What You Animate
Any finished shot is a clip waiting to happen
The workflow is two steps, not one. Get the still right first — then animate the still. These are real CatalogX on-model generations; each one is a valid starting frame for a video.
You already have the shot. The Video tool animates it — a model turning to show the back, a dupatta lifting in a breeze, a slow push-in on the embroidery — and hands you a 4, 6 or 8-second MP4. No filming, no crew, no editing timeline.
A good catalogue photo answers most of a buyer’s questions: what colour is it, what’s the print, where does the hem land. The one thing it structurally cannot answer is what the garment does — whether that georgette floats or hangs, whether the lehenga flares when she turns, whether the sequins catch light or sit flat. That is fabric behaviour, and it only exists in motion.
It is also the format the places you sell in are built around. Feeds are vertical and video-first; a listing with a short clip gives a shopper something a grid of stills doesn’t. The obstacle has never been that sellers don’t want video — it’s that video meant a second shoot: booking the model again, a videographer, a day of filming and an editor. For a 40-SKU drop that is not a line item most small labels can carry.
CatalogX starts from a photo you already have. Upload a finished on-model shot — including one you generated here with virtual try-on or a flat-lay conversion — pick how it should move, and the AI renders a short clip whose first frame is your photo. The product in the clip is the product you photographed, because the photo is where it starts.
The workflow is two steps, not one. Get the still right first — then animate the still. These are real CatalogX on-model generations; each one is a valid starting frame for a video.
The render runs in the background, so you can queue a clip and carry on working elsewhere in the app.
Upload a still up to 10MB, or pick one of your earlier generations straight from your gallery. On-model shots, flat lays, ghost-mannequin photos and close-up crops all work.
Five motion presets, 4 / 6 / 8 seconds, portrait or landscape. Add extra direction in plain words — “she lifts the dupatta as she turns” — or take the AI’s suggestion.
A 720p MP4 with the generated soundtrack unless you switched audio off, no watermark. Usually ready in a minute or two. If a render fails, your credits come straight back.
Different shots want different motion, and the wrong choice is what produces a melting sleeve: telling a model to turn when there is no model in the frame cannot end well. So the tool looks at your upload first — is this a full-body on-model shot, a cropped one, a flat lay, a ghost mannequin, a close-up? — and pre-selects the preset that suits it, along with a few suggested directions written from what is actually in your picture. You can override all of it.
Each preset names both the motion and what has to hold still — which is what keeps a clip usable rather than merely impressive.
The model turns smoothly through about 180° and back, so the garment is seen front, side and back in one clip. Camera locked off at full-body framing. The default, and the best single clip for a listing.
A few unhurried steps toward the camera with a natural gait, the garment moving as it would in real life. The camera holds position and keeps the full body in frame.
A still, poised pose and fabric that moves — dupatta, pleats and hem drifting as if in a light breeze. Minimal body movement, locked-off camera. The one that sells drape.
The camera pushes in slowly on fabric, embroidery and detailing, revealing texture and sheen. The subject stays still; only the camera moves. Pairs well with a close-up crop as the source.
No person appears. The camera orbits the garment slowly with a gentle parallax that reveals cut, texture and drape against the same clean background — for flat lays, ghost-mannequin shots and accessories.
A free-text box takes extra direction in your words, and it is added to the preset rather than replacing it — so the constraints that keep your garment intact still apply to a clip you directed yourself.
A video that quietly recolours your kurti is worse than no video, because it can be live on your product page before anyone notices. Every clip carries the same hard constraints.
Held constant, in every generation: the garment’s colour, print, pattern, embroidery, trims and proportions exactly as they are in the source photo; the model’s face, skin tone, hair and body; and the background, framing and lighting for the whole clip. Nothing is re-draped, shortened, lengthened or replaced, and no text, captions, logos or watermarks are added.
Blocked explicitly: colour shift, changing prints and patterns, morphing fabric, warped or restyled garments, a different outfit appearing mid-clip, distorted hands and extra limbs, identity changes, overlaid text or watermarks, jump cuts, camera shake, flicker and cartoon rendering. These go to the model as a separate negative instruction, which is a stronger channel than prompt text.
The same logic drives the suggested directions: a suggestion that told the AI to recolour or swap the garment is dropped before you ever see it, because the whole point of a tap-to-apply chip is that you can trust it. This is the same product-fidelity discipline behind our colour variant tool — the clip has to show the item the buyer will actually receive. That said, AI generation is not deterministic: watch the clip through once before you publish it.
No asterisks. This is the whole spec sheet.
| Setting | What’s available |
|---|---|
| Clip length | 4, 6 or 8 seconds. Eight seconds is the longest a single clip can be. |
| Resolution | 720p. |
| Framing | Portrait 9:16 or landscape 16:9. Auto picks whichever crops your photo least — portrait and square sources become 9:16, a clearly landscape source becomes 16:9. |
| File format | MP4, downloaded straight from the app. No watermark. |
| Audio | On by default — the soundtrack Veo generates is kept, and it is billed either way. Switch it off to have the track stripped for a silent clip, which is what a listing or a feed video with its own music needs. |
| Input | One still image, up to 10MB — uploaded, or picked from your CatalogX gallery. |
| Motion | Five presets, plus optional free-text direction of your own. |
| Credit cost | 3 credits per second: 12 for a 4-second clip, 18 for 6 seconds, 24 for 8 seconds. |
| Plan required | Business and above. |
| At once | Up to 2 clips rendering simultaneously per account. |
| Render time | Usually a minute or two; occasionally a few minutes. Runs in the background — you can leave the tab. |
| If it fails | Credits are reserved at the start and refunded in full on any failure or timeout. |
If you sell ethnic wear, motion is not a nice-to-have — it is the part of the product a photo has always failed to communicate. A Banarasi and a georgette saree can photograph almost identically and behave nothing alike. A lehenga’s flare is invisible until someone turns. A dupatta is, functionally, a piece of fabric defined entirely by how it falls.
