From Product Image to Promo Video: The Input-to-Video Pipeline Explained
Most AI video tools give you a text box and hope for the best. RunwayMotion takes a different approach: we start with your fashion product images, guide them through a structured pipeline, and produce motion content that actually looks like your brand.
This article walks through exactly how the pipeline works — from what makes a good input image to how the system generates, extends, and finalizes your promo video.
Whether you are a fashion brand evaluating tools or a creative team looking to understand the workflow before diving in, this guide covers the essentials.
The Input: What Makes a Good Source Image
The quality of your source image directly determines the quality of your output. RunwayMotion works best with clean, well-lit fashion product photography — similar to what you would use for an ecommerce product page or a lookbook.
While a single front-facing product shot can work, the system is designed to accept multiple views: front, side/back, and detail close-ups. The more visual information you provide, the more consistent the generated video will be.
- • Use clean, well-lit product photography
- • Front view is essential; back and detail views improve consistency
- • Avoid cluttered backgrounds or multiple products in one shot
- • A scene reference image helps anchor the environment
Preview: Your First 8 Seconds
Every RunwayMotion job starts with a Preview — an 8-second motion clip generated from your input images. This is where you validate the look, feel, and consistency before committing more credits.
The Preview focuses on the front-facing view of your product. If you upload multiple views, the system prioritizes the front shot for the Preview while using the others as reference anchors for fabric, structure, and detail.
- • Preview uses 1 free daily generation (or 3 credits)
- • Front-facing product view is the primary driver
- • Additional views serve as reference anchors
- • Review the Preview before deciding to Extend
Extend: Building the Full 30-Second Video
Once you are satisfied with the Preview, you can Extend to a full video. The extension generates three additional 8-second segments that connect to your Preview: a turn sequence, a back-view sequence, and a detail close-up sequence.
Each segment uses a specific combination of your uploaded images as reference. The turn segment uses front + side views, the back segment uses back + detail views, and the detail segment focuses on close-up texture and craftsmanship.
The four segments (Preview + Turn + Back + Detail) are then stitched together into a single, seamless 30-second video.
Structured Inputs: The Four Locks
To maintain visual consistency across all segments, RunwayMotion uses a "four-lock" system: character lock (the model/person stays the same), garment lock (the clothing structure is preserved), scene lock (the environment remains consistent), and detail lock (textures and trims are accurate).
This is why uploading multiple views matters. A single front shot gives the AI less information to maintain these locks. Multiple views act as additional constraints that improve consistency across the entire video.
- • Character Lock: Keep the same model across all segments
- • Garment Lock: Preserve clothing structure and drape
- • Scene Lock: Maintain consistent environment and lighting
- • Detail Lock: Ensure textures, trims, and craftsmanship stay accurate
The Role of Scene Reference Images
A scene reference image is an optional but powerful input. It tells the generation system what kind of environment, lighting, and mood you want — runway, studio, editorial, lifestyle.
Without a scene reference, the system defaults to a clean studio look. With one, you can guide the output toward a specific brand aesthetic. This is especially useful for campaign-specific content where the environment is part of the brand story.
Real Examples: Studio Product Motion, Social Ad Creative, and Evergreen Brand Assets
RunwayMotion ships with real, verifiable examples across three use-case categories: Promo (studio product motion and editorial teasers), Ads (social ad creative and retargeting variations), and Brand Assets (evergreen brand content and catalog motion).
Each example includes the original input image, the generated ending frame, and the actual video output. You can view these on the Examples page to understand what the pipeline produces before you create your own.
Visit the examples page to browse the full gallery.
Getting Started: Your First Run
Ready to try it? Head to the Studio, upload at least one product image (more views recommended), and click Generate. Your first Preview each day is free, so you can test the pipeline without any commitment.
Once you are happy with your Preview, you can extend to a full 30-second video. The entire process — from upload to final video — typically completes in a few minutes.