How to make a long AI video (over 30 seconds) that looks like one take
Every model in the AI Video Generator stops at 30 seconds, and most stop at 15. A 60-second ad is still easy: plan it in scenes, keep the model and style fixed, match the frames where clips meet, and join them free in your browser.
· 5 min read · 773 views · by the Cortex Digital Hub team
Make your first scene on a cheap draft model, then build the next scene from its last frame. The credit total shows before you press Generate. From ₹24 a clip · no watermark · runs in your browser
Make a video → → Opens AI Video Generator · runs in your browser
There is no 60-second button in any AI video model. The longest clips in the AI Video Generator are 30 seconds, and that is only on Vela 2.5, Orbit 3, Orbit Prime and Director 4; most models stop at 10 or 15. A long AI video is really a short film: four or five clips made the same way, cut on calm frames, and joined. Do that with a little care and nobody watching on a phone will spot where one clip ends and the next begins. Here is the method, with a costed plan for a 60-second ad.
Plan it in scenes before you spend a credit
Write the video as a shot list first. Each line is one scene with one subject and one camera move, which is also the way the tool's own tips say to prompt. A 60-second ad for a Hyderabad pickle brand might be: the jar on a kitchen counter, the pickle being made, a family at the table, the pack with the price, the logo. Decide which scenes carry the story and deserve 15 seconds, and which are 5-second beats.
Where a scene needs several angles, use a multi-shot model. Lumen Motion 3 and Orbit 2.6 can hold three or four scenes inside one 15-second clip, so write the cuts into the prompt ("close on hands mixing spices, cut to the jar being sealed, cut to a wide of the kitchen") and let the model handle the edit.
Keep scenes that will sit next to each other in the same location and the same light. A jump from warm kitchen light to blue evening light between clips reads as a different shoot, even if everything else matches.
Lock the model, the style words and the seed
Mixing models is the quickest way to make the joins visible. Each model has its own way of rendering skin, light, grain and colour, and the eye catches the change even when the subject is the same. Pick one model for the whole video and stay on it.
Write a style block of 20 to 30 words and paste it at the start of every prompt: "warm afternoon kitchen light, 35mm look, shallow depth of field, muted colours, slow handheld". Keep Resolution, Quality and Format identical across clips. In the Advanced options, type the same Seed number into every clip where the model offers it, and switch off "Let the model expand my prompt" so the model does not add its own ideas differently each time. Treat the sound toggle the same way: on for every clip or off for every clip, because a clip with ambience next to one without is the audio equivalent of a jump cut.
Match the frames where two clips meet
This is the part that makes a long AI video look like one take. The last frame of clip A becomes the first frame of clip B. Open clip A, pause on its final frame, and take a screenshot. Drop that screenshot into the next generation as the image: in image-to-video mode one image becomes the first frame, and the prompt then describes only the motion from there ("the camera keeps pushing in slowly as a hand lifts the jar").
On models with first and last frame control, such as Lumen Frames, Lumen Omni and Prism 6, you can go further. Give clip B both a start frame (clip A's end) and an end frame (where clip C should begin), and the model has to travel from one to the other. Plan those hand-off frames to be calm: a held wide shot, a still product, a hand resting. A hand-off in the middle of fast motion is where seams show.
Use Extend when the scene should simply continue
Extend video adds 4 to 30 seconds to a clip you already have, continuing the same motion. It is the right choice when the scene does not change, such as a camera orbiting a product that should go on orbiting, or a slow dolly that needs a few more seconds for the voice-over. Vela 2.5 is the model to use for this, and because Extend is billed on the input clip's length plus the new seconds, trim the input to the part you want continued before you send it.
For a change of scene, Extend is the wrong tool. Start a new clip from a matched frame instead.
A 60-second ad: five clips, costed
Here is the pickle-brand ad as a plan, all on Lumen Motion 3 at 1080p so the look stays constant. Prices are as of October 2026, at 1 credit = ₹0.50.
| Clip | What is in it | Model and setting | Seconds | Credits | Rupees |
|---|---|---|---|---|---|
| 1 | Jar on the counter, slow dolly in, steam from a pot behind | Lumen Motion 3, 1080p | 10 | 370 | ₹185 |
| 2 | Making the pickle, multi-shot: spices, mixing, sealing the jar | Lumen Motion 3, 1080p | 15 | 555 | ₹277.50 |
| 3 | Family at the table, multi-shot: serving, tasting, a laugh | Lumen Motion 3, 1080p | 15 | 555 | ₹277.50 |
| 4 | Pack close-up with the price card, slow orbit | Lumen Motion 3, 1080p | 10 | 370 | ₹185 |
| 5 | Logo on a plain wall, light sweep, hold for the end card | Lumen Motion 3, 1080p | 10 | 370 | ₹185 |
| Total | 60 | 2,220 | ₹1,110 |
Before spending that, draft the whole ad on Orbit 3 at 480p. At 14 credits a second the full 60 seconds is 840 credits, or ₹420, and it tells you whether the shot list works before the final render. Fix the prompts on the draft, then re-run them on the final model with the same seed. If a model fails or declines, the credits come back on their own. Expect 1080p clips of 10 to 15 seconds to take a few minutes each, and 30-second clips eight minutes or more; you can leave the page and they will appear under "Your videos".
Join the clips free, then export for where it is going
Download each clip within seven days. Open the Trim Video tool and cut the head and tail of each clip so the hand-off lands on the calm frame you planned; it cuts without re-encoding, so the quality of the render is untouched (more on that in trim video without losing quality). Then join the trimmed clips in order with the Video Converter, which runs on your device and never uploads the files.
Finish for the platform. For Reels and Shorts, generate at 9:16 in the first place rather than cropping; if you did not, the video resizer fixes it. For WhatsApp Business broadcasts, a 60-second 1080p ad is usually over the comfortable sharing size, so run it through compress video to 16 MB or follow the WhatsApp compression guide before you send it to channel partners. Then open the AI Video Generator, make clip 1 on the draft model, and build the rest from its last frame.
Frequently asked questions
Which model makes the longest AI video clip?
As of October 2026, Vela 2.5, Orbit 3, Orbit Prime and Director 4 go up to 30 seconds in one clip. Most other models stop at 10 or 15 seconds, so anything longer is made in parts and joined.
Can I make a 60-second AI video in one go?
No. Plan it as four or five clips on the same model with the same style words, make each clip start from the previous clip's last frame, and join them with the free trim and converter tools.
Does using the same seed make the clips identical?
No. The seed nudges the model towards the same look and grain, which helps the joins, but the content still follows your prompt. Matching the hand-off frame does far more for continuity than the seed.
How much does a 60-second AI video cost?
About 2,220 credits (₹1,110) for five clips on Lumen Motion 3 at 1080p, or 840 credits (₹420) as a full draft on Orbit 3 at 480p, as of October 2026. Credits are shown before you generate and refunded if a model fails.