{"id":263802,"date":"2026-08-25T18:51:07","date_gmt":"2026-08-26T01:51:07","guid":{"rendered":"https:\/\/picsart.com\/blog\/?p=263802"},"modified":"2026-08-25T18:51:07","modified_gmt":"2026-08-26T01:51:07","slug":"wan-3-0-user-guide","status":"publish","type":"post","link":"https:\/\/picsart.com\/blog\/wan-3-0-user-guide\/","title":{"rendered":"WAN 3.0 user guide: everything the model can do + prompts"},"content":{"rendered":"<p>WAN 3.0 turns one prompt into a single shot that runs a full half minute. It reads far more than prompts, though. Images, video, sound, files and live web pages all work as source. Output comes back at 1080p in an adaptive ratio.<\/p>\n<p>Most of the craft sits in one place. A 30-second shot needs a plan, not a picture. Five seconds can be one image in motion. Thirty seconds has a start, a middle and an end. The prompt has to carry all three.<\/p>\n<p>This guide walks the whole job. How long to run the clip, what to feed the model, how to build the prompt, 12 prompts to copy, and what to change when a take misses. It all runs in <a href=\"https:\/\/picsart.com\/ai-playground\/\">Picsart AI Playground<\/a>.<\/p>\n<h2><span id=\"How_long_the_clip_should_run\">How long the clip should run<\/span><\/h2>\n<p>Start with length. It sets the shape of the rest.<\/p>\n<p>WAN 3.0 runs to 30 seconds in a single take. Earlier models in the line stopped at half that. The jump matters less as a number than as a unit of work. A whole spot, a full how-to or a complete scene now lands in one run. Nothing has to be stitched together after.<\/p>\n<p>You do not have to pick the number yourself. Smart duration control reads the action and the pace you write, then suggests a length that fits. A slow build gets room to breathe. A quick demo does not get padded out.<\/p>\n<p>Thirty seconds is also not a ceiling. Every clip you make can be extended, so a long run grows out of a take you like. That beats starting over on something longer and hoping the good parts survive.<\/p>\n<h2><span id=\"What_to_feed_WAN_30\">What to feed WAN 3.0<\/span><\/h2>\n<p>Six kinds of input work as source. Four are familiar from earlier WAN models: text, image, audio and video. Two are new, and they change where a job can start.<\/p>\n<ul>\n<li><strong>Documents.<\/strong> Upload a .doc, .pdf, .ppt or .xls straight into the run.<\/li>\n<li><strong>Web pages.<\/strong> Point the model at a URL. Product pages, news pieces, papers and brand sites all count.<\/li>\n<\/ul>\n<p>The model reads that source and builds from it. So the deck you made turns into the film, and the live product page turns into the launch video. Nobody has to retype the facts into a prompt box first.<\/p>\n<figure class=\"wp-block-table\">\n<table style=\"border-collapse: collapse; width: 100%; table-layout: auto;\">\n<tbody>\n<tr>\n<th style=\"border: 1px solid #333333; padding: 10px 14px; text-align: left; vertical-align: top; color: #ffffff; background: #000000; font-weight: bold; white-space: nowrap;\">Input<\/th>\n<th style=\"border: 1px solid #333333; padding: 10px 14px; text-align: left; vertical-align: top; color: #ffffff; background: #000000; font-weight: bold; white-space: nowrap;\">Good for<\/th>\n<th style=\"border: 1px solid #333333; padding: 10px 14px; text-align: left; vertical-align: top; color: #ffffff; background: #000000; font-weight: bold; white-space: nowrap;\">What the prompt still adds<\/th>\n<\/tr>\n<tr>\n<td style=\"border: 1px solid #333333; padding: 10px 14px; text-align: left; vertical-align: top; color: #ffffff; background: #141414;\">Text only<\/td>\n<td style=\"border: 1px solid #333333; padding: 10px 14px; text-align: left; vertical-align: top; color: #ffffff; background: #141414;\">Original scenes with no existing assets<\/td>\n<td style=\"border: 1px solid #333333; padding: 10px 14px; text-align: left; vertical-align: top; color: #ffffff; background: #141414;\">Everything: frame, beats, pacing, lock<\/td>\n<\/tr>\n<tr>\n<td style=\"border: 1px solid #333333; padding: 10px 14px; text-align: left; vertical-align: top; color: #ffffff; background: #141414;\">Image<\/td>\n<td style=\"border: 1px solid #333333; padding: 10px 14px; text-align: left; vertical-align: top; color: #ffffff; background: #141414;\">Holding one subject or product identical<\/td>\n<td style=\"border: 1px solid #333333; padding: 10px 14px; text-align: left; vertical-align: top; color: #ffffff; background: #141414;\">The action across the take<\/td>\n<\/tr>\n<tr>\n<td style=\"border: 1px solid #333333; padding: 10px 14px; text-align: left; vertical-align: top; color: #ffffff; background: #141414;\">Video<\/td>\n<td style=\"border: 1px solid #333333; padding: 10px 14px; text-align: left; vertical-align: top; color: #ffffff; background: #141414;\">Matching a grade, a lens and a camera feel<\/td>\n<td style=\"border: 1px solid #333333; padding: 10px 14px; text-align: left; vertical-align: top; color: #ffffff; background: #141414;\">A new subject and a new set of beats<\/td>\n<\/tr>\n<tr>\n<td style=\"border: 1px solid #333333; padding: 10px 14px; text-align: left; vertical-align: top; color: #ffffff; background: #141414;\">Audio<\/td>\n<td style=\"border: 1px solid #333333; padding: 10px 14px; text-align: left; vertical-align: top; color: #ffffff; background: #141414;\">Timing action to an existing track<\/td>\n<td style=\"border: 1px solid #333333; padding: 10px 14px; text-align: left; vertical-align: top; color: #ffffff; background: #141414;\">The scene and where the energy lands<\/td>\n<\/tr>\n<tr>\n<td style=\"border: 1px solid #333333; padding: 10px 14px; text-align: left; vertical-align: top; color: #ffffff; background: #141414;\">Document<\/td>\n<td style=\"border: 1px solid #333333; padding: 10px 14px; text-align: left; vertical-align: top; color: #ffffff; background: #141414;\">Turning a deck, sheet or slide set into a film<\/td>\n<td style=\"border: 1px solid #333333; padding: 10px 14px; text-align: left; vertical-align: top; color: #ffffff; background: #141414;\">Which parts to use, and in what order<\/td>\n<\/tr>\n<tr>\n<td style=\"border: 1px solid #333333; padding: 10px 14px; text-align: left; vertical-align: top; color: #ffffff; background: #141414;\">Web page URL<\/td>\n<td style=\"border: 1px solid #333333; padding: 10px 14px; text-align: left; vertical-align: top; color: #ffffff; background: #141414;\">Building from a page that already sells<\/td>\n<td style=\"border: 1px solid #333333; padding: 10px 14px; text-align: left; vertical-align: top; color: #ffffff; background: #141414;\">The framing, the pacing and the lock<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/figure>\n<h2><span id=\"The_four_parts_of_a_WAN_30_prompt\">The four parts of a WAN 3.0 prompt<\/span><\/h2>\n<p>A prompt for a 30-second take is a timeline, not a snapshot. Name one moment and the model has 29 seconds left to fill on its own. So every prompt in this guide follows the same four-part shape.<\/p>\n<h3>The opening frame<\/h3>\n<p>Name the subject, the setting and the light in the first line. The rest gets measured against this frame. So it wants hard nouns, not adjectives. &#8220;A matte black bottle on a wet slate ledge, dawn light from frame left&#8221; beats &#8220;a stylish product shot&#8221;.<\/p>\n<h3>The beats<\/h3>\n<p>List what happens, in order, with a rough time span against each one. Three beats across 30 seconds is a good rhythm. It also maps onto the way most spots work. WAN 3.0 is built to parse detailed, multi-part instructions, so the beat list does real work and never reads as padding.<\/p>\n<h3>The pacing<\/h3>\n<p>Say how fast the beats should land. A prompt that says &#8220;steady and unhurried&#8221; gets a different length back from one that says &#8220;brisk&#8221;. Naming the pace also gives the beat list a rhythm to sit against, so the timing never stays vague.<\/p>\n<h3>The consistency lock<\/h3>\n<p>Close by naming what must stay the same. Reference detail holds at pixel level across characters, objects, audio, scenes and styles. The lock tells the model which of those matter here. Product shape, a face, a colour and a light angle are the four most worth spelling out.<\/p>\n<h2><span id=\"WAN_30_prompts_to_copy\">WAN 3.0 prompts to copy<\/span><\/h2>\n<p>The 12 prompts below are grouped by what you feed the model. That one call shapes the rest. Swap in your own subject and keep the four parts intact.