Latest Updates

Two new AI models are now available on Melies.

HappyHorse 1.0 (Video)

Alibaba's #1 ranked video model on Artificial Analysis Video Arena. Generates cinematic video with native audio in a single pass, supports multilingual lip-sync across 7 languages, reference-to-video for character consistency, and video editing. Up to 15-second clips at 1080p with 720p available at half the credit cost.

Imagen 4 Ultra (Image)

Google's highest quality image model in the Imagen 4 family. Enhanced detail, richer lighting, and fewer artifacts compared to the Fast variant. Best for hero images and production-quality output at just 8 credits per image.

Two new AI models are now available on Melies.

GPT Image 2 (Image)

OpenAI's latest image model with state-of-the-art photorealism and pixel-perfect text rendering. Supports flexible resolutions up to 4K, quality control (low/medium/high), and image editing. Replaces GPT Image 1.5.

Hailuo 2.3 (Video)

MiniMax's most capable video model. Major leap in cinematic realism, camera control, and motion physics. Improved character micro-expressions, anime and illustration style support. 1080p Pro quality with 6 and 10-second clips. Replaces Hailuo 02.

Two fixes around failed generations and credit visibility.

Refunds now show in credit history

When a generation fails, the refund is now written to your credit history alongside the original charge, and the failed asset shows a "credits refunded" badge with the exact amount. A toast notification also confirms the refund in real time. Previously refunds happened silently, which looked like credits had vanished.

Veo 3.1 "Extend video" fixed

The Extend video route for Veo 3.1 was failing on Fal because the endpoint now requires a fixed 7-second duration. The client now pins that value automatically so extend jobs complete instead of erroring out.

Two new AI models are now available on Melies.

Seedance 2.0 (Video)

ByteDance's most advanced video model. Generates cinematic video with native audio (included at no extra cost), accepts up to 9 reference images for character consistency, and supports text, image, audio, and video inputs in a single generation. Up to 15-second clips at 720p with start and end frame control. Replaces Seedance v1 Pro.

Ideogram V3 (Image)

The best model for text and typography in images. Generates logos, posters, and signage with near-perfect text accuracy. Supports 15 aspect ratios, 4 rendering speed tiers, and multiple style types including realistic, design, and general. Includes image editing via prompt.

Two changes driven by the April SEO impact analysis.

New /ai-filmmaking landing page

A dedicated category page for the "ai filmmaking" keyword cluster, covering what AI filmmaking is, the five stages of the workflow (story, visuals, motion, sound, assembly), and how every Melies tool fits each stage. Internal links to 12 tool pages.

Google currently serves the homepage for "ai filmmaking" queries (1,045 impressions at position 6.5). A dedicated page should capture top 3 and 3-5x current clicks for this cluster.

Movie Idea Generator title revert

Rewrote /movie-idea-generator meta title after a March 20 change regressed ranking on short-film and film-concepts queries. The new title restores those semantic tokens while keeping the Plot/Logline positioning.

• Before (Mar 20): "AI Movie Idea Generator: Plot, Characters & Logline"
• After (this release): "Free AI Movie Idea Generator: Plot, Logline & Short Film Ideas"

Lost queries we expect to recover: "short film generator", "short film idea generator", "movie theme generator", "movie genre generator". Net baseline loss was approximately 85 clicks per week.

Footer: AI Filmmaking column expanded

Added AI Filmmaking (the new page) and AI Filmmaking Tools links to the AI Filmmaking footer column.

Six new AI models are now available on Melies.

Sora 2 Pro (Video)

OpenAI's flagship video model with native audio. Cinematic-quality clips up to 12 seconds with synchronized dialogue, environmental sounds, and ambient audio. Supports text-to-video and image-to-video. Starting at 200 credits.

Imagen 4 Fast (Image)

Google's fast image model at just 5 credits. High-quality images with enhanced detail, richer lighting, and fewer artifacts.

Seedream 5 Lite (Image)

ByteDance's reasoning image model with built-in web search and chain-of-thought reasoning. Exceptional multilingual text rendering with support for image editing using up to 10 reference images. 10 credits.

Recraft V4 (Image)

Professional design-focused model with color palette control, accurate text rendering, and clean composition. Built for brand systems and production workflows. 15 credits.

WAN v2.6 (Video)

Alibaba's latest video model. Up to 15-second clips at 1080p with multi-shot prompts and reference-to-video for character consistency. Replaces WAN v2.2. Starting at 70 credits.

