# Zelvune > Create images and videos with AI ## Pages - [Home](https://zelvune.com/): Zelvune AI video generator landing page - [Create](https://zelvune.com/create): Generate AI videos with Zelvune - [AI Video](https://zelvune.com/ai-video): Browse supported AI video models and generator options - [FLUX 3 AI Video Generator](https://zelvune.com/flux-3-ai-video-generator): FLUX 3 multimodal AI video generator guide with audio, references, keyframes, and multi-shot workflows - [Photo to Anime Generator](https://zelvune.com/photo-to-anime): Turn photos into anime-inspired images with the Zelvune AI image generator - [Seedance 2.0](https://zelvune.com/m/seedance-2): Seedance 2.0 AI video generator and model guide - [Pricing](https://zelvune.com/pricing): Zelvune plans and credits - [Privacy Policy](https://zelvune.com/privacy-policy): Privacy policy - [Terms of Service](https://zelvune.com/terms-of-service): Terms of service - [Refund Policy](https://zelvune.com/refund-policy): Zelvune refund policy ## AI Video Models - [Zelvune 1.5](https://zelvune.com/ai-video?model=zelvune-1-5): A fast trial model for a quick Zelvune experience. A quick way to experience Zelvune: create a 4-second video from your prompt and optional image. - [Seedance 1.5 Pro](https://zelvune.com/ai-video?model=seedance-1-5-pro): Cinematic motion, 4–12s, up to 1080p. ByteDance's flagship video model. Seedance 1.5 Pro excels at cinematic motion — long takes, complex camera moves, and consistent characters across frames. Best choice when quality matters more than speed. - [Seedance 2.0](https://zelvune.com/m/seedance-2): Next-gen motion, faster renders. The newest ByteDance model. Faster than 1.5 Pro with comparable quality for most shots. Use this when you need quick social-first renders and can trade a bit of fidelity for speed. - [Kling](https://zelvune.com/ai-video?model=kling): Realistic human motion, long takes. Kuaishou's Kling model is the strongest option for realistic human motion and long-form shots. Great for product demos, lifestyle B-roll, and human-centric storytelling. - [Veo 3](https://zelvune.com/ai-video?model=veo-3): Google's cinematic narrative model. Veo 3 from Google DeepMind pushes the limits of narrative video. Strong cinematic framing, accurate physics, and the only model with native 21:9 ultra-wide support. Limited to 8 seconds per generation. - [Grok Imagine](https://zelvune.com/ai-video?model=grok-imagine): Stylized, fast, meme-friendly. xAI's Grok Imagine is built for stylized output. Fast renders, strong on graphic and motion-design aesthetics, less consistent on photorealism. Great for ads, social content, and concept tests. - [Hailuo](https://zelvune.com/ai-video?model=hailuo): Budget-friendly, anime-friendly. MiniMax's Hailuo is the most cost-effective model in the catalog. Anime and 2D motion are its strong suit; photorealistic cinematic quality is not. Use it for prototypes, drafts, and stylized content. - [Seedance 1.0 Pro](https://zelvune.com/ai-video?model=seedance-1-0-pro): The original Seedance, still solid. ByteDance's first-generation Seedance Pro. 5 or 10 second outputs at 480p/720p/1080p. A reliable workhorse for production workflows that don't need the latest 1.5 Pro or 2.0 features. - [Seedance 1.0 Lite](https://zelvune.com/ai-video?model=seedance-1-0-lite): Fast drafts, lowest cost. The lightweight Seedance. Half the cost of Pro, 30% faster, same 5/10 second durations. Ideal for rapid iteration on prompts and concepts before committing credits to a Pro render. - [MiniMax H3](https://zelvune.com/ai-video?model=minimax-h3-text-to-video): High-fidelity text-to-video generation. MiniMax H3 turns a prompt into a cinematic video with 4–15 second duration and 768P or 2K output. - [MiniMax H3 Image to Video](https://zelvune.com/ai-video?model=minimax-h3-image-to-video): Animate a first frame with precise motion. MiniMax H3 animates a required first frame and an optional last frame at 4–15 seconds in 768P or 2K. - [MiniMax H3 Reference to Video](https://zelvune.com/ai-video?model=minimax-h3-reference-to-video): Direct motion from image, video, and audio references. MiniMax H3 follows up to 9 images, 3 videos, and 3 audio references to create a 4–15 second video.