The AI video generator forevery viral trend.

A template is a ready-made prompt, sometimes with a sample opening frame. Pick one, change a few words to make it yours, and render. You are describing your own scene — dimly.art does not copy anyone else’s video.

dimly.art turns plain-language direction into a video prompt, picks the selected model and renders the clip. Attach a still to animate it. Everything stays in one conversation.

Usually a few minutes. Video models queue and render far slower than image models — a 5-second clip is typically 1–3 minutes, a 10-second one longer. You can keep chatting while it works.

Yes. Upload a still and describe what should move. The image becomes the opening frame and the model animates from there — a product turntable, a portrait that blinks, a landscape with drifting clouds.

MiniMax H3 and Seedance 2.5, both from the same chat. MiniMax supports up to 4K; Seedance supports up to 30 seconds and generated audio — pick per shot, or let the agent pick.

Each render costs credits based on the model and the clip length — a 10-second clip costs twice a 5-second one. The price is shown in the composer before you send, and a failed render is refunded automatically.

You can just buy credits. One-time packs need no subscription, work with every model and stay valid for 12 months. A subscription gives you a better rate per credit and refills every month, so it suits people who post every week. You can use both: pack credits are spent after your plan credits.

Yes. All paid plans include a full commercial license. Check the terms of the specific model you used if you plan to publish at scale.

Your uploads and clips are encrypted in transit and at rest. We never train on your content. You can permanently delete any conversation or clip at any time.

Re-render with one changed sentence. The agent keeps your session history, so you can say “same again, but faster zoom” — or open the trend guide’s prompt, adjust one detail, and render both versions to compare.