Find the next useful adjustment
AI Video Guides and Prompt Examples
These AI video guides cover the decisions around a short generated shot: the input, the motion and the result. Start with the question you have now. Each tutorial leads to the matching tool.
Only have an idea?
Read the text prompt examples, choose one scene and generate five seconds. You can decide whether to extend the duration after you see how the movement reads.
Already have a product photo and no preference about the camera? Open Product Agent. The manual prompting chapters are optional for that workflow.
Reading path 01
Before the first generation
Decide what the model should invent and what it should inherit. This is the shortest route if you have not chosen a mode or prepared an image.
You know the shot you need.
Should I start with a sentence or a photo? ↗
Compare the source of composition, identity and motion. The decision examples also show when Product Agent removes the need to write a prompt.
Choose a Generation ModeYou have a few candidate images.
Is this photo a useful starting frame? ↗
Inspect edges, background clutter, cropped parts and repeating details. Choose a photo before spending credits on a motion test.
Choose Your Starting ImageYou know where the video will appear.
Why is my output the wrong shape? ↗
See which modes offer a ratio selector and which inherit the image. Plan a portrait or wide composition without mistaking cropping for generation.
Video Aspect RatiosReading path 02
Put the movement into words
Choose the guide that matches your input. A prompt that builds an entire scene has a different job from one that animates a photo already on the screen.
Text to Video; no starting image.
How much of the scene should I describe? ↗
Build a prompt from a subject, action, setting and camera. Copy a complete example or make one deliberate revision after watching the output.
Text to Video PromptsImage to Video; a photo plus direction.
What should move in this photo? ↗
Separate subject movement from camera movement. Compare a fixed camera and a small pan, then try a prompt with fewer competing instructions.
Image to Video PromptsA product photo in Image to Video.
How do I direct a product shot myself? ↗
Use manual directions for a push, a light change or a material detail. The guide distinguishes this workflow from Product Agent, which prepares its own prompt.
Product Video PromptsA shot that needs a particular camera move.
Does “orbit” mean what I think it means? ↗
Compare a push, pull, pan and orbit by how the viewpoint changes. Choose a move whose hidden geometry the image can reasonably support.
Camera Movement PromptsReading path 03
Inspect the result and decide what to change
Keep the original export beside the source image. The next useful adjustment depends on whether the problem is size, identity or the destination frame.
You are choosing an output setting.
Is 768P worth using for this shot? ↗
Compare two actual exports from the same source and direction. Read their dimensions, file sizes and credit costs, then decide where the extra detail matters.
480P vs 768PA generated clip with a visible error.
Why did the object change halfway through? ↗
Classify the failure before trying again: geometry, texture, identity or movement. Work through the smallest useful revision instead of adding a long negative prompt.
Video DistortionProduct Agent in Single mode; two compatible views.
Can I give the clip a destination image? ↗
Prepare a starting and ending view of the same product. Check their lighting and framing, then inspect the transition rather than assuming every middle frame will be correct.
First & Last FramesUse a guide beside the generator
The prompt examples in these AI video guides can be copied or opened in a matching tool. Opening an example fills the prompt; it does not submit a generation. If that tool already has a different draft, choose whether to keep it or replace it.
Change one instruction before comparing two results. A different camera, source image and duration all at once make it difficult to tell which decision helped. Keep your source and downloaded clips so you can inspect them together.
Find the right level of help
The AI video guides answer a production question. If your question is about the finished placement, such as an Etsy listing or a Shopify product gallery, use the use case directory instead. Those pages cover the details to check before handing a clip to that destination.
Examples show individual outputs, not a promise that a repeated prompt will produce the same frames. Use what you can observe: a stable outline, a readable material, enough space around the subject and a motion that fits the intended edit.