Create AI video from a text prompt, animate a photo, connect two keyframes, or guide a shot with references. Choose the inputs your model supports, then review settings and credits in Studio.
AI-generated static illustration; not a rendered video or a guaranteed result.
Text to video
Turn a scene description into a short AI video with model-specific motion controls.
Image to video
Use an approved still or reference to guide framing, subject direction, and movement.
Transparent credits
Resolution, duration, audio, and model choice update the cost before you submit.
Workflow
How the AI video generator works
Start with a scene or an image, choose a supported workflow in Studio, then review the cost before rendering.
01
Describe the motion
Write the scene, camera movement, subject action, lighting, and final mood.
02
Choose video controls
Open the suite to set model, duration, resolution, aspect ratio, audio, and references.
03
Render
Confirm credits and render the video in your own workspace.
Model line-up
Fifteen video models, one prompt box
Available inputs, duration, resolution and audio depend on the selected model. Compare the model groups below, then check the current options in Studio.
Seedance 2.0 Mini
Cost-efficient motion drafts at 480p or 720p with all four creation workflows. The one to stay on while you are still checking whether a shot works.
Seedance 2.0 Fast · 2.0
Faster iteration at 480p and 720p; full Seedance 2.0 adds 1080p and 4K with text, frame and multimodal reference workflows.
Lumi Video 1.1 Pro
Our own tier at 480p, 720p or 1080p, up to 30 seconds, driven by text, first and last frames or multimodal references.
Seedance 2.5
Longer clips at 480p, 720p or 1080p, with first and last frames and multimodal references.
Wan 3.0 Prime · MiniMax H3
Advanced multimodal video at 480p, 720p or 1080p, driven by text, a still or a reference. MiniMax H3 adds a 768p and 2K path.
Controls
Four ways in, and everything after
Choose an input method for your shot. The illustrations below explain the workflow; available inputs and controls depend on the selected model.
Text to video
Start from a written scene when you do not have a source image. For example: a red tram crossing a rain-wet street at dusk, a slow tracking shot, warm window light. Name the subject, action, camera movement and lighting.
AI-generated static illustration; not a rendered video or a guaranteed result.
Image to video
Upload a photo you have the right to use and describe what should move. For this coastal portrait concept, ask for a gentle breeze in the hair and curtain with a steady camera. Start with a clear subject and a short, specific motion.
AI-generated static illustration; not a rendered video or a guaranteed result.
First and last frame
Use two stills to guide the beginning and end of a shot. The greenhouse concept shows closed doors in the first frame and open doors in the last. Choose a model with first/last-frame support; the generated transition may vary.
AI-generated static illustration; not a rendered video or a guaranteed result.
Reference to video
Use subject and visual references to guide a new shot. This concept pairs a blue bottle with foliage and warm light. Reference image, video and audio support differs by model; references guide the output rather than guarantee an exact match.
AI-generated static illustration; not a rendered video or a guaranteed result.
Duration and resolution
From 4 seconds up: to 15 on the Seedance 2.0 line, to 30 on Seedance 2.5, Wan 3.0 Prime and Lumi Video 1.1 Pro, at 480p, 720p or 1080p.
Framing and audio
16:9, 9:16, 1:1, 3:4, 4:3, 21:9 or adaptive, with an audio toggle on the models that support it.
Use cases
What the AI video generator is built to do
Text to video scenes
A written scene becomes a short clip — subject action, camera movement, lighting and mood in one pass.
Stills into motion
An image you already generated, or one you own, becomes the first frame of a moving shot.
Controlled transitions
First and last frames guide the beginning and end of a shot. Keep the subject and framing consistent when you want a coherent transition.
Vertical and wide cuts
9:16 for phone-first output, 21:9 for wide scenes — set before rendering rather than cropping a 16:9 clip afterwards.
Longer takes
Seedance 2.5, Wan 3.0 Prime and Lumi Video 1.1 Pro run up to 30 seconds, so a shot does not have to end at five.
Adult creative work
Uncensored models take on the adult scenes that general-purpose generators refuse, and treat them as ordinary work in your own workspace.
Credits
Video is billed by the second, and quoted first
Lumi Arts bills in credits, not seats. Video is billed per second of output, so duration moves the price directly — and the per-second rate itself rises with the model and the resolution, which is why 1080p costs several times what 480p does on the same model.
That is why the suite quotes the whole clip — model, resolution, duration and quantity together — before you submit, instead of showing one price per model and leaving you to do the multiplication.
If a render fails, the credits go back automatically — you are never charged for a clip that never arrives.
Creative boundaries
Uncensored video creation with clear rules
Lumi Arts supports adult creative work, but minors, non-consensual content, sexual deepfakes, harassment, and misuse of real people are prohibited.
FAQ
AI video generator FAQ
01
Can I generate uncensored AI videos here?
+
Choose a model that supports adult content in Studio. Model rules differ; general-audience models do not accept adult requests. Age confirmation is required for adult work.
02
Can I make video from an image?
+
Yes. In the suite, supported models can turn approved images or references into short videos.
03
Will this page start a render?
+
No. The prompt area only sends you to the suite. It does not submit a generation job on this page.
04
How is video cost calculated?
+
The suite calculates credits from the selected model, resolution, duration, quantity, and audio settings before submission.
05
Which AI video models does Lumi Arts use?
+
Fifteen public video models. Most carry their vendor's own name — Seedance 2.0 Mini, 2.0 Fast, 2.0 and 2.5, Wan 2.6, 2.6 Flash and 3.0 Prime, MiniMax H3, plus Grok and Gemini tiers. One model, Lumi Video 1.1 Pro, runs under our own name. The Grok, Gemini and promotional tiers are general-audience only and do not accept adult work.
06
How long can an AI video be?
+
From 4 seconds up. The Seedance 2.0 line runs to 15 seconds, and Seedance 2.5, Wan 3.0 Prime and Lumi Video 1.1 Pro run to 30. A new clip defaults to 5 seconds.
07
Can I make a video from two images?
+
Yes. First and last frame mode takes two stills and generates the motion between them. It is available on the Seedance 2.0 line and on Seedance 2.5.
08
Does the AI video generator produce audio?
+
There is an audio toggle on the models that support it, and reference-driven runs can take an audio track as input. Audio is off by default.