<\/p>\n<h3>Prompts that start from text<\/h3>\n<p>Text-only prompts carry the most weight, since nothing else anchors the output. Give the model a real frame, real beats and a real lock. That is enough to hold a full take.<\/p>\n<section class=\"try_prompt_block\" data-pulse-section-group=\"blog article\">\n    <h3 class=\"try_prompt_title\">Product spot, dawn light<\/h3>\n\n    <div class=\"try_prompt_card\" data-pulse-section=\"blog article_try prompt\">\n        <p class=\"try_prompt_text\" id=\"try-prompt-6a8e7a6b29217-text\">30-second continuous take, no cuts. OPENING FRAME: a matte black insulated water bottle standing on a wet slate ledge, dawn light from frame left, mist drifting behind it. BEAT 1, zero to eight seconds: the camera pushes in slowly as condensation beads and runs down the bottle. NEXT BEAT, eight to eighteen seconds: a hand enters from frame right, lifts the bottle, and turns it so the brushed lid catches the light. FINAL BEAT, eighteen to thirty seconds: the camera pulls back to reveal the ledge overlooking a valley as the sun clears the ridge. PACING, steady and unhurried throughout. CONSISTENCY, ABSOLUTE: the bottle shape, finish and lid proportions stay identical in every frame; the light direction never jumps.<\/p>\n\n        <button type=\"button\"\n                class=\"try_prompt_copy\"\n                data-copy-target=\"#try-prompt-6a8e7a6b29217-text\"\n                aria-label=\"Copy prompt\"\n                title=\"Copy prompt\"\n                data-pulse-name=\"try prompt - copy\">\n            <img class=\"try_prompt_copy_icon\"\n                 src=\"https:\/\/cdn-cms-uploads.picsart.com\/cms-uploads\/79a9b2ac-f4c9-436b-b388-a05a484adf00.png\"\n                 alt=\"\"\n                 width=\"20\" height=\"20\"\n                 loading=\"lazy\" decoding=\"async\">\n            <svg class=\"try_prompt_check_icon\" viewBox=\"0 0 24 24\" width=\"20\" height=\"20\" fill=\"none\" stroke=\"currentColor\" stroke-width=\"2.5\" stroke-linecap=\"round\" stroke-linejoin=\"round\" aria-hidden=\"true\">\n                <polyline points=\"5,12 10,17 19,7\"><\/polyline>\n            <\/svg>\n            <span class=\"try_prompt_copy_tooltip\" role=\"status\" aria-live=\"polite\">Copied<\/span>\n        <\/button>\n    <\/div>\n\n    <\/section>\n\n<h3>An explainer with words on screen<\/h3>\n<p>How-to clips are where on-screen text matters. Spell out every word you want, just as it should appear. Naming the text in loose terms will not get it there. On-screen text is one of the model&#8217;s stronger areas on dense, fact-heavy clips, so it is worth leaning on.<\/p>\n<section class=\"try_prompt_block\" data-pulse-section-group=\"blog article\">\n    <h3 class=\"try_prompt_title\">Chart explainer, exact labels<\/h3>\n\n    <div class=\"try_prompt_card\" data-pulse-section=\"blog article_try prompt\">\n        <p class=\"try_prompt_text\" id=\"try-prompt-6a8e7a6b2923a-text\">30-second continuous take. OPENING FRAME: a clean studio desk, soft top light, a tablet propped upright showing a simple bar chart. BEAT 1, zero to ten seconds: the camera drifts right as the words QUARTERLY REVENUE render across the top of the screen in a bold sans-serif. NEXT BEAT, ten to twenty seconds: three bars grow in sequence, each labelled Q1, Q2 and Q3, with the figures 1.2M, 1.8M and 2.4M appearing beside them. FINAL BEAT, twenty to thirty seconds: the camera settles as the line UP 40 PERCENT YEAR ON YEAR resolves beneath the chart. TEXT, EXACT: every label renders exactly as written here, legible and unwarped. PACING, calm and even.<\/p>\n\n        <button type=\"button\"\n                class=\"try_prompt_copy\"\n                data-copy-target=\"#try-prompt-6a8e7a6b2923a-text\"\n                aria-label=\"Copy prompt\"\n                title=\"Copy prompt\"\n                data-pulse-name=\"try prompt - copy\">\n            <img class=\"try_prompt_copy_icon\"\n                 src=\"https:\/\/cdn-cms-uploads.picsart.com\/cms-uploads\/79a9b2ac-f4c9-436b-b388-a05a484adf00.png\"\n                 alt=\"\"\n                 width=\"20\" height=\"20\"\n                 loading=\"lazy\" decoding=\"async\">\n            <svg class=\"try_prompt_check_icon\" viewBox=\"0 0 24 24\" width=\"20\" height=\"20\" fill=\"none\" stroke=\"currentColor\" stroke-width=\"2.5\" stroke-linecap=\"round\" stroke-linejoin=\"round\" aria-hidden=\"true\">\n                <polyline points=\"5,12 10,17 19,7\"><\/polyline>\n            <\/svg>\n            <span class=\"try_prompt_copy_tooltip\" role=\"status\" aria-live=\"polite\">Copied<\/span>\n        <\/button>\n    <\/div>\n\n    <\/section>\n\n<h3>A character scene with a turn in it<\/h3>\n<p>Motion and feeling both got stronger in this version. So a turn in mood belongs in the beats. Name the moment the change lands, and name what sets it off. One take is long enough to carry a small arc, which a five-second clip never is.<\/p>\n<section class=\"try_prompt_block\" data-pulse-section-group=\"blog article\">\n    <h3 class=\"try_prompt_title\">Character scene, single unbroken shot<\/h3>\n\n    <div class=\"try_prompt_card\" data-pulse-section=\"blog article_try prompt\">\n        <p class=\"try_prompt_text\" id=\"try-prompt-6a8e7a6b2924c-text\">30-second continuous take, single unbroken shot. OPENING FRAME: a woman in a mustard raincoat stands at a tram stop at dusk, rain falling, shop signs glowing behind her. BEAT 1, zero to nine seconds: she checks her watch, then looks down the empty track, breath visible. NEXT BEAT, nine to nineteen seconds: a tram arrives and its headlights sweep across her face; she straightens and her expression shifts from impatience to relief. FINAL BEAT, nineteen to thirty seconds: she steps aboard and the camera holds on the closing door as the tram pulls away. CONSISTENCY, ABSOLUTE: face, hair, and the exact shade and cut of the raincoat stay identical from the first frame to the last. PACING, unhurried, with the emotional turn landing on the headlight sweep.<\/p>\n\n        <button type=\"button\"\n                class=\"try_prompt_copy\"\n                data-copy-target=\"#try-prompt-6a8e7a6b2924c-text\"\n                aria-label=\"Copy prompt\"\n                title=\"Copy prompt\"\n                data-pulse-name=\"try prompt - copy\">\n            <img class=\"try_prompt_copy_icon\"\n                 src=\"https:\/\/cdn-cms-uploads.picsart.com\/cms-uploads\/79a9b2ac-f4c9-436b-b388-a05a484adf00.png\"\n                 alt=\"\"\n                 width=\"20\" height=\"20\"\n                 loading=\"lazy\" decoding=\"async\">\n            <svg class=\"try_prompt_check_icon\" viewBox=\"0 0 24 24\" width=\"20\" height=\"20\" fill=\"none\" stroke=\"currentColor\" stroke-width=\"2.5\" stroke-linecap=\"round\" stroke-linejoin=\"round\" aria-hidden=\"true\">\n                <polyline points=\"5,12 10,17 19,7\"><\/polyline>\n            <\/svg>\n            <span class=\"try_prompt_copy_tooltip\" role=\"status\" aria-live=\"polite\">Copied<\/span>\n        <\/button>\n    <\/div>\n\n    <\/section>\n\n<h3>Prompts that start from a document<\/h3>\n<p>Upload the file, then write a prompt that says which parts to use and in what order. The prompt stops carrying the facts and starts carrying the plan. Tell the model to pull the numbers straight from the source.<\/p>\n<p><section class=\"try_prompt_block\" data-pulse-section-group=\"blog article\">\n    <h3 class=\"try_prompt_title\">Deck to explainer<\/h3>\n\n    <div class=\"try_prompt_card\" data-pulse-section=\"blog article_try prompt\">\n        <p class=\"try_prompt_text\" id=\"try-prompt-6a8e7a6b29259-text\">Use the attached deck as the source. Build a 30-second continuous explainer that follows the deck in order. BEAT 1, zero to eight seconds: open on the deck&#8217;s headline claim rendered as clean on-screen type over a plain studio background. NEXT BEAT, eight to twenty seconds: move through the three supporting figures from the deck, one at a time, each rendered legibly beside a simple matching visual. FINAL BEAT, twenty to thirty seconds: close on the deck&#8217;s final call to action in the same type treatment. TEXT, EXACT: pull every number and label verbatim from the deck and invent nothing. PACING, steady, one idea at a time.