AI Learn
Runway · Practical Tutorial

Build a 20–30 second product ad in Runway from multiple short shots

Storyboard the ad as several short generations, lock product and visual continuity with references, then edit the clips into one coherent sequence instead of asking for a single long generation.

Reviewed September 20, 2026Independent tutorialFeatures may vary
Not official support. AI Learn is an independent education site. Check the provider’s current documentation for plan-specific limits, pricing and feature availability.
Before you start

Longer AI videos are more controllable when you design them shot by shot. Plan the story first, create consistent product or character plates, and reserve editing, pacing, sound and text for the assembly stage.

1. Write a six-shot storyboard where every shot has one job in the story

Write a six-shot storyboard where each shot has one narrative job—hook, reveal, product proof, use case, emotional beat or close. This prevents six visually attractive clips from becoming a storyless montage.

2. Create clean product or character reference images and reuse them across shots

Create a clean product or character reference before video generation and reuse it across shots. Keep the reference free of unnecessary background clutter so identity is easier to preserve.

3. Generate each shot separately with one dominant subject motion and one camera behavior

Generate each shot with one primary subject motion and one camera motion. Short prompts with clear motion hierarchy are easier to iterate than a sentence containing several conflicting actions.

4. Review adjacent clips for identity, lighting, direction of movement and screen position

Place adjacent clips side by side and inspect identity, lighting direction, travel direction, screen position and scale. Fix continuity problems before adding sound or final copy, because editing polish cannot hide a broken cut.

5. Assemble a rough cut, trim for rhythm, then add sound, copy and final color adjustments

Assemble a rough cut first, adjust shot length and rhythm, then add voice/music, brand copy and color treatment. Export only after checking that the story remains understandable with the sound muted.

Try this prompt

Create shot [#] of a six-shot product ad. Reference asset: [product/character]. Story job: [what the viewer learns]. Subject motion: [one action]. Camera: [one movement]. Duration target: [5–8s]. Preserve product shape, color and branding from the reference. Do not add text; copy will be added in editing.

Common mistakes

Trying to generate the whole ad in one prompt; changing product reference halfway through; putting several actions and camera moves into every shot; adding tiny AI-generated copy inside frames; checking clips individually but not adjacent continuity.

Finish check

The rough cut tells a clear story in 20–30 seconds, the product remains recognizable across shots, motion cuts together cleanly, and text/audio are added in the editing stage where they can be controlled.

开始之前

AI 长视频按镜头设计通常更可控。先规划故事,建立稳定的产品/角色基础参考,再把节奏、声音、字幕和最终连续性留到剪辑阶段处理。

1. 先写一个大约六镜头的分镜,每个镜头只承担一个叙事任务

先写 6 镜头 Storyboard,并让每个镜头只承担一个叙事职责,例如 Hook、揭示、产品证明、使用场景、情绪节点或收尾。这样可以避免 6 个好看片段最后只是没有故事的拼接。

2. 准备干净的产品或角色参考图,并在多个镜头中重复使用

生成视频前先制作干净的产品或角色 Reference,并在各镜头重复使用。参考图尽量减少无关背景和复杂元素,这样主体身份更容易跨镜头保持一致。

3. 每个镜头单独生成,只安排一个主要主体动作和一个镜头运动

每个镜头只描述一个主要主体动作和一个镜头运动。动作层级越清楚,越容易迭代;把多个互相冲突的动作塞进一句 Prompt 通常会降低可控性。

4. 相邻镜头逐个检查产品/角色身份、光线、运动方向和画面位置是否连续

把相邻 Clip 并排检查主体身份、光线方向、移动方向、画面位置和比例。连续性问题要在加声音和最终文案之前修,因为后期包装无法掩盖断裂的剪辑。

5. 先做 Rough Cut,按节奏裁剪,再加入声音、文案和最终色彩调整

先拼 Rough Cut,调整镜头时长和节奏,再添加配音/音乐、品牌文案和统一调色。最终导出前把声音关掉看一遍,确认纯画面仍能讲清故事。

可以直接套用的提示词

生成六镜头产品广告中的第 [#] 镜。参考资产:[产品/角色]。本镜头的叙事任务:[观众要知道什么]。主体动作:[一个动作]。镜头:[一个运动]。目标时长:[5–8 秒]。保持参考中的产品形状、颜色和品牌特征,不要生成文字,文案留到剪辑阶段加入。

常见错误

一条提示词直接生成整支广告;做到一半换产品参考;每个镜头塞进多个动作和镜头运动;让生成模型在画面里做大量小文字;只单独看每个片段,不检查前后连续性。

完成检查

Rough Cut 能在 20–30 秒内讲清故事,产品跨镜头仍可辨认,运动衔接自然,文字和声音都在可控的剪辑阶段加入。

開始之前

AI 長影片按鏡頭設計通常更可控。先規劃故事,建立穩定的產品/角色基礎參考,再把節奏、聲音、字幕和最終連續性留到剪輯階段處理。

1. 先寫一個大約六鏡頭的分鏡,每個鏡頭只承擔一個敘事任務

先寫 6 鏡頭 Storyboard,並讓每個鏡頭只承擔一個敘事職責,例如 Hook、揭示、產品證明、使用場景、情緒節點或收尾。這樣可以避免 6 個好看片段最後只是沒有故事的拼接。

2. 準備乾淨的產品或角色參考圖,並在多個鏡頭中重復使用

生成視頻前先製作乾淨的產品或角色 Reference,並在各鏡頭重復使用。參考圖盡量減少無關背景和複雜元素,這樣主體身份更容易跨鏡頭保持一致。

3. 每個鏡頭單獨生成,只安排一個主要主體動作和一個鏡頭運動

每個鏡頭只描述一個主要主體動作和一個鏡頭運動。動作層級越清楚,越容易迭代;把多個互相衝突的動作塞進一句 Prompt 通常會降低可控性。

4. 相鄰鏡頭逐個檢查產品/角色身份、光線、運動方向和畫面位置是否連續

把相鄰 Clip 並排檢查主體身份、光線方向、移動方向、畫面位置和比例。連續性問題要在加聲音和最終文案之前修,因為後期包裝無法掩蓋斷裂的剪輯。

5. 先做 Rough Cut,按節奏裁剪,再加入聲音、文案和最終色彩調整

先拼 Rough Cut,調整鏡頭時長和節奏,再添加配音/音樂、品牌文案和統一調色。最終導出前把聲音關掉看一遍,確認純畫面仍能講清故事。

可以直接套用的提示詞

生成六鏡頭產品廣告中的第 [#] 鏡。參考資產:[產品/角色]。本鏡頭的敘事任務:[觀眾要知道甚麼]。主體動作:[一個動作]。鏡頭:[一個運動]。目標時長:[5–8 秒]。保持參考中的產品形狀、顏色和品牌特徵,不要生成文字,文案留到剪輯階段加入。

常見錯誤

一條提示詞直接生成整支廣告;做到一半換產品參考;每個鏡頭塞進多個動作和鏡頭運動;讓生成模型在畫面裡做大量小文字;只單獨看每個片段,不檢查前後連續性。

完成檢查

Rough Cut 能在 20–30 秒內講清故事,產品跨鏡頭仍可辨認,運動銜接自然,文字和聲音都在可控的剪輯階段加入。

Official references

Check current product details at the source

Related practical tutorials

Continue with another single-task workflow.

All tutorials →