👉 Runway improves image generation with Frames

Runway has launched Frames, a new AI image model that enhances stylistic control and visual fidelity. This model allows users to create custom worlds by integrating stylistic choices consistently across outputs, a challenge many previous models faced. Some examples of its capabilities include generating 1980s SFX-inspired visuals, or Japanese Zine aesthetics. Runway plans to gradually roll out Frames within its Gen-3 Alpha and API in the coming weeks.

👉 Luma AI launches new creative platform

Luma AI has expanded its Dream Machine into a full creative platform, introducing a mobile app alongside a new image generation model called Luma Photon. Key features include multi-image prompting, single-image character references, and the ability to upload reference images for tailored outputs. Luma Photon stands out for its efficiency and quality, generating high-resolution images rapidly while using advanced AI architecture. The subscription model offers four tiers, starting at $9.99 and providing custom pricing for enterprise solutions.

👉 The first full AI video game is a Minecraft simulation

The world's first fully AI-generated game, Oasis, has been developed by Decart, offering a unique Minecraft-inspired experience that evolves based on player actions. Players interact with the environment in real-time, shaping the game world without a traditional game engine. Initially, sessions were limited to five minutes but have been extended to 15 minutes, with players queuing for access. The game allows customization, including uploading images to create unique worlds. YouTuber YuB described the experience as "kinda creepy," highlighting the dynamic and unpredictable nature of the gameplay.

👉 IMAX leverages AI for global content localization

IMAX is partnering with camb.ai to use AI for localizing its original content, aiming to capture the growing demand for non-English films and shows. This initiative comes as the global entertainment industry reached $2.8 trillion in 2023, with significant growth expected in non-English content viewing. camb.ai's technology will facilitate AI-based speech translation for IMAX's documentaries and films. The partnership will enhance accessibility and potentially reduce translation costs, with plans to roll out AI translations in stages.

👉 ElevenLabs introduces AI Podcasts

ElevenLabs has unveiled GenFM, a new feature for its iOS app that enables users to create multispeaker podcasts by uploading various content types. This tool supports 32 languages and automatically selects voices for generating podcasts, adding natural speech elements like "ums" and pauses to enhance realism. The approach contrasts with traditional tools that focus on removing such elements, aiming for a more human-like conversational flow. Future updates will introduce greater customization options for users.

Let’s dive into the style reference feature on Magnific👇

The feature uses Mystic model to generate an output image following a reference image you upload. You can upload a simple picture, but we suggest to upload a grid with multiple images to maximize the style adherence.

Reference image → The image that transfers the style. For example, if it's an image with a cartoon style, you'll transfer that style to the input image.

Where to find this feature? On mystic, go on Mystic v2.5 tab, then, under the prompt tab you will find the style reference feature.

First, you need to input a prompt, then you can upload a style reference image, we will try 2 different ones for our test. You can also adjust settings:

Prompt adherence: (from 0 to 100%), increasing this value will make your generation more faithful to the prompt, but it may transfer the style a bit less accurately. Lower values will give you more creative images and closer to the style reference. you can

HDR: (from 0 to 100%), increasing this value can give you a more detailed image, at the cost of a more 'AI look' and slightly worse style transfer. Lower values have a more natural and artistic look but may increase artifacts.

prompt “A modern interior design featuring a family dining area. The space has a high arched ceiling, and an elegant, mustard yellow oval-shaped dining table in the center. On the table a carafe and some drinking glass. Around the table, there are 6 green designer chairs. Above the table hangs a contemporary, asymmetrical white pendant light. The flooring is stone with a light speckled pattern. The background includes a door framed by minimalist wall art, and natural light floods into the room.”

First, we tried with a style reference grid made from previously generated images in Midjourney, which has dark tones and atmosphere, that works well with landscapes and creepy subjects.

Here is the result 👇

Then, we tried with another sref grid we generated in Midjourney, which has vibrant colors, mostly pink, that works well with daylight, human subjects and textures.

Here is the result 👇

Let’s see the 2 renders side by side

Pretty cool, right?

Our conclusion is that the style reference in Magnific works very well, as long as the prompt you use follows the mood of your style reference image, (eg. if you use a very dark style reference and you want a summer picture) the render will follow the aesthetic with great adherence.

👉 Try it here

👋 See you next week for a tutorial

🎉 Get The Mahazine Issue 3! 🎉

What Awaits You Inside?

🔥Global Trends: Stay ahead with insights into the evolving AI creative landscape.
👀Visual Inspirations: Fuel your creativity with captivating imagery.
📣Trending AI News: Keep up with the latest digital campaigns and advancements.
🎒AI Tutorials: Master new techniques and enhance your skills.
⭐And More...: Explore a wealth of resources tailored for creative minds

We truly hope you will enjoy the insights and inspiration we’ve packed into this edition. Your support means the world to us!

Thank you for being a part of our journey. We can’t wait to hear what you think!

Best,
The Mahazine Team