<\/p>\n\n        <button type=\"button\"\n                class=\"try_prompt_copy\"\n                data-copy-target=\"#try-prompt-6a8e7a6b29259-text\"\n                aria-label=\"Copy prompt\"\n                title=\"Copy prompt\"\n                data-pulse-name=\"try prompt - copy\">\n            <img class=\"try_prompt_copy_icon\"\n                 src=\"https:\/\/cdn-cms-uploads.picsart.com\/cms-uploads\/79a9b2ac-f4c9-436b-b388-a05a484adf00.png\"\n                 alt=\"\"\n                 width=\"20\" height=\"20\"\n                 loading=\"lazy\" decoding=\"async\">\n            <svg class=\"try_prompt_check_icon\" viewBox=\"0 0 24 24\" width=\"20\" height=\"20\" fill=\"none\" stroke=\"currentColor\" stroke-width=\"2.5\" stroke-linecap=\"round\" stroke-linejoin=\"round\" aria-hidden=\"true\">\n                <polyline points=\"5,12 10,17 19,7\"><\/polyline>\n            <\/svg>\n            <span class=\"try_prompt_copy_tooltip\" role=\"status\" aria-live=\"polite\">Copied<\/span>\n        <\/button>\n    <\/div>\n\n    <\/section>\n<br \/>\n<section class=\"try_prompt_block\" data-pulse-section-group=\"blog article\">\n    <h3 class=\"try_prompt_title\">Spec sheet to product film<\/h3>\n\n    <div class=\"try_prompt_card\" data-pulse-section=\"blog article_try prompt\">\n        <p class=\"try_prompt_text\" id=\"try-prompt-6a8e7a6b29282-text\">Use the attached product spec sheet as the source. Build a 30-second continuous product film. OPENING FRAME: the product alone on a seamless white cyclorama under soft studio light. BEAT 1, zero to ten seconds: a slow orbit around the product as its three headline specifications from the sheet appear one at a time as on-screen type. NEXT BEAT, ten to twenty-two seconds: the camera moves in on the single feature the sheet describes at greatest length. FINAL BEAT, twenty-two to thirty seconds: pull back to the full product with the model name rendered exactly as written on the sheet. CONSISTENCY, ABSOLUTE: proportions, colour and materials match the spec sheet in every frame.<\/p>\n\n        <button type=\"button\"\n                class=\"try_prompt_copy\"\n                data-copy-target=\"#try-prompt-6a8e7a6b29282-text\"\n                aria-label=\"Copy prompt\"\n                title=\"Copy prompt\"\n                data-pulse-name=\"try prompt - copy\">\n            <img class=\"try_prompt_copy_icon\"\n                 src=\"https:\/\/cdn-cms-uploads.picsart.com\/cms-uploads\/79a9b2ac-f4c9-436b-b388-a05a484adf00.png\"\n                 alt=\"\"\n                 width=\"20\" height=\"20\"\n                 loading=\"lazy\" decoding=\"async\">\n            <svg class=\"try_prompt_check_icon\" viewBox=\"0 0 24 24\" width=\"20\" height=\"20\" fill=\"none\" stroke=\"currentColor\" stroke-width=\"2.5\" stroke-linecap=\"round\" stroke-linejoin=\"round\" aria-hidden=\"true\">\n                <polyline points=\"5,12 10,17 19,7\"><\/polyline>\n            <\/svg>\n            <span class=\"try_prompt_copy_tooltip\" role=\"status\" aria-live=\"polite\">Copied<\/span>\n        <\/button>\n    <\/div>\n\n    <\/section>\n<\/p>\n<h3>Prompts that start from a live web page<\/h3>\n<p>Point the model at a URL and it reads the page as source. So the copy, the product shots and the claims are already there. The prompt only has to set the framing, the order and the lock.<\/p>\n<p><section class=\"try_prompt_block\" data-pulse-section-group=\"blog article\">\n    <h3 class=\"try_prompt_title\">Product page URL to vertical spot<\/h3>\n\n    <div class=\"try_prompt_card\" data-pulse-section=\"blog article_try prompt\">\n        <p class=\"try_prompt_text\" id=\"try-prompt-6a8e7a6b2928e-text\">Use the linked product page as the source. Build a 30-second continuous social spot in vertical framing. BEAT 1, zero to nine seconds: open on the product in the setting the page shows it in, lit to match the page photography. NEXT BEAT, nine to twenty seconds: work through the two benefits the page leads with, each shown as an action rather than stated as text. FINAL BEAT, twenty to thirty seconds: close on the product with the page&#8217;s own headline rendered legibly across the lower third. CONSISTENCY, ABSOLUTE: colourway, packaging and logo placement match the page exactly. PACING, brisk and cut-free.<\/p>\n\n        <button type=\"button\"\n                class=\"try_prompt_copy\"\n                data-copy-target=\"#try-prompt-6a8e7a6b2928e-text\"\n                aria-label=\"Copy prompt\"\n                title=\"Copy prompt\"\n                data-pulse-name=\"try prompt - copy\">\n            <img class=\"try_prompt_copy_icon\"\n                 src=\"https:\/\/cdn-cms-uploads.picsart.com\/cms-uploads\/79a9b2ac-f4c9-436b-b388-a05a484adf00.png\"\n                 alt=\"\"\n                 width=\"20\" height=\"20\"\n                 loading=\"lazy\" decoding=\"async\">\n            <svg class=\"try_prompt_check_icon\" viewBox=\"0 0 24 24\" width=\"20\" height=\"20\" fill=\"none\" stroke=\"currentColor\" stroke-width=\"2.5\" stroke-linecap=\"round\" stroke-linejoin=\"round\" aria-hidden=\"true\">\n                <polyline points=\"5,12 10,17 19,7\"><\/polyline>\n            <\/svg>\n            <span class=\"try_prompt_copy_tooltip\" role=\"status\" aria-live=\"polite\">Copied<\/span>\n        <\/button>\n    <\/div>\n\n    <\/section>\n<br \/>\n<section class=\"try_prompt_block\" data-pulse-section-group=\"blog article\">\n    <h3 class=\"try_prompt_title\">Article URL to summary video<\/h3>\n\n    <div class=\"try_prompt_card\" data-pulse-section=\"blog article_try prompt\">\n        <p class=\"try_prompt_text\" id=\"try-prompt-6a8e7a6b292aa-text\">Use the linked article as the source. Build a 30-second continuous summary video. BEAT 1, zero to ten seconds: open on the article&#8217;s central claim rendered as on-screen type over a relevant establishing scene. NEXT BEAT, ten to twenty-two seconds: illustrate the two pieces of evidence the article rests on, each as its own short visual beat. FINAL BEAT, twenty-two to thirty seconds: close on the article&#8217;s conclusion in one legible line. TEXT, EXACT: quote figures and names verbatim from the article. PACING, measured and documentary.<\/p>\n\n        <button type=\"button\"\n                class=\"try_prompt_copy\"\n                data-copy-target=\"#try-prompt-6a8e7a6b292aa-text\"\n                aria-label=\"Copy prompt\"\n                title=\"Copy prompt\"\n                data-pulse-name=\"try prompt - copy\">\n            <img class=\"try_prompt_copy_icon\"\n                 src=\"https:\/\/cdn-cms-uploads.picsart.com\/cms-uploads\/79a9b2ac-f4c9-436b-b388-a05a484adf00.png\"\n                 alt=\"\"\n                 width=\"20\" height=\"20\"\n                 loading=\"lazy\" decoding=\"async\">\n            <svg class=\"try_prompt_check_icon\" viewBox=\"0 0 24 24\" width=\"20\" height=\"20\" fill=\"none\" stroke=\"currentColor\" stroke-width=\"2.5\" stroke-linecap=\"round\" stroke-linejoin=\"round\" aria-hidden=\"true\">\n                <polyline points=\"5,12 10,17 19,7\"><\/polyline>\n            <\/svg>\n            <span class=\"try_prompt_copy_tooltip\" role=\"status\" aria-live=\"polite\">Copied<\/span>\n        <\/button>\n    <\/div>\n\n    <\/section>\n<\/p>\n<h3>Prompts that start from an image or a video<\/h3>\n<p>Name the reference in the first line. Then say which of its traits the model has to protect. Pixel-level hold is what makes a reference worth using across a take this long.<\/p>\n<p><section class=\"try_prompt_block\" data-pulse-section-group=\"blog article\">\n    <h3 class=\"try_prompt_title\">Image reference, subject locked<\/h3>\n\n    <div class=\"try_prompt_card\" data-pulse-section=\"blog article_try prompt\">\n        <p class=\"try_prompt_text\" id=\"try-prompt-6a8e7a6b292bc-text\">Use the attached image as the reference for the subject. Build a 30-second continuous take that keeps that subject identical throughout. OPENING FRAME: the subject exactly as shown in the reference, placed in a sunlit kitchen with plants along the window. BEAT 1, zero to ten seconds: the subject moves through the room and the camera tracks alongside at the same height. NEXT BEAT, ten to twenty seconds: the subject stops at the counter and begins working with both hands. FINAL BEAT, twenty to thirty seconds: the camera settles into a slow push toward a three-quarter framing. CONSISTENCY, ABSOLUTE: face, proportions, wardrobe and colour match the reference image in every frame.