Full tutorial: put yourself into a Drake video with ChatGPT Image 2.5 + Higgsfield Genjutsu
chrisfirst · x · 2026-09-20
A creator shared a complete three-step workflow for inserting himself into a Drake music video using Higgsfield Genjutsu:
- Grab the first frame: screenshot the opening frame of the target video and prepare a reference photo of yourself
- Image swap: feed the screenshot plus your photo into any tool with ChatGPT Image 2.5, prompting it to replace Drake while preserving scene details, matching lighting, and keeping your outfit
- Motion transfer: import the generated image and the original video into Genjutsu, prompting it to use the image as first-frame reference while transferring the original motion
No training required — just image editing plus video motion transfer.
More from Multimodal
- LTX Video 2.5 Measured: 121-Second Single-Pass Generation, Resolution-Duration Equation Derived — Robotman2100 · 2026-09-20
- SPEED V2: Retraining-Free MiniMax-H3 Speedup Adds 5 Samplers and Fixes a Degradation Bug — antipode_insights · 2026-09-20
- Frustrated Suno user switches to open-source YuE2, matches old model quality locally — lazyspock · 2026-09-20
- Running Comfy workflows via a frontier model: one dev's local image-gen pipeline — mccoypauley · 2026-09-20
- ntsc-td plugin emulates the VHS/NTSC signal itself, not an overlay — PurzBeats · 2026-09-20
- Creator tests H3 on a music video: solid car physics, engine sounds and anime-style character consistency — Mr__Earthling · 2026-09-20