The 8-second wall
Every Veo 3.1 generation stops at 8 seconds. That is the model, not your plan — no subscription anywhere unlocks a longer single render. What gets you past it is chaining shots with continuity frames, which carries one scene to 148 seconds. It comes with a trade we will not bury: the long version is 720p.
7-day trial · 50 tokens · no credit card
The limit
It is worth knowing exactly what you are up against before you pay anyone to get around it, because most of the offers to get around it are clips glued together and sold as a feature.
The ceiling
Veo 3.1 accepts a duration of 4, 6 or 8 seconds per generation. 8 is the ceiling. Every product built on Veo inherits it, including this one — we are not going to pretend otherwise.
Coherence decays with length. Past a few seconds a video model drifts off the face, the room and the light it started with, and the compute bill climbs with every second. Eight seconds is where Google drew the line between a shot that holds and one that wanders.
A larger plan buys more generations, not longer ones. There is no tier here, or anywhere else on Veo, that returns a single 148-second render. Anyone selling "unlimited length" is selling you clips joined end to end.
Handing the last frame of one shot back to the model as the first frame of the next. Veo is still only asked for 8 seconds at a time; the chain is what makes the result read as one continuous take instead of four separate ones.
How you pass it
Scene extension is a loop, not a setting. Four steps run for every 8 seconds of finished footage, and the only one you pay video rates for is the third.
The loop
Nano Banana 2
A still image of the first moment of the scene, generated from your description with your cast injected as reference images. It costs a token or two and it anchors everything downstream.
Veo 3.1
Veo films the shot from that frame — dialogue, sound effects and room tone generated inside the clip, not layered on afterwards.
Continuity frame
The final frame of that shot becomes the conditioning image for the next generation. Same face, same room, same lens, same light. This is the whole trick.
Draft and Quality modes
Chained this way one scene runs to 148 seconds and plays as a single take. Fast mode renders shots in parallel for speed, so it has no previous frame to pass on and does not chain.
Loop: last frame of shot N → first frame of shot N+1
The same loop is what runs between scenes as well as inside one, which is how a production reaches 99 shots without the cast changing. The full pipeline is written out on how VeoStudio works.
The honest part
Veo’s scene extension runs at 720p. That is a constraint of the model, not a paywall in this app — there is no tier, no plan and no price here that returns a 148-second scene at 1080p. So the choice is real, and you have to make it per scene.
Most films want both, in different places: extend for the long take, cut at 1080p for the hero shots. What you should not do is buy anything — here or elsewhere — expecting a 148-second 1080p render, because nothing on Veo produces one.
One more limit worth stating plainly: Fast mode renders shots in parallel to save time, so it never has a previous frame to pass forward. Chaining runs on Draft and Quality only.
Beyond length
Two minutes of drifting faces is worse than four 8-second shots that match. The things that make long-form watchable are not length features at all.
What holds it together
Characters are generated once, given an ID, and injected as reference images into every frame and every shot that uses them. Without that, the person at minute two is a different person.
1–4 tokens per stillThe link between shots is an actual image handed from one generation to the next, plus the closing action of one scene written into the opening action of the next. Prompt wording alone does not hold a room together.
Automatic, no extra chargeVeo produces dialogue, sound effects and ambience inside each clip, so sound does not slide out of sync as shots are chained. Music is laid over the finished timeline afterwards.
In-clip, per shotUp to 99 shots, each with its own mode, aspect ratio, duration and prompt. When shot twelve is wrong you re-render shot twelve, not the film.
Up to 99 shotsWhat it costs
A chained scene costs exactly what its seconds cost — there is no surcharge for extending, and no discount either. Stills for the frames run 1–4 tokens each depending on the image model.
| What | Mode | Resolution | Tokens |
|---|---|---|---|
| One 8s shot | Draft | 720p | 12 |
| One 8s shot | Fast | 720p | 30 |
| One 8s shot | Quality | 1080p | 80 |
| One 148s chained scene | Draft | 720p | 222 |
| One 148s chained scene | Quality | 720p | 1,480 |
In plain terms: the 7-day trial’s 50 tokens are about 33 seconds of chained Draft footage, enough to write a production, cast it, draw the frames and film a scene. Pro is $25 a month for 300 tokens, roughly 200 seconds in Draft, and it is the plan that opens Quality mode. Top-ups are available at any time — the full ladder is on the pricing page.
Straight answers
Not in one generation. Veo 3.1 accepts 4, 6 or 8 seconds and 8 is the ceiling, on every plan and in every tool built on it. Longer footage is made by chaining: the last frame of one shot becomes the first frame of the next, which carries a single scene to 148 seconds and a full production to 99 shots.
Coherence and cost both degrade with length. Past a few seconds a video model starts losing the face, the room and the lighting it began with, and every extra second multiplies the compute. 8 seconds is where Google set the boundary for Veo 3.1. It is a model limit, not a subscription limit — no plan unlocks a longer single generation.
148 seconds, built from chained 8-second shots with continuity frames between them. Beyond that you cut to a new scene; a VeoStudio production holds up to 99 shots stitched into one film, so total runtime is not the constraint.
No. Veo's scene extension runs at 720p only, so 148 seconds and 1080p are mutually exclusive. A 148-second continuous scene is 720p; 1080p is available on Quality mode's native 8-second shot, which you cut between rather than extend.
Draft mode charges 1.5 tokens per second, so a full 148-second chained scene at 720p is 222 tokens. Quality mode charges 10 tokens per second, or 1,480 tokens for the same length. The 7-day trial grants 50 tokens, and the Pro plan is $25 a month for 300 tokens.
Try the chain
The 7-day trial grants 50 tokens and asks for no card. At 1.5 tokens a second in Draft that is around 33 seconds of chained, continuous 720p footage — long enough to see for yourself whether the continuity holds.
7-day trial · 50 tokens · no credit card