<\/p>\n\n        <button type=\"button\"\n                class=\"try_prompt_copy\"\n                data-copy-target=\"#try-prompt-6a8e7a6b292bc-text\"\n                aria-label=\"Copy prompt\"\n                title=\"Copy prompt\"\n                data-pulse-name=\"try prompt - copy\">\n            <img class=\"try_prompt_copy_icon\"\n                 src=\"https:\/\/cdn-cms-uploads.picsart.com\/cms-uploads\/79a9b2ac-f4c9-436b-b388-a05a484adf00.png\"\n                 alt=\"\"\n                 width=\"20\" height=\"20\"\n                 loading=\"lazy\" decoding=\"async\">\n            <svg class=\"try_prompt_check_icon\" viewBox=\"0 0 24 24\" width=\"20\" height=\"20\" fill=\"none\" stroke=\"currentColor\" stroke-width=\"2.5\" stroke-linecap=\"round\" stroke-linejoin=\"round\" aria-hidden=\"true\">\n                <polyline points=\"5,12 10,17 19,7\"><\/polyline>\n            <\/svg>\n            <span class=\"try_prompt_copy_tooltip\" role=\"status\" aria-live=\"polite\">Copied<\/span>\n        <\/button>\n    <\/div>\n\n    <\/section>\n<br \/>\n<section class=\"try_prompt_block\" data-pulse-section-group=\"blog article\">\n    <h3 class=\"try_prompt_title\">Video reference, new subject<\/h3>\n\n    <div class=\"try_prompt_card\" data-pulse-section=\"blog article_try prompt\">\n        <p class=\"try_prompt_text\" id=\"try-prompt-6a8e7a6b292ca-text\">Use the attached video as the style reference. Build a new 30-second continuous take that matches its grade, grain and camera behaviour, with a different subject. OPENING FRAME: a cyclist waiting at a junction in a city at night. BEAT 1, zero to ten seconds: the lights change and the cyclist pulls away as the camera follows from behind. NEXT BEAT, ten to twenty-two seconds: the ride continues past lit storefronts, reflections moving across the frame. FINAL BEAT, twenty-two to thirty seconds: the cyclist slows at a canal and the camera drifts wide. CONSISTENCY, ABSOLUTE: hold the reference video&#8217;s colour treatment and lens character across every frame.<\/p>\n\n        <button type=\"button\"\n                class=\"try_prompt_copy\"\n                data-copy-target=\"#try-prompt-6a8e7a6b292ca-text\"\n                aria-label=\"Copy prompt\"\n                title=\"Copy prompt\"\n                data-pulse-name=\"try prompt - copy\">\n            <img class=\"try_prompt_copy_icon\"\n                 src=\"https:\/\/cdn-cms-uploads.picsart.com\/cms-uploads\/79a9b2ac-f4c9-436b-b388-a05a484adf00.png\"\n                 alt=\"\"\n                 width=\"20\" height=\"20\"\n                 loading=\"lazy\" decoding=\"async\">\n            <svg class=\"try_prompt_check_icon\" viewBox=\"0 0 24 24\" width=\"20\" height=\"20\" fill=\"none\" stroke=\"currentColor\" stroke-width=\"2.5\" stroke-linecap=\"round\" stroke-linejoin=\"round\" aria-hidden=\"true\">\n                <polyline points=\"5,12 10,17 19,7\"><\/polyline>\n            <\/svg>\n            <span class=\"try_prompt_copy_tooltip\" role=\"status\" aria-live=\"polite\">Copied<\/span>\n        <\/button>\n    <\/div>\n\n    <\/section>\n<\/p>\n<h3>Prompts that start from a sound<\/h3>\n<p>Audio works as a reference too, and it is the one that sets timing rather than looks. Give it a scene, then say where the energy should land.<\/p>\n<section class=\"try_prompt_block\" data-pulse-section-group=\"blog article\">\n    <h3 class=\"try_prompt_title\">Audio reference, action on the beat<\/h3>\n\n    <div class=\"try_prompt_card\" data-pulse-section=\"blog article_try prompt\">\n        <p class=\"try_prompt_text\" id=\"try-prompt-6a8e7a6b292dc-text\">Use the attached audio as the reference for pacing and mood. Build a 30-second continuous take whose action lands with the audio. OPENING FRAME: an empty rehearsal studio, low warm light, dust in the air. BEAT 1, zero to twelve seconds: a dancer enters and begins moving slowly as the camera circles at a fixed radius. NEXT BEAT, twelve to twenty-four seconds: the movement grows as the camera rises. FINAL BEAT, twenty-four to thirty seconds: the dancer holds a final position and the camera settles. PACING, follow the reference audio, and land every change in energy on screen with a change in the audio.<\/p>\n\n        <button type=\"button\"\n                class=\"try_prompt_copy\"\n                data-copy-target=\"#try-prompt-6a8e7a6b292dc-text\"\n                aria-label=\"Copy prompt\"\n                title=\"Copy prompt\"\n                data-pulse-name=\"try prompt - copy\">\n            <img class=\"try_prompt_copy_icon\"\n                 src=\"https:\/\/cdn-cms-uploads.picsart.com\/cms-uploads\/79a9b2ac-f4c9-436b-b388-a05a484adf00.png\"\n                 alt=\"\"\n                 width=\"20\" height=\"20\"\n                 loading=\"lazy\" decoding=\"async\">\n            <svg class=\"try_prompt_check_icon\" viewBox=\"0 0 24 24\" width=\"20\" height=\"20\" fill=\"none\" stroke=\"currentColor\" stroke-width=\"2.5\" stroke-linecap=\"round\" stroke-linejoin=\"round\" aria-hidden=\"true\">\n                <polyline points=\"5,12 10,17 19,7\"><\/polyline>\n            <\/svg>\n            <span class=\"try_prompt_copy_tooltip\" role=\"status\" aria-live=\"polite\">Copied<\/span>\n        <\/button>\n    <\/div>\n\n    <\/section>\n\n<h3>Prompts that extend a clip<\/h3>\n<p>The base clip already works. So the second prompt has two jobs. Say what happens next, and protect the join.<\/p>\n<p><section class=\"try_prompt_block\" data-pulse-section-group=\"blog article\">\n    <h3 class=\"try_prompt_title\">Extend, invisible join<\/h3>\n\n    <div class=\"try_prompt_card\" data-pulse-section=\"blog article_try prompt\">\n        <p class=\"try_prompt_text\" id=\"try-prompt-6a8e7a6b292e9-text\">Continue the attached clip forward without a cut. Match the grade, grain, framing and camera speed of the final second exactly. BEAT 1: the camera keeps moving in the same direction at the same rate for a further four seconds. NEXT BEAT: the subject completes the action already in progress. FINAL BEAT: the camera settles into a static wide and holds. CONSISTENCY, ABSOLUTE: subject, wardrobe, lighting direction and background architecture stay identical to the source clip. The join between the source and the continuation is invisible.<\/p>\n\n        <button type=\"button\"\n                class=\"try_prompt_copy\"\n                data-copy-target=\"#try-prompt-6a8e7a6b292e9-text\"\n                aria-label=\"Copy prompt\"\n                title=\"Copy prompt\"\n                data-pulse-name=\"try prompt - copy\">\n            <img class=\"try_prompt_copy_icon\"\n                 src=\"https:\/\/cdn-cms-uploads.picsart.com\/cms-uploads\/79a9b2ac-f4c9-436b-b388-a05a484adf00.png\"\n                 alt=\"\"\n                 width=\"20\" height=\"20\"\n                 loading=\"lazy\" decoding=\"async\">\n            <svg class=\"try_prompt_check_icon\" viewBox=\"0 0 24 24\" width=\"20\" height=\"20\" fill=\"none\" stroke=\"currentColor\" stroke-width=\"2.5\" stroke-linecap=\"round\" stroke-linejoin=\"round\" aria-hidden=\"true\">\n                <polyline points=\"5,12 10,17 19,7\"><\/polyline>\n            <\/svg>\n            <span class=\"try_prompt_copy_tooltip\" role=\"status\" aria-live=\"polite\">Copied<\/span>\n        <\/button>\n    <\/div>\n\n    <\/section>\n<br \/>\n<section class=\"try_prompt_block\" data-pulse-section-group=\"blog article\">\n    <h3 class=\"try_prompt_title\">Extend, second character enters<\/h3>\n\n    <div class=\"try_prompt_card\" data-pulse-section=\"blog article_try prompt\">\n        <p class=\"try_prompt_text\" id=\"try-prompt-6a8e7a6b292f4-text\">Continue the attached clip and carry the scene into its next beat. Match the source clip&#8217;s lighting, lens and colour exactly at the join. BEAT 1: hold the existing framing for two seconds so the join disappears. NEXT BEAT: a second character enters from frame right and the two acknowledge each other. FINAL BEAT: the camera widens to hold both of them. CONSISTENCY, ABSOLUTE: the original subject&#8217;s face, wardrobe and position stay continuous with the source. Nothing about the environment changes across the join.