A step-by-step look at turning a simple prompt into a short AI-generated video, plus the workaround needed to reach the platform in the first place.
AI video tools have gotten fast enough that a usable 15-second clip can now come together in a few minutes, provided you know how to structure the prompt correctly. Seedance 2.5 is one of the newer models making this possible, but the interface it runs on doesn’t include the usual dropdown menus for aspect ratio or visual style — those details have to be written directly into the prompt. Here’s how the process works, from writing the prompt to downloading the finished clip, along with the separate steps needed to get into the platform itself.
Setting Up the Prompt
Inside the platform, under the Pro tier, there’s a Skills option. Scrolling down from there brings up Generate Videos, which is where the process starts.
Unlike most AI video generators, this interface doesn’t offer built-in controls for aspect ratio or style — there’s simply no menu for either. That means both have to be spelled out inside the prompt itself, along with which model to use. Duration doesn’t need to be specified, since 15 seconds is the default.
So a complete prompt needs to cover three things:
- Aspect ratio — written out manually, since there’s no selector
- Style — same situation, no dropdown available
- Model — in this case, Seedance 2.5
Once the prompt is written with those details included, it gets copied and pasted into the main chat window, then sent.
Confirming and Generating
After the prompt is submitted, the tool responds with a written summary — duration, ratio, style, and model, echoing back what was specified — along with a rough breakdown of how the shots will be structured. At that point, it asks for a go-ahead, accepting responses like “yes,” “generate,” or “start.”
Typing “yes” and sending it kicks off generation. Because this is a free tool, it can take a little while, and it may prompt for additional confirmation partway through — those prompts need a response to keep things moving.
There’s also a plus icon in the interface that allows for image uploads, suggesting an image-to-video option is either in beta or simply not live yet. For now, the tool only supports text-to-video.
The generation process itself happens in two stages. First, it produces a handful of still images — typically five or six — based on the prompt. Once those are ready, it asks whether to convert them into a video. Replying with the same confirmation line (“yes,” “generate,” etc.) triggers that conversion, and the video begins generating.
When it’s done, a Save button appears next to the finished clip. Clicking it downloads the 15-second video.
Getting Into the Platform
Reaching this tool in the first place requires a couple of extra steps, since the site doesn’t load without a workaround.
Start by searching “Urban VPN Chrome extension” on Google and installing it — click Add to Chrome, then Add extension. Once it’s installed, pin the extension for easy access, open it, and accept its terms. At this stage, there’s no need to connect to any specific country yet.
Next, search “Dola AI” on Google and click the first result. The page won’t load on its own. Open the Urban VPN extension, select Pakistan as the location, and connect. Once the VPN connects, refreshing the Dola AI page brings up a Fast option, which needs to be clicked, followed by Pro.
From there, selecting Continue with Google brings up an account picker. After choosing a Google account, the platform asks for age confirmation — clicking Confirm completes the login. Once signed in, clicking Fast and then Pro again lands back on the interface where the video generation steps above take place.
A Few Things Worth Knowing
Access to unlimited generations through this method isn’t guaranteed to last — free tiers and workarounds like this one tend to get restricted or shut down without much notice, sometimes within days or weeks of a tool gaining attention. It’s also worth being cautious about routing a Google login through a VPN-accessed, region-restricted site, since that combination isn’t something every platform’s terms of service are designed to accommodate. If the tool disappears or changes, the underlying prompt structure — specifying ratio, style, and model manually — is a useful pattern to carry over to whatever comes next.