Seeking Workflow for Consistent Talking Head Videos on RTX 5060 Ti

NewnovaMae · reddit · 2026-08-28

A user is seeking a workflow to generate 3-5 second talking head videos with natural micro-expressions and head movements without video guidance, using an RTX 5060 Ti (8GB VRAM) and 16GB RAM. The user has experimented with Juggernaut, LoRA, and Flux 1 dev but struggles with maintaining character consistency. Despite limited hardware resources, the user has made progress with GGUF and is looking for a method to achieve perfect character consistency.

Original post →

More from Multimodal

Multimodal channel →