<\/p>\n\n        <button type=\"button\"\n                class=\"try_prompt_copy\"\n                data-copy-target=\"#try-prompt-6a8e7a6b292f4-text\"\n                aria-label=\"Copy prompt\"\n                title=\"Copy prompt\"\n                data-pulse-name=\"try prompt - copy\">\n            <img class=\"try_prompt_copy_icon\"\n                 src=\"https:\/\/cdn-cms-uploads.picsart.com\/cms-uploads\/79a9b2ac-f4c9-436b-b388-a05a484adf00.png\"\n                 alt=\"\"\n                 width=\"20\" height=\"20\"\n                 loading=\"lazy\" decoding=\"async\">\n            <svg class=\"try_prompt_check_icon\" viewBox=\"0 0 24 24\" width=\"20\" height=\"20\" fill=\"none\" stroke=\"currentColor\" stroke-width=\"2.5\" stroke-linecap=\"round\" stroke-linejoin=\"round\" aria-hidden=\"true\">\n                <polyline points=\"5,12 10,17 19,7\"><\/polyline>\n            <\/svg>\n            <span class=\"try_prompt_copy_tooltip\" role=\"status\" aria-live=\"polite\">Copied<\/span>\n        <\/button>\n    <\/div>\n\n    <\/section>\n<\/p>\n<h2><span id=\"When_a_take_misses\">When a take misses<\/span><\/h2>\n<p>Five misses, and the prompt change that fixes each one. None of them needs a different model.<\/p>\n<ul>\n<li><strong>The subject drifts across the take.<\/strong> The lock is doing too little. Name the face, the shape, the colour and the light angle outright, because a reference holds only for what the prompt names.<\/li>\n<li><strong>On-screen words come out garbled.<\/strong> The prompt named the text instead of spelling it. Type the exact words. On-screen text is strongest on dense, fact-heavy clips, and the model can only draw what the prompt holds.<\/li>\n<li><strong>The pace feels rushed or padded.<\/strong> The beats have no timing against them. Add a rough span to each beat, then read back the length smart duration suggests.<\/li>\n<li><strong>The facts are wrong.<\/strong> The prompt carried them instead of the source. Attach the deck, the sheet or the page, and tell the model to pull every figure word for word.<\/li>\n<li><strong>The take ends too early.<\/strong> Extend it. The second prompt only has to match the join and carry the next beat.<\/li>\n<\/ul>\n<h2><span id=\"How_to_use_WAN_30_in_Picsart\">How to use WAN 3.0 in Picsart<\/span><\/h2>\n<ol>\n<li><strong>Open AI Playground.<\/strong><br \/>\n<a href=\"https:\/\/picsart.com\/ai-playground\/\">WAN 3.0 runs in Picsart&#8217;s AI Playground<\/a>, the sandbox where one prompt can go to several models at once.<\/li>\n<li><strong>Select WAN 3.0.<\/strong><br \/>\nThe <a href=\"https:\/\/picsart.com\/ai-models\/wan-3-0\/\">WAN 3.0 model page<\/a> lists what it accepts. The rest of the roster sits alongside it under <a href=\"https:\/\/picsart.com\/ai-models\/\">AI video models<\/a>.<\/li>\n<li><strong>Add the reference.<\/strong><br \/>\nAttach the image, video, audio, document or page URL the film should be built from. A reference is what holds a subject or a product steady across the take.<\/li>\n<li><strong>Write the prompt in four parts.<\/strong><br \/>\nOpening frame, beats in order, pacing, then the consistency lock. Label each beat and keep them in order, so the instruction stays easy to parse.<\/li>\n<li><strong>Read the length it suggests.<\/strong><br \/>\nSmart duration control suggests a length from the action and pace you wrote. Read it back as a check on whether the beats fit the time you had in mind.<\/li>\n<li><strong>Generate at 1080p.<\/strong><br \/>\nOutput comes back at 1080p in an adaptive ratio. Write for the framing named in the prompt.<\/li>\n<li><strong>Extend the take.<\/strong><br \/>\nRun a second prompt against the finished clip to build a longer run from a base that already works.<\/li>\n<\/ol>\n<p><section class=\"tips_block\" data-pulse-section-group=\"blog article\">\n    <h3 class=\"tips_title\">Tips for getting more out of WAN 3.0<\/h3>\n\n    <div class=\"tips_list\" data-pulse-section=\"blog article_tips\">\n                    <article class=\"tips_item\">\n                <div class=\"tips_item_header\">\n                    <span class=\"tips_item_icon\" aria-hidden=\"true\"><\/span>\n                    <h4 class=\"tips_item_title\">Number the beats<\/h4>\n                <\/div>\n                <p class=\"tips_item_body\">WAN 3.0 parses detailed, multi-part instructions and honors them in the output. So a numbered list of actions lands better than one dense sentence.<\/p>\n            <\/article>\n                    <article class=\"tips_item\">\n                <div class=\"tips_item_header\">\n                    <span class=\"tips_item_icon\" aria-hidden=\"true\"><\/span>\n                    <h4 class=\"tips_item_title\">Let the model suggest the length first<\/h4>\n                <\/div>\n                <p class=\"tips_item_body\">Smart duration control reads the action and pace you wrote. Treat the number it offers as a check on the beat list.<\/p>\n            <\/article>\n                    <article class=\"tips_item\">\n                <div class=\"tips_item_header\">\n                    <span class=\"tips_item_icon\" aria-hidden=\"true\"><\/span>\n                    <h4 class=\"tips_item_title\">Spell out on-screen text<\/h4>\n                <\/div>\n                <p class=\"tips_item_body\">The model can only draw words the prompt holds, so type them the way they should appear.<\/p>\n            <\/article>\n                    <article class=\"tips_item\">\n                <div class=\"tips_item_header\">\n                    <span class=\"tips_item_icon\" aria-hidden=\"true\"><\/span>\n                    <h4 class=\"tips_item_title\">Send the document instead of summing it up<\/h4>\n                <\/div>\n                <p class=\"tips_item_body\">Uploading the deck or the sheet keeps the figures right. It also cuts the step where facts get retyped by hand.<\/p>\n            <\/article>\n                    <article class=\"tips_item\">\n                <div class=\"tips_item_header\">\n                    <span class=\"tips_item_icon\" aria-hidden=\"true\"><\/span>\n                    <h4 class=\"tips_item_title\">Name what has to stay the same<\/h4>\n                <\/div>\n                <p class=\"tips_item_body\">Reference detail holds across characters, objects, audio, scenes and styles. The lock sets which of those the model protects.<\/p>\n            <\/article>\n                    <article class=\"tips_item\">\n                <div class=\"tips_item_header\">\n                    <span class=\"tips_item_icon\" aria-hidden=\"true\"><\/span>\n                    <h4 class=\"tips_item_title\">Grow a long run from a clip you like<\/h4>\n                <\/div>\n                <p class=\"tips_item_body\">Extending a base you signed off on costs less than starting a longer take from scratch.<\/p>\n            <\/article>\n                    <article class=\"tips_item\">\n                <div class=\"tips_item_header\">\n                    <span class=\"tips_item_icon\" aria-hidden=\"true\"><\/span>\n                    <h4 class=\"tips_item_title\">Write one lock per thing that matters<\/h4>\n                <\/div>\n                <p class=\"tips_item_body\">A face, a logo and a light angle are three locks, and stacking them into one clause blurs all three.<\/p>\n            <\/article>\n            <\/div>\n<\/section>\n<br \/>\n<section class=\"section_faq\" id=\"faq-faq-6a8e7a6b29776\">\n            <h2 class=\"faq_title\" id=\"Get_answers_to_common_questions\">Get answers to common questions<\/h2>\n    \n    <div class=\"faq_items\">\n                    <div class=\"faq_item faq_item--active\">\n                <button type=\"button\" class=\"faq_question\" aria-expanded=\"true\">\n                    <span class=\"faq_question_text\">What does WAN 3.0 do?<\/span>\n                    <svg class=\"faq_chevron\" width=\"24\" height=\"24\" viewBox=\"0 0 24 24\" fill=\"none\" xmlns=\"http:\/\/www.w3.org\/2000\/svg\">\n                        <path d=\"M6 9L12 15L18 9\" stroke=\"currentColor\" stroke-width=\"1.5\" stroke-linecap=\"round\" stroke-linejoin=\"round\"\/>\n                    <\/svg>\n                <\/button>\n                <div class=\"faq_answer\" aria-hidden=\"false\">\n                    <div class=\"faq_answer_content\"><p>It makes video from a prompt plus an optional reference, in one unbroken shot of up to 30 seconds. Alibaba builds it, and it is the current release in the WAN line. Output comes back at 1080p in an adaptive ratio.<\/p>\n<\/div>\n                <\/div>\n                <div class=\"faq_divider\"><\/div>\n            <\/div>\n                    <div class=\"faq_item \">\n                <button type=\"button\" class=\"faq_question\" aria-expanded=\"false\">\n                    <span class=\"faq_question_text\">Does the prompt need to say how many seconds?