@DominicCarterLA@yapper_so I don't know if Yapper pulled the control, but I still lock an approved still as the first frame, and set a last frame when the camera path is already boarded. Without those, 2.5 tends to renegotiate the shot mid-clip. I check the last frame against the still before I keep it.
@ohneisserdemy@liluocheng13 A 17s train move is where I stop trusting the opening. I still give Seedance an end frame so the aisle can't smear halfway, and I often cut the path into short pieces. A clean mid-shot can hide seats that have already melted.
@EmperorSandy Same pattern here. Extra control: freeze a hero still of the face first, then I2V with motion-only words — don’t re-describe the look mid-prompt or the model renegotiates the face. One verb beats an epic ask every time.
@NabhaniMehdi Doors and stairs are where AI invents a third hand or floating foot. Cut before contact, or shoot two shorts from a locked still: approach plate, then hand on handle / foot on step. One continuous walk-through-the-door ask invites weird hands.
One full-body photo → you lead a zombie crowd down a night street. 14 seconds, dance music built in. 🧟
Made with AI: this is the example clip from MixVio's AI Thriller Dance page.
1/4
One prompt. One take. Sound included.
Kling 4.0 makes a single 3–30s clip with native stereo audio. This ultrawide lobby scene is one generation. Made with AI.
#Kling4#AIvideo
3/4
The sound is generated with the picture. Name what you want to hear: "solo grand piano, heavy rain, waves on the hull."
Audio is always on, so describe it instead of hoping.
Official fashion show? The FAQ says: "No. This is an independent catwalk effect. It is not affiliated with a designer, a show or a brand."
8 s, 16:9 or 9:16. 480p 52 / 720p 112 / 1080p 268 credits.
Make my catwalk video: https://t.co/mbCJBuGCel
An 8-second AI catwalk from one front-facing photo: you walk a lit runway in an emerald satin gown, turn once, blow a kiss and wave. 👗
Made with AI: sample clip from MixVio's AI Catwalk Video page.
Your photo decides the last 3 seconds. Those are close-ups, so a sharp, large, front-lit face keeps you recognisable. Skip side profiles, group shots and heavy filters.
One photo on a gravel path. Fifteen seconds later you are chipping a keeper under floodlights. ⚽
Photo → chip → pause, all AI-generated from a single full-body picture. 🧵
What stays: the green goalkeeper and both defenders.
What changes: only the player on the ball.
You wait out two defenders, then side-foot it into an empty net. Hands and fast legs can vary, so watch the whole clip.
@felam216 Hands and transitions are where we see the most drift too, so that would be really useful. Feel free to DM us whenever you want to swap notes.
@felam216 Thanks! We don't share a public roadmap, but on the full-body side we keep tightening the photo checks so fewer runs drift, and we post new effects here first. Happy to compare notes anytime.
@felam216 Agreed. If the first frame is clean, the rest of the clip has something solid to follow. Thanks for sharing what you're seeing on your side too.