consistent characters + mood in Google Veo 3... it’s possible ✨ I used to spent hours refining text-to-image prompting (mostly in midjourney) trying to keep character and tone consistent – now I’m trying the same in T2V, seeing how far I can push visual continuity using just words here's... what's been working for me so far 🧵show more

here's my Veo 3 prompt set-up: → shot type → year → setting/scene description → character description → lighting → color palette → sound → mood → optional: voice over text

character consistency: be very specific in details, and repeat the full description in every prompt! example: A 10-year-old boy, slim frame, buzzcut light brown hair. He wears a white and navy-blue vertically striped t-shirt, loosely tucked into high-waisted denim shorts, and white crew socks slouched into worn red converse sneakers it's not a magic spell, but it helps. especially when you can't use I2V for continuity

color palette + mood: stick to one tone unless the scene calls for change. keep your color palette tight and your mood aligned across prompts – less "chaos magic", more control

Does veo work well with specified cameras, lenses, or film types? Is it possible to consistency that way? Combinations like: —Cinestill 800T —Angénieux Optimo Prime 75mm T1.8 —Arri Alexa 35 I'd love to hear your experience with this stuff!

I'll test these tomorrow! 🥹

“Getting motivated to go to the gym was always hard for me. But PowerBeatsVR has helped me reach a weight I never thought I'd reach!" ⭐⭐⭐⭐⭐ — PowerBeatsVR User

It is clearly an improvment for scene consistency 👏👏 But I think its almost impossible to create production ready results with only prompts(T2V), they'll either need to make the duration longer, or make it able to process I2V consistently which is worse rn with Veo 3.

extend in the scenebuilder is pretty usable, I can create pretty consistent scenes up to 20 seconds so far with it without it completely losing it's pace – definitely getting there :)

Thank you, Julie! I hope that soon I can use my Midjourney images to generate videos with sound and sfx. Fingers crossed!

Just use TwinAI for better character consistency.