Added functions to Enable/Disable Speech-To-Text and to load the model:

- When STT is enabled and a track is selected, a subtitle track is automatically created.
- When the STT subtitle track is selected, the model is loaded, and the input is buffered.

Create a Loader interface to asynchronously load the desired Speech-to-Text model.

Use a boolean to automatically select the new ES or not.

We assume the first display must be in HDR for Windows to enable SDR. We don't know
which adapter will be used at this point nor on which output.

When forcing or disabling HDR we don't want to match the display

Only one of them needs special handling between DComp/HWND.

We can avoid some scaling rounding if it's not applied.

At this stage all regions are picture-based (ie not text regions). We don't need to check it anymore.
Even if someday display modules can render text regions on their own, they can surely do the scaling too.

We already require these display modules to be able to place and scale/stretch the video picture.

A display module may set a value but not be used in the end.

The zooming factor is applied once in the core, no need to make the display module compute it.

The local blend_fmt was just exactly the same.

The picture format already has a way to set cropping values in the picture.
We just need a clone of the picture to change the metadata.

This is how the main picture is handled as well

They are only shown in the display area, regardless of the picture cropping.

The core cannot give us a null scaling.

Maxime Chapelet authored 1 year ago and Steve Lhomme committed 1 year ago

* OpenGL compatibility only for macOS.
* OpenGLES compatibility everywhere else but visionOS

Given this, every decoded buffers will be compatible for all available
Darwin compatible vout renderers.