Hi there, I talk about this in pretty great detail in the #ai-image-prompts channel of the Discord. You can join here: https://discord.gg/vcBG244DzC
In short: I use Illustrious (Wai, v17) via ComfyUI for the anime images and then Wan 2.6 for some light editing, and a combination of a bunch of models for the IRL ones (sometimes Z-Image Turbo via ComyUI, other times GPT-2, other times Seedream v5, sometimes Nano Banana 2 via third-party cloud services).
I share my LoRA (Illustrious) on the Discord to recreate the style, along with my prompts + some suggestions on best models to use.