First-frame/last-frame (also called start-end frame) is the most powerful control technique in AI video: give the model two images — where the shot begins and where it ends — and it generates the journey between them. This guide explains the workflow, the prompt patterns that make transitions seamless, and how to practice it free with single-image animation in your browser.
Transform static photos into dynamic, lifelike videos with advanced AI animation technology
Choose a mode to get started
Drag & Drop, Paste, or Click to Upload
PNG, JPG, JPEG, WEBP, Max 24 MB
Why directors, real-estate creators, and VFX artists all use the two-frame trick.
Renovated rooms, seasonal changes, aging, day-to-night, old-to-new — lock frame one as 'before' and frame two as 'after', and the AI renders the transformation between them.
Single-image generation is a roll of the dice about where motion ends. Two-frame control means you choose the final composition precisely — essential for storyboards and ad endings.
The viral trick: use a real video's first frame as the start, generate to a styled end frame, then cut back to real footage. The shared frame makes the AI segment blend invisibly.
The golden rule: both frames must share the same camera angle, framing, and perspective. A tripod-like match gives a smooth morph; mismatched angles give warped, melting transitions.
With start/end set, your prompt should describe the journey — 'smooth continuous transformation, no cuts' — not re-describe the images. Over-prompting is the #1 cause of crossfades instead of motion.
The playground above animates a single image with full camera language — the core skill. Master prompt-driven motion first, then apply it to two-frame workflows in any tool that supports them.
Shoot or generate your starting image — clear subject, deliberate composition, steady angle.
Edit or AI-generate the destination image from the exact same angle. Same framing, changed content.
'Continuous smooth transformation from the first frame to the last, no camera shake, no cuts, coherent motion throughout'.
Review the in-between motion — iterate until the journey feels intentional, not just the endpoints.
Clear answers for people comparing AI image to video tools, old photo animation, and photo to video workflows.
It's a control technique where you provide two reference images — the video's opening frame and its final frame — and the AI model generates the motion connecting them. It's sometimes called start-end frame or keyframe video generation.
Single-image generation decides the destination on its own. Two frames give you exact control over where the shot ends — critical for before/after reveals, brand endings, and stitching AI footage with real footage.
Kling, Vidu, and several workflow tools (ComfyUI setups, DomoAI and others) offer first/last-frame modes. The technique matters more than the tool: matched angles plus transition-focused prompts.
Usually over-prompting or mismatched frames. Keep both images at identical angles and framing, and prompt the journey ('smooth continuous transformation, no dissolve') rather than describing the images again.
Same camera position, same focal length, same framing — only the content changes. Think tripod time-lapse, not handheld walk. The closer the match, the more magical the morph.
Yes — the playground above animates a single image with full cinematic control, which builds the prompt skills the two-frame workflow depends on. Free credits included, watermark-free exports.
The playground above is optimized for single-image animation. The guide's prompt patterns and matching rules transfer directly to any tool with native start-end frame support.
Build your prompt-directing skills in the playground, then apply them to two-frame workflows anywhere.
