Native 9:16 Portrait Support and Best Practices for Dynamic Backgrounds

Hi HeyGen Engineering Team,

We are developing a mobile-first AI tutoring application and are currently evaluating the LiveAvatar and Streaming SDK. Our app is designed specifically for a 9:16 Portrait orientation. To optimize our asset production, we have a few technical questions regarding aspect ratios and background handling:

Native 9:16 Portrait Support
The official documentation recommends 1920x1080 (16:9 Landscape).

Does the LiveAvatar training engine officially support a native 1080x1920 (9:16 Portrait) MP4 upload?

If we upload a 9:16 video, will it be processed at full vertical resolution, or is a 16:9 landscape source strictly required?

If 16:9 is the only option, does the SDK offer a server-side Center-Crop feature for 9:16 displays?

Dynamic Backgrounds vs. Multiple Avatar IDs
We need to use various backgrounds during a single session.

Do we need to create separate Avatar IDs for every different background? (Does each version consume an additional Avatar Slot?)

Alternatively, is it technically recommended to record the source footage on a Green Screen (#00FF00) and perform Client-side Chroma Keying in our app (e.g., using Flutter shaders) to swap backgrounds dynamically?

Are there any known issues with color consistency or compression artifacts during WebRTC streaming that might affect chroma key precision?

Recommended Specs for Portrait Cropping
If we must use a 16:9 source and crop it to 9:16 on the client-side, what are the recommended Bitrate and framing guidelines during recording to maintain high pixel density for the portrait area?

We look forward to your professional guidance to help us build a high-quality user experience.

Best regards