@Joshua Guerrero I struggled with that for a long time but I figured out a few tricks

@Carrie Crews Start with a character sheet for your AI Twin and use that as your main reference. Include front, side and ¾ views, plus different facial expressions like neutral, smiling, laughing, surprised, serious and talking. Lock the identity into EVERY prompt. Describe the face, hair, eye colour, skin tone, age, body proportions and any other features that must not change. I also tell it not to beautify, de-age, reshape or substitute the character. Don’t rely on the reference image alone. I use Google Flow and create one clip at a time. When the clip is right, I capture the final frame and use that as the starting reference for the next clip along with the next prompt. That keeps the character, clothing, lighting, background and position flowing naturally between clips. A few other things that help: • Keep an identity description saved and paste the exact same wording into every prompt. Don’t keep rewriting it. • Lock the outfit too if it’s meant to be the same scene. Clothing loves mysteriously redesigning itself. • Describe the environment consistently, including lighting, time of day and important background details. • Be specific about body movement, hand movement and facial expression. Don’t just say “she talks to camera.” • Keep camera instructions consistent. If one clip ends as a medium shot, don’t accidentally start the next one as a close-up unless you actually want the change. • For speaking clips, specify natural mouth movement, blinking and subtle facial movement so the Twin doesn’t look frozen apart from the lips. • Write each clip as one small action rather than trying to cram an entire scene into one generation. • Save your successful prompts. Once you find wording that keeps your Twin consistent, reuse it rather than making AI prove it can break something that already works. My basic formula is: Character sheet + fixed identity description + scene continuity + one clear action + final frame from the previous clip.

A beginners guide to AI video generation 😁

@Joshua Guerrero its in the image prompts classroom for the moment

@Suzie Kemper

@Dr. Robert Gault of course

F for me. I’ve been working on turning more of what I create into repeatable systems instead of having to start from scratch every time.

Welcome @Georg Malinovski