needed to match the vertical size of the computer signal.

Check your work by zooming in to ensure that you are pixel-accurate. I recommend creating your template with a fake computer signal instead of a PowerPoint presentation. This fake signal can simply be a solid color, which I like to use because if your PowerPoint template has a border or a background that matches your background, you may not be able to accurately discern the edges until you go live and realize your error when a background changes. I’m speaking from experience here.

Another pro tip that makes the often tricky work of resizing and cropping inputs a bit easier is to create a template with transparent alpha channels where your video and computer input would show through. Then, place the template on the front or top layer, with the video and computer signals below. This ensures that you have clean edges, which is especially important when you are working with nonstandard aspect ratios.

I want to emphasize the importance of working with nonstandard aspect ratios and signals, considering the pandemic-induced increase in remote presentations. If your presenters are connecting with you remotely, you may need to scale and crop their video signal to remove elements that are added by the video conference service.

Creating a 40/60 PxP Look

The 50/50 look works when you occasionally show a wide shot and need all of the horizontal real estate in a 16:9 video signal (Figure 4, below). The reality is that half of your screen is not being used by the video or computer presentation. The workaround to fill the screen is to add titles, graphics, logos, and an interesting background, but this leaves something to be desired if you want the presentation to be larger and more readable.

Figure 4. The 50/50 look (top), and the settings used to create it (bottom)

For a single presenter, I like to use a 40/60 weighting, with both inputs scaled to 60%. To achieve the 40% weighting on the video signal, you need to crop 320 pixels from each side of the video. Having said this, if you plan your layers with the front and back in mind, you may not need to crop anything.

Start with the video as the bottom layer. Scale to 60%, and pan your X-axis to -0.6. In my sample template, the cutout for the Y-axis is not centered, so I have to adjust my Y-axis by -0.21 below the center. I did add 320 pixels of cropping on the X1 axis and cropped to 1,600 pixels on the X2 axis, but this step isn’t required, and I am only showing it in case you are working with nonstandard-sized inputs or you got your layer order wrong and need to make additional adjustments. The left side is automatically clipped because 320 pixels of content is outside my work area, and I am going to place the computer presentation in the layer on top of the video, which will clip the additional 320 pixels on the right-hand side that I want to cover. What I am left with is a scaled 1280x1080 signal, which is effectively 768x648 within my 1920x1080 workspace.

Scaling the computer presentation to 60% reduces it to a 1152x648 element. I can quickly check my math by adding the horizontal size of both my video (768) and computer (1,152) elements to equal 1920. I then need to move the computer input along the X-axis to 0.4 and adjust my Y-axis to the same -0.21 as my video input.

Adding Elements Like Logos, Graphics, and Titles

If you use a 50/50 or 40/60 weighting, you still have 50% or 40% of your screen left for graphic elements. Depending on your hardware or software, you might be able to add these elements to help fill the screen. I like to engage my clients’ graphic design team to create a template in advance that works for both of us. It doesn’t always go exactly as planned, as clients are prone to getting you the files too late in the process for you to send them back for changes. But properly done, they usually look better than what I can quickly create in the field. I often still need to run the team’s template through Photoshop to make a few quick changes, including replacing the video and slides placeholder with a transparent alpha channel and correcting the size of the placeholder if the team got the math wrong.

However, most of my clients leave it up to me to create their look, so I think it is important to have their logo in at least two formats. I like to have a version that is good on light and dark backgrounds—always with a transparent background—and one for RGB color spaces. PNG files work well for this, but I often have to convert Adobe Illustrator or Encapsulated PostScript files myself.

The same principle that we used to scale and move the video and computer inputs applies to the graphics: Keep the order of the layers in mind so your graphics appear on top of the background and not behind it. Some hardware solutions might require you to use a downstream keyer (DSK).

Creating a PxP Look With Multiple Video Inputs

Creating a single video next to a single computer presentation is usually the limit for hardware solutions, but software solutions often allow you to display multiple live video inputs at the same time. This is especially important when you are working with multiple remote presenters over a video conferencing solution.

I used to use Google Meet and Zoom a lot as a source of remote presenters. I would pin individuals and bring them into my software by pinning individual participants and by using NDI or HDMI outputs from one or multiple computers. Now I am moving my remote productions to Microsoft Teams, as it has native NDI support and I don’t need as many computers to manage my remote participants with. Figure 5 (below) shows one look I can create using the same techniques that we used before, except now with more video inputs.

Figure 5. APxP look with four video inputs (top), and the settings used to achieve it (bottom)

I started this one with the same 60% scaling on the computer presentation as in my previous example and a 40% weighting on the four video inputs. For a 2x2 grid, each video input needs to be scaled to 30%, and each needs to be positioned in a grid. The same cropping amounts of 320/1,600 that we previously used will work here as well. Start by making sure the tops and bottoms align to match the computer inputs. Then, move the four video inputs, and adjust the cropping as needed until they fill the 40% video space. Here are the X and Y coordinates I used to make my 4x4 video grid:

  • Upper Left: 0.8/0.12
  • Upper Right: -0.4/0.12
  • Lower Left: -0.8/-0.48
  • Lower Right: -0.4/-0.48

My final pro tip is to get creative when designing your multiview templates. Make your own templates in Photoshop, and don’t limit yourself to rectangles for video signals. Also, plan looks for different combinations of the total number of presenters you will display simultaneously. The reason for this is that despite your best intentions, you may not always want to display all of the available remote presenters because their signal could disappear on you, as often happens with remote connections.