No, you do not. Omitting certain components grants the model creative freedom to produce your video. We recommend starting with a simple prompt that focuses on the most critical visual and motion components and then adding more detail to refine as-needed.

Both work, but natural language usually gives you more control. When you write in full sentences, you provide context that helps guide exactly how elements appear in your scene.

Keywords are useful for setting a general direction—think of them as suggestions. The model will include what you've specified, but interpretation and incorporation may vary between generations.

Using a consistent prompt structure makes iteration much faster and easier. Rather than hunting through scattered details to find what you need to change, you'll know exactly where each component lives.

This means you can quickly adjust specific elements without accidentally disrupting other parts of your prompt or missing details buried in the middle of a paragraph.

There's no ideal prompt length—focus on clarity rather than word count. Overly long prompts can create conflicting requests and may limit the creative freedom that produces surprising, dynamic shots.

However, Text to Video prompts are typically longer than Image to Video prompts because they need to describe both the visual elements and the motion in detail.

No. The order in which elements are introduced in a prompt do not matter.

Timestamp prompting helps control when specific actions occur by specifying timestamps for each action. Though it may not be perfectly precise, it effectively guides the general sequence and timing.

A needle-felted orange and white Corgi character wearing a yellow, green, and orange sweater stands in a grocery store aisle, initially facing away from the lens. The Corgi abruptly turns its head to face the camera, triggering a rapid crash zoom directly into its shiny black bead eyes as it squints suspiciously and its woolly brow furrows deeply. The background features blurred shelves stocked with colorful red and blue products under bright, linear fluorescent ceiling lights. The lighting is soft and diffuse, highlighting the fuzzy, fibrous texture of the felted wool against the bokeh of the supermarket. [00:00 through 00:02] looking away, then turns towards camera [00:02 through 00:03] rapidly crash zoom to closely frame his eyes [00:03 through 00:04] black eyes squint [00:04 through 00:07] brow slowly furrows

As seen in the example above, we recommend pairing timestamp prompts with a natural language prompt for the most control.

For accuracy, consider how long an action would reasonably take to complete. For example, if you're generating a video of someone walking across a room, setting timestamps 00:00 through 00:00.5 wouldn't give enough time for the action to unfold naturally—a more realistic timeframe might be 00:00 to 00:03.

Text to Video allows you to generate footage directly without first creating an image. This works well when exact character or scene consistency isn't the priority—like creating B-roll, stock effects, or background plates.

If a component isn't present or executed in an initial generation, try iterating your prompt to reinforce it through natural language. Iteration is a normal part of the process when working with generative media, much like the drafting phases of other creative processes.

In example, if we didn't receive a high angle when prompting High angle of a koi fish pond for our first generation, we would reinforce the angle by iterating with High angle looking down at a koi fish pond.