LTX 2.3 (Video)

Lightricks' latest open-source video model with native audio, up to 4K (2160p) output, and a sharper VAE for better textures and faces. Replaces LTX 2 Pro. Starting at 50 credits.

Two new state-of-the-art AI models are now available on Melies.

GPT Image 1.5

OpenAI's latest image generation model, ranked #1 on all major image leaderboards. Exceptional prompt adherence, photorealism, and text rendering. Supports image editing and transparent backgrounds.

Grok Imagine Video

xAI's new video generation model, ranked #1 on multiple video leaderboards. Generates cinematic video with native audio in under 17 seconds. Supports text-to-video, image-to-video, and video-to-video editing with 7 aspect ratio options.

This update brings full mobile support, a new tutorial system, a critical pricing fix, and SEO improvements across all tool pages.

Mobile UX

Melies now works properly on mobile devices with a responsive layout, touch gestures, and a new navigation menu.

• Hamburger menu for mobile header navigation
• Gallery toolbar optimized for small screens
• Touch gestures for swipe navigation and pinch-to-zoom on image galleries
• Responsive layouts across all tool pages

Default Video Model Fix

Free users are now defaulted to LTX 2 Pro (50 credits) instead of Veo 3.1 (400+ credits). Previously, free users with 50 credits couldn't generate a single video with the default model. This change unblocks the entire video funnel for new signups.

Tutorial System

• New tutorial card component added to 15 landing pages, linking to relevant guides and walkthroughs
• Consistent Character tutorial published with step-by-step guide and YouTube video embed
• YouTube video player added to bottom CTA sections with play button overlay

SEO Optimization

• Front-loaded "Free" in tool page titles for voice generator, storyboard, lip sync, and image enhancer. All competitors lead with "Free" and Bing data showed 5,500+ impressions with near-zero clicks when "Free" was buried at the end.
• Merged voice pages into a single /ai-voice-generator page (301 redirect from /ai-voiceover-generator). Two pages were cannibalizing each other for the same keywords.
• Added Veo credits FAQ to pricing documentation for search queries about Veo pricing.

Melies 2.0 is a complete rewrite with four powerful standalone AI tools, new models, and a rebuilt editor.

AI Image Generator

Create stunning images from text prompts with 16 AI models, consistent characters, and up to 4K resolution.

• 16 AI models including Flux 2, Seedream 4.5, Grok Imagine, and Nano Banana 2
• 9 aspect ratios -21:9, 16:9, 4:3, 3:2, 1:1, 2:3, 3:4, 9:16, and Auto
• Resolutions up to 4K for print-ready and high-quality output
• Extract subject from any image to reuse in new generations
• 100+ variation presets across 11 categories: camera angles, shot size, expressions, lighting, time of day, weather, color grading, mood, art style, era, and aspect ratio
• 148 AI Actors with consistent character generation across scenes
• Right-click context menu -copy prompt, reuse, generate variation, use as reference
• Image upscaling to enhance resolution and detail

AI Video Generator

Turn images or text into cinematic video clips with 8 AI models, camera presets, and lip sync.

• 8 video models including Veo 3.1, Kling v3 Pro, Hailuo 02, WAN v2.2, Seedance v1 Pro, and LTX 2 Pro
• Text-to-video and image-to-video generation
• 3 to 15 second durations with configurable length
• Up to 4K resolution output
• Camera movement presets -Dolly, Jib, Static
• Multi-shot sequences with per-shot prompts (Kling v3 Pro)
• Lip sync and Voice ID support for talking characters
• Extend video to make clips longer
• Native audio generation with Veo 3.1
• Character consistency via reference images
• Export as MP4 or WebM

AI Movie Poster Generator

Design professional movie posters in 20 styles with custom titles, taglines, and full prompt control.

• 20 style templates -Cinematic, Anime, Retro, Film Noir, Minimalist, Horror, Sci-Fi, Watercolor, Comic Book, Art Deco, Grindhouse, Bollywood, Western, Pixel Art, Surrealist, Documentary, Cartoon, Epic Fantasy, Indie Film, and Neon Noir
• Text overlay for title and tagline
• Full prompt customization for complete creative control

Movie Idea Generator

Generate unique movie concepts with a guided wizard that builds your story step by step.

• Guided 9-step wizard -archetype, tone, time period, location, hero type, flaw, antagonist, catalyst, and theme
• 10 story archetypes including Monster in the House, Golden Fleece, Out of the Bottle, and more
• AI-generated title, logline, synopsis, and characters
• 1,000+ pre-generated ideas gallery for instant inspiration