<\/span>\n                    <svg class=\"faq_chevron\" width=\"24\" height=\"24\" viewBox=\"0 0 24 24\" fill=\"none\" xmlns=\"http:\/\/www.w3.org\/2000\/svg\">\n                        <path d=\"M6 9L12 15L18 9\" stroke=\"currentColor\" stroke-width=\"1.5\" stroke-linecap=\"round\" stroke-linejoin=\"round\"\/>\n                    <\/svg>\n                <\/button>\n                <div class=\"faq_answer\" aria-hidden=\"true\" data-collapsed>\n                    <div class=\"faq_answer_content\"><p>No. Smart duration control reads the action and pace you wrote, then suggests a length that fits. Naming the rhythm of the beats gives the model what it needs to get that number right.<\/p>\n<\/div>\n                <\/div>\n                <div class=\"faq_divider\"><\/div>\n            <\/div>\n                    <div class=\"faq_item \">\n                <button type=\"button\" class=\"faq_question\" aria-expanded=\"false\">\n                    <span class=\"faq_question_text\">Can a deck or a web page be the source?<\/span>\n                    <svg class=\"faq_chevron\" width=\"24\" height=\"24\" viewBox=\"0 0 24 24\" fill=\"none\" xmlns=\"http:\/\/www.w3.org\/2000\/svg\">\n                        <path d=\"M6 9L12 15L18 9\" stroke=\"currentColor\" stroke-width=\"1.5\" stroke-linecap=\"round\" stroke-linejoin=\"round\"\/>\n                    <\/svg>\n                <\/button>\n                <div class=\"faq_answer\" aria-hidden=\"true\" data-collapsed>\n                    <div class=\"faq_answer_content\"><p>Yes. Documents go in as .doc, .pdf, .ppt or .xls. A URL can point at a product page, a news piece, a paper or a brand site, and the model builds from what it reads there.<\/p>\n<\/div>\n                <\/div>\n                <div class=\"faq_divider\"><\/div>\n            <\/div>\n                    <div class=\"faq_item \">\n                <button type=\"button\" class=\"faq_question\" aria-expanded=\"false\">\n                    <span class=\"faq_question_text\">What happens when 30 seconds is not enough?<\/span>\n                    <svg class=\"faq_chevron\" width=\"24\" height=\"24\" viewBox=\"0 0 24 24\" fill=\"none\" xmlns=\"http:\/\/www.w3.org\/2000\/svg\">\n                        <path d=\"M6 9L12 15L18 9\" stroke=\"currentColor\" stroke-width=\"1.5\" stroke-linecap=\"round\" stroke-linejoin=\"round\"\/>\n                    <\/svg>\n                <\/button>\n                <div class=\"faq_answer\" aria-hidden=\"true\" data-collapsed>\n                    <div class=\"faq_answer_content\"><p>Extend the clip. Any finished take can be carried further, so a long run grows out of a base that already works instead of one huge render.<\/p>\n<\/div>\n                <\/div>\n                <div class=\"faq_divider\"><\/div>\n            <\/div>\n                    <div class=\"faq_item \">\n                <button type=\"button\" class=\"faq_question\" aria-expanded=\"false\">\n                    <span class=\"faq_question_text\">Will WAN 3.0 render exact words on screen?<\/span>\n                    <svg class=\"faq_chevron\" width=\"24\" height=\"24\" viewBox=\"0 0 24 24\" fill=\"none\" xmlns=\"http:\/\/www.w3.org\/2000\/svg\">\n                        <path d=\"M6 9L12 15L18 9\" stroke=\"currentColor\" stroke-width=\"1.5\" stroke-linecap=\"round\" stroke-linejoin=\"round\"\/>\n                    <\/svg>\n                <\/button>\n                <div class=\"faq_answer\" aria-hidden=\"true\" data-collapsed>\n                    <div class=\"faq_answer_content\"><p>Yes, and on-screen text is one of its stronger areas on dense, fact-heavy clips. Type the words in the prompt just as they should appear, then ask for them to render clearly.<\/p>\n<\/div>\n                <\/div>\n                <div class=\"faq_divider\"><\/div>\n            <\/div>\n                    <div class=\"faq_item \">\n                <button type=\"button\" class=\"faq_question\" aria-expanded=\"false\">\n                    <span class=\"faq_question_text\">How do you stop a character changing mid-clip?<\/span>\n                    <svg class=\"faq_chevron\" width=\"24\" height=\"24\" viewBox=\"0 0 24 24\" fill=\"none\" xmlns=\"http:\/\/www.w3.org\/2000\/svg\">\n                        <path d=\"M6 9L12 15L18 9\" stroke=\"currentColor\" stroke-width=\"1.5\" stroke-linecap=\"round\" stroke-linejoin=\"round\"\/>\n                    <\/svg>\n                <\/button>\n                <div class=\"faq_answer\" aria-hidden=\"true\" data-collapsed>\n                    <div class=\"faq_answer_content\"><p>Name the features in the prompt. Reference detail holds at pixel level across characters, objects, audio, scenes and styles, and the lock sets which of those the model protects.<\/p>\n<\/div>\n                <\/div>\n                <div class=\"faq_divider\"><\/div>\n            <\/div>\n                    <div class=\"faq_item \">\n                <button type=\"button\" class=\"faq_question\" aria-expanded=\"false\">\n                    <span class=\"faq_question_text\">Where does WAN 3.0 run in Picsart?<\/span>\n                    <svg class=\"faq_chevron\" width=\"24\" height=\"24\" viewBox=\"0 0 24 24\" fill=\"none\" xmlns=\"http:\/\/www.w3.org\/2000\/svg\">\n                        <path d=\"M6 9L12 15L18 9\" stroke=\"currentColor\" stroke-width=\"1.5\" stroke-linecap=\"round\" stroke-linejoin=\"round\"\/>\n                    <\/svg>\n                <\/button>\n                <div class=\"faq_answer\" aria-hidden=\"true\" data-collapsed>\n                    <div class=\"faq_answer_content\"><p>In [AI Playground](https:\/\/picsart.com\/ai-playground\/), where a single prompt can go to several models at once.<\/p>\n<\/div>\n                <\/div>\n                <div class=\"faq_divider\"><\/div>\n            <\/div>\n            <\/div>\n<\/section>\n\n<script type=\"application\/ld+json\">\n{\n    \"@context\": \"https:\/\/schema.org\",\n    \"@type\": \"FAQPage\",\n    \"mainEntity\": [\n        {\n            \"@type\": \"Question\",\n            \"name\": \"What does WAN 3.0 do?\",\n            \"acceptedAnswer\": {\n                \"@type\": \"Answer\",\n                \"text\": \"It makes video from a prompt plus an optional reference, in one unbroken shot of up to 30 seconds. Alibaba builds it, and it is the current release in the WAN line. Output comes back at 1080p in an adaptive ratio.\"\n            }\n        },\n        {\n            \"@type\": \"Question\",\n            \"name\": \"Does the prompt need to say how many seconds?\",\n            \"acceptedAnswer\": {\n                \"@type\": \"Answer\",\n                \"text\": \"No. Smart duration control reads the action and pace you wrote, then suggests a length that fits. Naming the rhythm of the beats gives the model what it needs to get that number right.\"\n            }\n        },\n        {\n            \"@type\": \"Question\",\n            \"name\": \"Can a deck or a web page be the source?\",\n            \"acceptedAnswer\": {\n                \"@type\": \"Answer\",\n                \"text\": \"Yes. Documents go in as .doc, .pdf, .ppt or .xls. A URL can point at a product page, a news piece, a paper or a brand site, and the model builds from what it reads there.\"\n            }\n        },\n        {\n            \"@type\": \"Question\",\n            \"name\": \"What happens when 30 seconds is not enough?\",\n            \"acceptedAnswer\": {\n                \"@type\": \"Answer\",\n                \"text\": \"Extend the clip. Any finished take can be carried further, so a long run grows out of a base that already works instead of one huge render.\"\n            }\n        },\n        {\n            \"@type\": \"Question\",\n            \"name\": \"Will WAN 3.0 render exact words on screen?\",\n            \"acceptedAnswer\": {\n                \"@type\": \"Answer\",\n                \"text\": \"Yes, and on-screen text is one of its stronger areas on dense, fact-heavy clips. Type the words in the prompt just as they should appear, then ask for them to render clearly.\"\n            }\n        },\n        {\n            \"@type\": \"Question\",\n            \"name\": \"How do you stop a character changing mid-clip?