Auto-Pre-Mix

Automatically Pre-Mixing Inputs

Allow one to auto-pre-mix (aka pre-render or flattening) source inputs with the help of two intermediate Mix-type input(s) in order to further embed the result onto a layer of a target input.

Problem

vMix is an awesome video mixing software, but it is still rather limited (especially in contrast to OBS Studio) when it comes to its input layering: when creating complex input trees the maximum layering depth is 2, and when applying effects like positioning there is only the single, inherited, original positioning. The reason is the design weakness of vMix that no pre-rendering/flattening of layers actually happens.

In practice the following scenario of a greenscreen-based and virtual-PTZ-based scenario illustrates the problem: The N physical cameras are on inputs PCAM-N. For applying a chroma-key filter (without changing the original input), N virtual inputs VCAM-N are created from the PCAM-N inputs. For replacing the greenscreen with N pre-rendered background images (inputs BG-N) and performing K virtual pan/tilt/zoom (PTZ), NxK VPTZ-N-xxx inputs (of type Virtual Set / Blank) are created with each having the background image BG-N on layer 1, the VCAM-N on layer 2 and an individual Position for the PTZ. X scene inputs (X <= NxK) SCENE-X can now be created with each having VPTZ-N-xxx on layer 1, and e.g. slides and titles on additional layers.

Unfortunately, when a slide is too unreadable and you want to exchange the virtual camera VPTZ-N-xxx and the slide by creating a special scene SCENE-PIP with the slide on layer 1 and the virtual camera VPTZ-N-xxx on layer 2 (as a PIP) you cannot achieve the PIP effect because the Position of layer 2 interferes with the Position already in the Virtual Set VPTZ-N-xxx. Even a virtual input from VPTZ-N-xxx doesn't help here.

Solution

Fortunately, vMix 4K/Pro have Mix input types which actually perform pre-rendering/flattening of inputs -- even if they are primarily intended for switching between inputs. Unfortunately, vMix provides only a maximum of just 3(!) of those Mix input types. But we can re-use a Mix input type by dynamically re-configuring it to show (and pre-render) a particular input when it a scene containing it comes into PREVIEW. In order to not change the PROGRAM, two Mix inputs are used in total.

For our practical scenario, you create two Mix-type inputs PRERENDER1 and PRERENDER2 and in the PIP-scene SCENE-PIP you now place the slide on layer 1, PRERENDER1 on layer 2 and the virtual camera VPTZ-N-xxx on layer 3, but disabled. The Position for the PIP is now done on layer 2 instead of 3. When the SCENE-PIP comes into PREVIEW, the auto-pre-mix script detects the scenario and automatically switches PRERENDER1 to VPTZ-N-xxx. In case PROGRAM already uses PRERENDER1, the SCENE-PIP is re-configured to use PRERENDER2 and then this Mix is switched.

Usage

Create two Mix-type inputs PRERENDER1 and PRERENDER2 and on any target input, place any Mix input onto layer N and any source input on layer N+1.

Alternative

An alternative for pre-mixing is to use a VirtualSet with a single layer based on a full-screen UV Map (1:1 mapping). See vMix-VirtualSet-Flatten for a realization.