Because of COVID-19 we have shifted many of shows to a videoconference format, leveraging NDI inputs and Skype’s native NDI support as our way to do shows and interviews. So far it has worked like a charm.
Now, I’m trying to achieve some things, like border, shadow and rounded video sources, like social network avatars, mostly as a proof of concept that I could implement in following shows.
Is there a way to achieve this in live video sources using MIXER commands or something? I’m open to ideas.
I’ve been thinking of using png cutouts and overlay them for borders and shadow, which I’ve tried and works fine so far, as long as transitions are direct cuts. But I have many shows which leverage transition times for transforms and crops, so this would mean I’d also have to manipulate the overlays transitions, transforms and crops if needed. It would take some time to have a show prepared for this, though.
As @hreinnbeck said you can use defere and commit or you could use a virtual channel where you build your composition and route the whole channel down to al layer on the output. You can then use the MIXER FILL commss as md to animate the whole thing ar once. Could be easier to do all the calculations.
Here is that template modified to accommodate four inputs, each one into a different canvas inside a circle:
The template has some extra bits that a client makes use of located in helpers.min.js (I included the non-minified version). There are some files missing: gsap.min.js and CSSRulePlugin.min.js from GreenSock and seriously.js files. You can download them at GitHub.