\",\n            \"acceptedAnswer\": {\n                \"@type\": \"Answer\",\n                \"text\": \"Name the features in the prompt. Reference detail holds at pixel level across characters, objects, audio, scenes and styles, and the lock sets which of those the model protects.\"\n            }\n        },\n        {\n            \"@type\": \"Question\",\n            \"name\": \"Where does WAN 3.0 run in Picsart?\",\n            \"acceptedAnswer\": {\n                \"@type\": \"Answer\",\n                \"text\": \"In [AI Playground](https:\/\/picsart.com\/ai-playground\/), where a single prompt can go to several models at once.\"\n            }\n        }\n    ]\n}<\/script>\n\n<script>\n(function() {\n    var container = document.getElementById('faq-faq-6a8e7a6b29776');\n    if (!container) return;\n\n    var items = container.querySelectorAll('.faq_item');\n    items.forEach(function(item) {\n        var button = item.querySelector('.faq_question');\n        var answer = item.querySelector('.faq_answer');\n        if (!button || !answer) return;\n\n        button.addEventListener('click', function() {\n            var isActive = item.classList.contains('faq_item--active');\n\n            if (isActive) {\n                item.classList.remove('faq_item--active');\n                button.setAttribute('aria-expanded', 'false');\n                answer.setAttribute('aria-hidden', 'true');\n                answer.setAttribute('data-collapsed', '');\n            } else {\n                items.forEach(function(other) {\n                    var otherBtn = other.querySelector('.faq_question');\n                    var otherAnswer = other.querySelector('.faq_answer');\n                    other.classList.remove('faq_item--active');\n                    if (otherBtn) otherBtn.setAttribute('aria-expanded', 'false');\n                    if (otherAnswer) {\n                        otherAnswer.setAttribute('aria-hidden', 'true');\n                        otherAnswer.setAttribute('data-collapsed', '');\n                    }\n                });\n                item.classList.add('faq_item--active');\n                button.setAttribute('aria-expanded', 'true');\n                answer.removeAttribute('data-collapsed');\n                answer.setAttribute('aria-hidden', 'false');\n            }\n        });\n    });\n})();\n<\/script>\n<\/p>\n<h2><span id=\"Start_with_a_30-second_brief\">Start with a 30-second brief<\/span><\/h2>\n<p>Pick the prompt closest to the film you need, paste it into <a href=\"https:\/\/picsart.com\/ai-playground\/\">Picsart AI Playground<\/a>, and swap in your own subject. Keep the four parts intact. The beats and the lock are what hold a 30-second take together. Then attach the deck or the page you have, and let the model read the brief instead of retyping it.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>WAN 3.0 turns one prompt into a single shot that runs a full half minute. It reads far more than prompts, though. Images, video, sound, files and live web pages all work as source. Output comes back at 1080p in an adaptive ratio. Most of the craft sits in one place. A 30-second shot needs &hellip; <\/p>\n<p class=\"link-more\"><a href=\"https:\/\/picsart.com\/blog\/wan-3-0-user-guide\/\" class=\"more-link\">Continue reading<span class=\"screen-reader-text\"> &#8220;WAN 3.0 user guide: everything the model can do + prompts&#8221;<\/span><\/a><\/p>\n","protected":false},"author":146,"featured_media":263660,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"_yoast_wpseo_title":"WAN 3.0 User Guide: Everything the Model Can Do and Prompts %%page%% %%sep%% %%sitename%%","_yoast_wpseo_metadesc":"How long a WAN 3.0 clip runs, the six inputs it reads, and 12 prompts to copy. A full user guide to 30-second AI video, with everything you need to start.","faq_show":true,"faq_enable_schema":true,"how_to_show":false,"how_to_show_on_single":false,"how_to_enable_schema":false,"how_to_is_upload":false,"faq_title":"Get answers to common questions","how_to_title":"How to use WAN 3.0 in Picsart","how_to_layout":"default","how_to_cta_text":"","how_to_cta_url":"https:\/\/picsart.com\/ai-playground\/","how_to_image_alt":"","how_to_display_image":0,"faq_items":null,"how_to_steps":[],"prompt_box_show":false,"prompt_box_placeholder":"","prompt_box_deeplink":"","prompt_box_submit_label":"","footnotes":""},"categories":[3181,1669],"tags":[],"class_list":["post-263802","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai","category-inspiration","entry"],"acf":{"footer_banner_name":"Start creating","footer_banner_link_":"https:\/\/picsart.com\/ai-playground\/","footer_banner_button_text_":"Try it now","faq_show":true,"faq_title":"Get answers to common questions","faq_enable_schema":true,"faq_items":[{"question":"What does WAN 3.0 do?","answer":"It makes video from a prompt plus an optional reference, in one unbroken shot of up to 30 seconds. Alibaba builds it, and it is the current release in the WAN line. Output comes back at 1080p in an adaptive ratio."},{"question":"Does the prompt need to say how many seconds?","answer":"No. Smart duration control reads the action and pace you wrote, then suggests a length that fits. Naming the rhythm of the beats gives the model what it needs to get that number right."},{"question":"Can a deck or a web page be the source?","answer":"Yes. Documents go in as .doc, .pdf, .ppt or .xls. A URL can point at a product page, a news piece, a paper or a brand site, and the model builds from what it reads there."},{"question":"What happens when 30 seconds is not enough?","answer":"Extend the clip. Any finished take can be carried further, so a long run grows out of a base that already works instead of one huge render."},{"question":"Will WAN 3.0 render exact words on screen?","answer":"Yes, and on-screen text is one of its stronger areas on dense, fact-heavy clips. Type the words in the prompt just as they should appear, then ask for them to render clearly."},{"question":"How do you stop a character changing mid-clip?","answer":"Name the features in the prompt. Reference detail holds at pixel level across characters, objects, audio, scenes and styles, and the lock sets which of those the model protects."},{"question":"Where does WAN 3.0 run in Picsart?","answer":"In [AI Playground](https:\/\/picsart.com\/ai-playground\/), where a single prompt can go to several models at once."}],"how_to_show":false,"how_to_show_on_single":false,"how_to_title":"How to use WAN 3.0 in Picsart","how_to_layout":"default","how_to_steps":null,"how_to_enable_schema":false,"how_to_is_upload":true,"how_to_cta_text":"","how_to_cta_url":"https:\/\/picsart.com\/ai-playground\/","how_to_display_image":null,"how_to_image_alt":"","prompt_box_show":false,"prompt_box_placeholder":"","prompt_box_deeplink":"https:\/\/picsart.com\/create\/editor?category=miniapps&app=com.picsart.aura","prompt_box_submit_label":"Create","try_prompt_show":false,"try_prompt_title":"Try this prompt","try_prompt_text":"","try_prompt_deeplink":"","tips_show":true,"tips_title":"Tips for getting more out of WAN 3.0","tips_items":[{"title":"Number the beats","body":"WAN 3.0 parses detailed, multi-part instructions and honors them in the output. So a numbered list of actions lands better than one dense sentence."},{"title":"Let the model suggest the length first","body":"Smart duration control reads the action and pace you wrote. Treat the number it offers as a check on the beat list."},{"title":"Spell out on-screen text","body":"The model can only draw words the prompt holds, so type them the way they should appear."},{"title":"Send the document instead of summing it up","body":"Uploading the deck or the sheet keeps the figures right. It also cuts the step where facts get retyped by hand."},{"title":"Name what has to stay the same","body":"Reference detail holds across characters, objects, audio, scenes and styles. The lock sets which of those the model protects."},{"title":"Grow a long run from a clip you like","body":"Extending a base you signed off on costs less than starting a longer take from scratch."},{"title":"Write one lock per thing that matters","body":"A face, a logo and a light angle are three locks, and stacking them into one clause blurs all three."