Create an AI commercial from a phone photo

Step by step

Run the workflow

1. 01

   Capture a clean source frame

   Create the clear source image used by the rest of the workflow.

   1. Set a three-to-five-second timer and take a still photo, or record a five-second phone video.
   2. For video, use 4K at 60 fps when your phone supports it, then select the sharpest frame.
   3. Keep the subject clear and fully visible, with enough light to preserve important details.
   4. Export the selected frame as a PNG before adding filters or effects.

   

   Common failure modes

   • A blurred or compressed source causes unstable faces, hands, text, or product details.
   • Reflections, private documents, bystanders, or protected branding remain visible in the background.
2. 02

   Plan with DirectorGPT

   Turn the goal, tone, and duration into one selected concept and a complete shot plan.

   • Click here to open SIA's Video Prompt Director (opens in a new tab)Direct link supplied on page 5 of the original PDF.

   1. Open SIA's Video Prompt Director and select 'I want to create a video'.
   2. Answer Goal, Tone, and Duration in short bullet points.
   3. Review the three narrative paths it proposes.
   4. Choose the concept that best matches the brand and audience.
   5. Send the concept selection to receive the full shot list, script, camera direction, and Kling-ready prompts.

   

   

   

   

   

   

   Common failure modes

   • The selected concept does not match the original goal, brand, or audience.
   • The plan introduces claims, people, characters, or assets that are not approved.
3. 03

   Complete the image stage

   Prepare the approved still that the video stage will use.

   • Click here to open Google Flow (opens in a new tab)Official tool link. The PDF names Flow but does not include a Flow link or walkthrough.

   1. The source protocol names Google Flow for image generation, but does not include a detailed Flow walkthrough.
   2. If the selected concept needs a generated or refined still, open Flow, choose Image, set the aspect ratio, and add the approved source as a reference when needed.
   3. Generate the smallest useful set, review the options, and export only the approved image.
   4. If the original source frame already matches the shot plan, keep it and continue to Kling.

   Common failure modes

   • An unnecessary image-generation pass changes details that were already correct in the source.
   • A generated still alters identity, packaging, logos, text, proportions, or product behaviour.
4. 04

   Generate the video in Kling

   Map each approved shot prompt to the correct source reference and generate a consistent first cut.

   • Click here to open Kling AI (opens in a new tab)Affiliate link supplied on page 9 of the original PDF. The Workflow Lab may earn a commission.

   1. Open Kling AI, choose Video 3.0 Omni, and enable the custom multi-shot workflow if it is available on your account.
   2. Upload only the approved reference images needed for this commercial.
   3. Paste the approved prompts into the matching shots in order.
   4. Use the @ selector in each shot prompt to map the correct uploaded subject or product image. Confirm the mapping before generation.
   5. Choose the resolution, duration, audio setting, and number of generations deliberately. Start with a low-cost test before requesting more variants.
   6. Review the complete result once for story and once frame by frame for identity, hands, text, product shape, background, and continuity.

   

   

   

   

   

   

   Common failure modes

   • A prompt is mapped to the wrong reference image, causing identity or product drift.
   • More generations are purchased before the shot mapping and prompt structure have been tested.
   • A visually attractive result changes packaging, logos, product function, or a person's appearance.
5. 05

   Polish the edit in CapCut

   Use editing, pacing, and colour to finish the commercial without hiding generation errors.

   • Click here to open CapCut (opens in a new tab)Affiliate link supplied on page 12 of the original PDF. The Workflow Lab may earn a commission.

   1. Import the approved clips into CapCut and arrange them in the final narrative order.
   2. Use Speed > Curve only where a speed ramp improves rhythm or emphasises a planned moment.
   3. Apply glow sparingly if it supports the intended look. Effect names and availability can vary by version.
   4. Use the Adjustment controls to correct exposure, white balance, contrast, and colour consistency across shots.
   5. Add only licensed or approved music, voice, captions, logos, and calls to action.
   6. Watch the complete export with sound, then check every frame, detail, claim, and permission before publishing.

   

   

   

   

   

   

   

   

   

   

   

   

   

   

   Common failure modes

   • Effects and colour grading make the edit busier but do not fix weak pacing or unclear messaging.
   • Speed changes create jitter, unreadable text, or distorted audio.
   • The final export is published without checking captions, claims, permissions, or platform specifications.

Before you use it

Final checks

• Check every frame, detail, and claim.
• Clear all people, voices, music, locations, and brand assets.
• Never upload confidential material without approval.