A useful brief starts with the right photo, a clear action and a manageable shot. Build a focused first version, then change one thing at a time.
VideoTrendKit prepares effect-specific production plans and local photo previews. New generation is paused until each effect workflow is validated. Earlier technical samples verify the account and credit pipeline, not the named video effects.
Use a sharp, well-lit JPEG, PNG or WebP up to 10 MB in the recipe studio. For body movement, leave space around the entire body and keep hands visible. For a performance close-up, make the face readable without sunglasses or heavy occlusion.
Prepare separate labelled images for each role. Decide who appears on the left, who moves first and whether anyone crosses in front. A single still photo does not define the timing of a duet.
Use a clip you have permission to use. Prefer one continuous shot with a clear performer, stable camera and visible limbs. A trend name or song title describes an idea; it does not supply the original choreography.
Decide between an original recording, licensed audio or a silent first test. A web link is not proof of permission. Our local recipe tools do not download or extract media from a link.
Local photo previews stay in this browser. Copying or downloading a recipe includes text, not the photos.
Plan a subject image plus a permitted action clip. Keep body framing compatible between the photo and reference. Start with one performer and one short action; the reference carries timing and movement.
Find motion guides ↗Define what stays: setting, camera direction, props and background action. Define what changes: the lead's appearance. Review face detail, body proportions and interactions through the entire shot.
Find scene guides ↗Write a start, action and end for each shot. Keep a shared character description, outfit and lighting. Multiple people, sudden transformations and sports interactions may need separate renders and editing.
Find advanced guides ↗A remake follows a reference performance. Original rap needs your own words and audio. A music video needs a sequence of shots and a soundtrack plan. These are different inputs and workflows.
Choose a rap workflow ↗Write the subject, one action, the setting, camera framing, light and what must remain consistent. A motion-reference workflow also needs the actual reference clip; a longer prompt cannot replace it. Avoid changing the person, camera and background all at once in a first test.
Subject: one adult performer from my reference photo. Action: take two relaxed steps, then pause facing the camera. Setting: a simple studio with an uncluttered background. Camera: a stable full-body frame with room around both hands. Light: soft side light, consistent throughout the shot. Keep: the same face, outfit and body proportions. Sound: silent first test. Use only permitted audio later.
This describes creative direction, not a tested result or a guarantee. Check the chosen provider's supported inputs and settings before submitting it elsewhere.
Build an editable recipeTry a sharper reference with a clear face. Reduce extreme head turns, rapid camera moves and heavy occlusion in the first test. Keep the same reference across shots. These changes may help, but identity consistency varies by model.
Give the entire body space in frame. Simplify overlapping gestures and avoid hands covering the face. Shorten the action and inspect several frames before spending credits on a longer render.
Check that the provider supports motion-reference input and that the correct clip is attached. Match subject framing, use one clear performer and remove unnecessary scene changes. A song name alone cannot encode choreography.
Label each reference and specify positions, entry order and interaction. Start with a wider shot and less crossing. If the chosen model does not support reliable multi-person control, plan separate shots and edit them together.
Confirm whether the chosen workflow actually supports audio-driven performance or lip sync. Keep vocals clear and start with a short excerpt you own. Generic image-to-video and motion transfer do not automatically provide word-level synchronization.
Reuse character descriptions, references, wardrobe and light. Write the last pose of one shot as the start of the next. Check transitions in an editor instead of assuming independent generations will join seamlessly.
Reading guides and planning a recipe do not require an account. The account preview lets you test email login and credit bookkeeping separately from video generation.
The sandbox grants 20 signup credits once, valid for 7 days. A manual daily check-in grants 5 credits once per UTC day, valid for 7 days. The balance distinguishes expiring rewards from permanent pack credits.
In the ledger sandbox, a technical failure releases reserved credits exactly once. A source reward that has already expired gets a limited refund window. This differs from disliking a successfully generated result, and it is not a cash refund policy.
Pack selection and top-ups in this preview are simulations. No payment is charged and no video provider is called. The proposed credit estimates vary by configuration and do not prove video quality.