That is what the fabric-flow preset is for: the model holds a still pose while dupatta, pleats and hem drift slowly, with a locked-off camera and minimal body movement. It is the closest thing to handing a buyer the fabric. Turntable is the other ethnic-wear workhorse — one clip that shows the pallu, the back of the blouse and the border run, which otherwise costs you three separate photos.
CatalogX is built around these garments rather than adapted to them — pleat and pallu direction, border continuity and hem length are things we measure on the still before anything animates it. Start with the still-photo pages if you haven’t generated one yet: saree photography, lehenga photography and kurti photography. There’s also a guide to getting saree drape right with AI.
Portrait 9:16 is the shape the vertical feeds use; landscape 16:9 suits a product page or a site banner.
A 6-second vertical clip is a drop-in for Instagram Reels and YouTube Shorts — switch audio off if you want your own track or a trending sound to sit on top cleanly. Selling through your Instagram profile? See the Instagram shop guide.
Vertical product video in the same 9:16 frame as the rest of the feed. Our TikTok Shop page covers the image side of a listing.
Drop a turntable clip next to the stills on your Shopify or WooCommerce product page — it answers the fit-and-drape question buyers would otherwise ask in a DM.
A short MP4 is small enough to forward through a broadcast list or a reseller group without it being compressed into mush. More on this in our WhatsApp catalogue guide.
Video creative for an ad set, generated from the same catalogue shots as the still creative, so the product looks identical across both.
Check each platform’s current video specs before you upload — requirements change, and we’d rather you verify than take our word for it.
We’ll be straightforward about this: video generation is not available on Free, Starter or Pro. It sits on the Business plan, which is ₹3,000/month for 250 credits, and on Agency above it.
The reason is cost, not packaging. One video clip is roughly 150 times the unit cost of one AI image at the provider we generate through. Every other capability at CatalogX is on every plan — premium quality, ecommerce packs, batches of up to 20 garments, AI shadows — because they are all the same order of magnitude. Video isn’t, so it gets a gate.
Credits still meter each clip on top of the plan: 3 credits per second, so 12 for a 4-second clip, 18 for 6 seconds and 24 for 8. Unused credits roll over while your plan is active, and a failed render is refunded automatically rather than silently spent.
You don’t need Business to start. Signing up gives you 5 free credits with no card, which is enough to generate the on-model stills that a clip would animate, and to judge whether the image quality is worth building a video workflow on. Upgrade when the stills have convinced you. Full plan detail on the pricing page.
Open the Video tool in CatalogX, upload a still product photo (or pick one of your earlier generations), choose a motion preset and a clip length of 4, 6 or 8 seconds, and generate. The result is a 720p MP4 you can download and post. Your photo is the first frame, so the clip only ever shows the product you actually photographed.
Video is a Business plan feature — ₹3,000/month for 250 credits — and is also included on Agency. It is not available on Free, Starter or Pro. One clip costs roughly 150× what one AI image costs to produce, which is why it sits there rather than being included everywhere. See pricing.
4, 6 or 8 seconds — eight is the maximum for a single clip. Short is also the right length: it’s the format vertical feeds and product-page video slots actually use, and a longer generated clip has more opportunity to drift away from your product.
Three credits per second of video: 12 credits for 4 seconds, 18 for 6, 24 for 8. Credits are reserved when the render starts and refunded in full if it fails or times out.
720p, in portrait 9:16 or landscape 16:9. Auto framing picks whichever crops your source least: portrait and square photos become 9:16, and a clearly landscape photo becomes 16:9.
It is instructed not to. Every clip carries fixed constraints to keep the garment’s colour, print, pattern, embroidery, trims and proportions exactly as the source photo shows them, keep the model’s face and body identical, and hold the background and lighting steady — plus a separate negative instruction blocking colour shift, morphing prints, restyled outfits and any added text, logo or watermark. AI generation isn’t deterministic, so watch the clip once before you publish it.
Yes, by default — Veo generates a soundtrack and CatalogX now keeps it, so the MP4 arrives with audio. Switch audio off before rendering if you’d rather have a silent clip, which is usually what you want for a listing or a feed video you’ll put your own music on.
Yes, and it’s arguably where video earns its keep. The fabric-flow preset holds a still pose while the dupatta, pleats and hem drift — the one thing a photo of a saree can never show. Turntable is the other common pick, because it gets the pallu and the back of the blouse into the same clip as the front.
Yes — that’s the product-only preset. No person appears; the camera orbits the garment slowly against the same background. Use it for flat lays, ghost-mannequin shots and accessories.
Usually a minute or two, sometimes a few minutes — considerably slower than an image, because video is a much heavier job. It renders in the background, so you can leave the tab and come back. Anything that stalls past the timeout is failed and refunded automatically rather than left hanging. You can have up to two clips rendering at once.
Yes. You keep ownership of what you upload and what CatalogX generates from it, and you may use generated output commercially — product listings, marketing and social. Files come with no watermark. The full wording is in our terms.
Start with the stills — five free credits, no card. When they’re right, Business turns any one of them into a clip.