The Video Pipeline turns a prompt, and optionally an image, into a video. You must be signed in. Generation runs as a queued job, so the page can confirm that the job was queued before a playable video exists. Open the finished result from your own video listings.
1 · Check the keys for your chosen model
Open your API settings before starting. The pipeline reads keys saved for your account; this guide supplies no key. It requires an OpenAI key even when you provide the starting image. For Kling V1.6, also save a Replicate key. For Kling V2.0, Veo 3 (Fast) or Veo 3, save a FAL AI key. These providers can charge for generation under your own account terms. If a needed key is absent, the queue acknowledgement does not mean the job can finish. The form itself does not check key readiness for you.
2 · Open the pipeline and choose a type
Go to Video Pipeline. Under Video Type, choose Text-To-Video to start from words or Image-To-Video to animate an image. The form opens with Image-To-Video selected. Choose the type deliberately before filling the fields: the image controls appear for the image path, while the text path offers an idea prompt.
3 · Pick a model and describe the clip
Under Video Model, choose one of the models for which you have the matching key. Set Video duration and Aspect Ratio. Enter the motion and scene you want in Video Generation Prompt; you can also use Negative Prompt to name things to avoid. If you chose the text path, add an idea in Idea Example Prompt. The duration menu expresses your requested length, while generation can take longer; it is not a completion timer.
4 · Add an image when using the image path
For Image-To-Video, upload a PNG or JPG in Start Image. You may also add an End Image. The starting image gives the generation job its first frame. Without one, this path has no starting file to animate. If you have only an idea in words, go back to Text-To-Video instead. Review the prompt against the image before queuing: a clear description of the desired motion is more useful than simply repeating what is already visible.
5 · Decide whether the clip needs sound
Add Background Music opens extra choices for music; Add Voice Narration opens narration choices. Leave them off for a first clip if you do not have the related inputs and provider keys. AI generated music uses a saved Sonauto key. AI narration uses the saved OpenAI key. Selecting either option adds another generation step, and each can fail separately from video generation. For a Veo model, Generate Audio (Veo 3) is another option; its default is No.
6 · Queue the job
Press Create Video once. The page first says it is queuing the video, then shows a success message when the job has been accepted. That message is a queue receipt, not a finished result. If the form reports an error, review the selected type, the image upload and your account keys before trying again. Do not expect a finished video to appear immediately after the button press.
7 · Find and watch the result
Use See My Creation at the top of the form, or open your Video Pipeline listings. A finished item has a thumbnail and Watch; View Video opens its listing. If the page says No videos found, the result has not appeared there yet. Return later and check the listing again. Generate New takes you back to the creation form when you want another clip.
What this does not do
This guide does not create a key, guarantee that a queued job succeeds, or publish the video to any social account. Video generation depends on your saved provider keys and on those providers being available. The video waits in your own pipeline listing for you to review; any later use or sharing is a separate choice you make.