Seedance 2.5 can use iPhone-captured references in an image to video workflow, but a camera roll is not automatically a useful reference set. Mixed lighting, inconsistent angles and unexplained clips make it hard for a creative collaborator to understand what must stay accurate. A short preparation pass turns casual phone captures into clear production inputs.
Seedance 2.5 can use prompts and optional image, video and audio references to guide a concept. Prepare the captures on your phone, then use the supported files in the browser-based creator. Available input rules and file limits should always be checked in the creator interface before upload.
Decide what each phone capture must explain
Start with a shot list. For a small product, capture the front, back, side profile, controls, important texture and the way a person handles it. If scale matters, include a neutral comparison or a hand in one frame. For a location, record the entrance, main working area, key background and natural movement through the space.
Each asset should answer a question. “Front label reference” is useful. “IMG_4821” is not. Rename or log the captures according to their role before anyone builds the prompt.
Capture clean references on the iPhone
Wipe the lens, use stable framing and avoid digital zoom when a closer position is possible. Keep the product and background lighting consistent across stills. Locking exposure or focus can help prevent a sequence of reference shots from changing brightness, but the exact camera choices depend on the phone model and capture conditions.
For motion, record a short clip of the action at a comfortable speed. A reference clip can show how a lid opens or how a hand approaches the product. It does not need cinematic editing; it needs to make the action unambiguous.
Select and label the approved assets
Remove duplicates, blurred frames and images that contradict the chosen look. Keep the smallest set that explains the subject, action and setting. Record whether an item guides identity, product detail, composition, movement, color or sound.
Confirm that the team has permission to use every visible logo, person, artwork and audio recording. Avoid uploading private notifications, contact details or customer information that may appear in a screenshot. In the creator, upload only the selected files that the current interface accepts, assign each one a clear role, and check any rejected file against the displayed format and size requirements before trying again.
Build the concept around a simple sequence
For an image-to-video concept, describe the subject first, followed by the action, camera, environment, timing and intended use. A hypothetical example for a travel mug might open on the mug beside a bag, show a hand securing the lid, then follow the mug into a morning commute setting.
For a first image-to-video attempt, choose one clear product photo. For a workflow combining several types of reference, choose the corresponding media-reference mode and follow its input requirements. Do not assume that the two modes accept the same asset combination.
A small reference list might contain:
mug-front: product color, silhouette and logo position.
mug-side: handle shape and proportions.
lid-action: movement reference, if the selected mode accepts video.
kitchen-setting: composition and morning-light direction.
Use the reference controls or tags shown in the creator to connect each selected asset to its role. Then choose the output settings and review the credit estimate before generating.
Example prompt: “Use the selected mug reference for the product’s appearance. The mug stands upright beside a canvas bag on a kitchen counter. A hand gently places the closed mug closer to the bag. Slow lateral camera movement, soft morning window light and quiet room ambience. Keep the handle and lid visible, with no added screen text.”
This starts with a manageable product scene. Compare the result with the original photo before adding a more complex transition or additional action.
Review the concept beyond visual appeal
Check whether the product shape, controls and branding remain accurate. Watch hand contact, reflections, background continuity and the timing of the key action. Inspect generated text separately. Compare the concept with the shot list rather than approving it simply because it looks polished.
If one element needs revision, describe that element precisely and compare the new result with the prior version. Regeneration may change other details, even when the instruction changes only one variable.
Decide what belongs in real production
Use the concept to identify shots that can remain illustrative and shots that require real footage. A stylized transition may work as generated material, while a close-up demonstrating a safety feature may need a real product and controlled shoot.
The value of the iPhone video workflow lies in preparation. Clear captures, assigned reference roles and a defined review standard give the generated concept a practical purpose: helping the team make production decisions earlier.