}],"cta_banner_show":false,"cta_banner_title":"Need more space?","cta_banner_subtitle":"Extend any image in any direction with AI.","cta_banner_button_label":"Expand image","cta_banner_button_url":"","related_tools_title":"Related tools","related_tools_items":null,"post_level":""},"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v25.5 - https:\/\/yoast.com\/wordpress\/plugins\/seo\/ -->\n<title>WAN 3.0 User Guide: Everything the Model Can Do and Prompts - Picsart Blog<\/title>\n<meta name=\"description\" content=\"How long a WAN 3.0 clip runs, the six inputs it reads, and 12 prompts to copy. A full user guide to 30-second AI video, with everything you need to start.\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/picsart.com\/blog\/wan-3-0-user-guide\/\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"WAN 3.0 User Guide: Everything the Model Can Do and Prompts - Picsart Blog\" \/>\n<meta property=\"og:description\" content=\"How long a WAN 3.0 clip runs, the six inputs it reads, and 12 prompts to copy. A full user guide to 30-second AI video, with everything you need to start.\" \/>\n<meta property=\"og:url\" content=\"https:\/\/picsart.com\/blog\/wan-3-0-user-guide\/\" \/>\n<meta property=\"og:site_name\" content=\"Picsart Blog\" \/>\n<meta property=\"article:publisher\" content=\"https:\/\/www.facebook.com\/picsart\" \/>\n<meta property=\"article:published_time\" content=\"2026-08-26T01:51:07+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/cdnblog.picsart.com\/2026\/08\/Wan0.3_Blog_1200x800.png\" \/>\n\t<meta property=\"og:image:width\" content=\"1200\" \/>\n\t<meta property=\"og:image:height\" content=\"800\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/png\" \/>\n<meta name=\"author\" content=\"Julia Tovmasyan\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:creator\" content=\"@PicsArtStudio\" \/>\n<meta name=\"twitter:site\" content=\"@PicsArtStudio\" \/>\n<meta name=\"twitter:label1\" content=\"Written by\" \/>\n\t<meta name=\"twitter:data1\" content=\"Julia Tovmasyan\" \/>\n\t<meta name=\"twitter:label2\" content=\"Est. reading time\" \/>\n\t<meta name=\"twitter:data2\" content=\"13 minutes\" \/>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"WAN 3.0 User Guide: Everything the Model Can Do and Prompts - Picsart Blog","description":"How long a WAN 3.0 clip runs, the six inputs it reads, and 12 prompts to copy. A full user guide to 30-second AI video, with everything you need to start.","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/picsart.com\/blog\/wan-3-0-user-guide\/","og_locale":"en_US","og_type":"article","og_title":"WAN 3.0 User Guide: Everything the Model Can Do and Prompts - Picsart Blog","og_description":"How long a WAN 3.0 clip runs, the six inputs it reads, and 12 prompts to copy. A full user guide to 30-second AI video, with everything you need to start.","og_url":"https:\/\/picsart.com\/blog\/wan-3-0-user-guide\/","og_site_name":"Picsart Blog","article_publisher":"https:\/\/www.facebook.com\/picsart","article_published_time":"2026-08-26T01:51:07+00:00","og_image":[{"width":1200,"height":800,"url":"https:\/\/cdnblog.picsart.com\/2026\/08\/Wan0.3_Blog_1200x800.png","type":"image\/png"}],"author":"Julia Tovmasyan","twitter_card":"summary_large_image","twitter_creator":"@PicsArtStudio","twitter_site":"@PicsArtStudio","twitter_misc":{"Written by":"Julia Tovmasyan","Est. reading time":"13 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/picsart.com\/blog\/wan-3-0-user-guide\/#article","isPartOf":{"@id":"https:\/\/picsart.com\/blog\/wan-3-0-user-guide\/"},"author":{"name":"Julia Tovmasyan","@id":"https:\/\/picsart.com\/blog\/ko\/#\/schema\/person\/74b70f3125250c23596a5306775b702d"},"headline":"WAN 3.0 user guide: everything the model can do + prompts","datePublished":"2026-08-26T01:51:07+00:00","mainEntityOfPage":{"@id":"https:\/\/picsart.com\/blog\/wan-3-0-user-guide\/"},"wordCount":2912,"publisher":{"@id":"https:\/\/picsart.com\/blog\/ko\/#organization"},"image":{"@id":"https:\/\/picsart.com\/blog\/wan-3-0-user-guide\/#primaryimage"},"thumbnailUrl":"https:\/\/cdnblog.picsart.com\/2026\/08\/Wan0.3_Blog_1200x800.png","articleSection":["AI","Inspirational"],"inLanguage":"en-US"},{"@type":"WebPage","@id":"https:\/\/picsart.com\/blog\/wan-3-0-user-guide\/","url":"https:\/\/picsart.com\/blog\/wan-3-0-user-guide\/","name":"WAN 3.0 User Guide: Everything the Model Can Do and Prompts - Picsart Blog","isPartOf":{"@id":"https:\/\/picsart.com\/blog\/ko\/#website"},"primaryImageOfPage":{"@id":"https:\/\/picsart.com\/blog\/wan-3-0-user-guide\/#primaryimage"},"image":{"@id":"https:\/\/picsart.com\/blog\/wan-3-0-user-guide\/#primaryimage"},"thumbnailUrl":"https:\/\/cdnblog.picsart.com\/2026\/08\/Wan0.3_Blog_1200x800.png","datePublished":"2026-08-26T01:51:07+00:00","description":"How long a WAN 3.0 clip runs, the six inputs it reads, and 12 prompts to copy. A full user guide to 30-second AI video, with everything you need to start.","breadcrumb":{"@id":"https:\/\/picsart.com\/blog\/wan-3-0-user-guide\/#breadcrumb"},"inLanguage":"en-US","potentialAction":[{"@type":"ReadAction","target":["https:\/\/picsart.com\/blog\/wan-3-0-user-guide\/"]}]},{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/picsart.com\/blog\/wan-3-0-user-guide\/#primaryimage","url":"https:\/\/cdnblog.picsart.com\/2026\/08\/Wan0.3_Blog_1200x800.png","contentUrl":"https:\/\/cdnblog.picsart.com\/2026\/08\/Wan0.3_Blog_1200x800.png","width":1200,"height":800},{"@type":"BreadcrumbList","@id":"https:\/\/picsart.com\/blog\/wan-3-0-user-guide\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/picsart.com\/blog\/"},{"@type":"ListItem","position":2,"name":"WAN 3.0 user guide: everything the model can do + prompts"}]},{"@type":"WebSite","@id":"https:\/\/picsart.com\/blog\/ko\/#website","url":"https:\/\/picsart.com\/blog\/ko\/","name":"Picsart Blog","description":"Keep up with the latest news in photo editing, digital photography, and art trends.","publisher":{"@id":"https:\/\/picsart.com\/blog\/ko\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/picsart.com\/blog\/ko\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"en-US"},{"@type":"Organization","@id":"https:\/\/picsart.com\/blog\/ko\/#organization","name":"PicsArt Inc.","url":"https:\/\/picsart.com\/blog\/ko\/","logo":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/picsart.com\/blog\/ko\/#\/schema\/logo\/image\/","url":"https:\/\/cdnblog.picsart.com\/2016\/02\/PicsArt-logo.png","contentUrl":"https:\/\/cdnblog.picsart.com\/2016\/02\/PicsArt-logo.png","width":195,"height":43,"caption":"PicsArt Inc."},"image":{"@id":"https:\/\/picsart.com\/blog\/ko\/#\/schema\/logo\/image\/"},"sameAs":["https:\/\/www.facebook.com\/picsart","https:\/\/x.com\/PicsArtStudio","https:\/\/www.instagram.com\/picsart","https:\/\/www.linkedin.com\/company\/picsart-photo-studio","https:\/\/www.pinterest.com\/picsart"]},{"@type":"Person","@id":"https:\/\/picsart.com\/blog\/ko\/#\/schema\/person\/74b70f3125250c23596a5306775b702d","name":"Julia Tovmasyan","image":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/picsart.com\/blog\/ko\/#\/schema\/person\/image\/","url":"https:\/\/cdnblog.picsart.com\/2026\/03\/3285C16C-FD87-4868-A2F0-04B6A0815CE1-150x150.jpg","contentUrl":"https:\/\/cdnblog.picsart.com\/2026\/03\/3285C16C-FD87-4868-A2F0-04B6A0815CE1-150x150.jpg","caption":"Julia Tovmasyan"}}]}},"featured_image":{"url":"https:\/\/cdnblog.picsart.com\/2026\/08\/Wan0.3_Blog_1200x800.png","dimensions":[]},"_links":{"self":[{"href":"https:\/\/picsart.com\/blog\/wp-json\/wp\/v2\/posts\/263802","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/picsart.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/picsart.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/picsart.com\/blog\/wp-json\/wp\/v2\/users\/146"}],"replies":[{"embeddable":true,"href":"https:\/\/picsart.com\/blog\/wp-json\/wp\/v2\/comments?post=263802"}],"version-history":[{"count":16,"href":"https:\/\/picsart.com\/blog\/wp-json\/wp\/v2\/posts\/263802\/revisions"}],"predecessor-version":[{"id":263818,"href":"https:\/\/picsart.com\/blog\/wp-json\/wp\/v2\/posts\/263802\/revisions\/263818"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/picsart.com\/blog\/wp-json\/wp\/v2\/media\/263660"}],"wp:attachment":[{"href":"https:\/\/picsart.com\/blog\/wp-json\/wp\/v2\/media?parent=263802"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/picsart.com\/blog\/wp-json\/wp\/v2\/categories?post=263802"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/picsart.com\/blog\/wp-json\/wp\/v2\/tags?post=263802"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}