Home / Image to video / Inputs
Image to video · input workflow
A single-image animation asks one source image to carry the whole clip. A storyboard gives each beat its own source frame. Pick the input workflow before you write a motion prompt or spend a generation credit.
Use one image when the clip is one subject in one setting. Use a storyboard when the viewer must see a new angle, a new place, or a deliberate sequence of events.
A two-minute input check
Choose the closest statement. This does not predict a specific service’s result; it helps you decide which source material to prepare.
Prepare one strong source image. Keep the subject, background, and intended motion in the same composition. A one-image brief is usually the clearer starting point for one contained moment.
The decision behind the prompt
Start here for an avatar, old photo, product close-up, or a short moment where the setting should stay recognizably the same.
Start here when the clip needs the viewer to understand a transition: wide shot to close-up, product detail to use case, or one event followed by another.
The missing-evidence test
Finish this sentence before you make the prompt: “The viewer must see ___.” If the blank is already visible inside the first composition, keep one image and direct the motion. If it is a new angle, a changed state, or the result of an earlier action, make that proof a separate source frame.
Ask whether a crop, a camera push, or a small movement can reveal the fact without leaving the original view. If yes, the first image is still doing its job.
When the first image cannot prove the fact, name the exact thing the next frame must establish: a different side, a changed object, or a consequence.
Every added frame should answer one visible question. If it does not add proof, remove it and keep the input set smaller.
A practical production rule
A prompt can direct motion, but it is a weak substitute for source material the audience must actually see. Build the minimum input set first, then write a prompt that only asks for change within that set.
If the outcome is a single continuous moment, choose one image. If it relies on a cut, angle change, or ordered action, list the beats and prepare a frame for each.
For a storyboard, compare the subject, clothing, product state, lighting direction, and camera side before rendering. This prevents a later frame from quietly changing the story.
Use a short test to answer one question, such as whether a face holds or whether a product label stays readable. Change one input variable at a time.
Why this distinction exists
A 2021 TokkingHeads product page described a one-photo minimum for its animation workflow and routed text, voice, and video into animated portrait outcomes. That historical input boundary is useful because it separates a one-photo character or portrait task from a multi-shot video plan.
This page does not treat a 2021 product description as a claim about current image-to-video services. It uses the archived constraint for a narrower lesson: when your plan only contains one photo, define the clip as a contained animation. When the story needs several views, prepare a storyboard instead of hoping a single image will imply every missing beat.
Source: TokkingHeads archived product page, 2021. The historical facts are rewritten and attributed; archived copy is not reproduced.
Limit: A storyboard is not a guarantee that a chosen generation service accepts multiple references or preserves continuity. Check that service’s current input controls and terms before uploading or paying for a render.
Questions before you generate
You can ask a generation service for more than one change, but a single source frame cannot show every viewpoint with the same control as separate reference frames. If a later shot matters to the story, prepare it.
It is a source image assigned to one visible beat, not a decorative mood board. A useful frame tells you what the viewer must see at that point in the sequence.
No. A contained one-subject moment can be clearer and faster with one well-composed image. Add frames only when the viewer needs a meaningful new view or ordered event.
No. It is a planning guide. It does not receive images, call a video model, or publish a clip.
