Google goes bananas, Kling start & end frames & Midjourney x Meta partnership
Examples of everything inside:
Total reading time about 4 minutes.
Welcome to Visually AI!
Gemini 2.5 Flash, aka Nano Banana
Gemini 2.5 Flash, also known as Nano Banana, is Google’s lightweight image editing model built for creators who need fast, flexible adjustments.
It can handle tasks like removing or replacing backgrounds, shifting camera angles, adding or removing objects, and even combining characters into a single scene.
Compared to existing options, it works in a similar space as Flux Kontext (precise style transfer and visual consistency), Qwen Image Edit (fine-grained object-level edits), and Runway References (image-to-image editing for continuity), but is designed to be smaller, faster, and integrated directly into Google’s ecosystem for quick creative workflows
You can use it in the Gemini app, Google AI Studio, and many platforms such as:
Here’s a peek from the Gemini 2.5 Flash landing page:
Here are a couple of my first results, using the same woman in two different scenes:
Kling 2.1 Start & End Frames
Kling launched 2.1 Start & End frames for precision control. Upload a start and end frame and prompt the motion in between frames.
This is a huge improvement over the previous Start & End frame feature in version 1.6.
Kling’s integration of DeepSeek makes it simple to upload two images and click to get a polished prompt to create a smooth transition video with intelligent motion interpolation between keyframes, without morphing or visual discontinuities.
Example 1:
Example 2:
Higgsfield Draw to Video
Higgsfield recently released a new Draw to Video feature. The interface includes annotation tools to make it easier for you to visually describe what you want to see in the output video.
You can see an example of what it can do and how it can be incorporated into creative ideas below:
Meta x Midjourney
Meta’s Chief AI Officer, Alexandr Wang announced Meta’s partnership with Midjourney a few days ago.
We haven’t seen many details of what this partnership involves and Midjourney has not made a public announcement as of the time of this article, but it will be interesting to see future developments…
🚀 Runway Game Worlds
Runway’s new Game Worlds tool is now available for all users.
You can create a full video game with text prompts in this feature. The AI agent will guide you through the process as you describe the type of game, characters, rewards, etc.
Check out some of the games that have already been created here.
🚀 My Recent Top AI Tools Picks
Glif.app: Glif has a huge assortment of the latest generative models neatly packaged into ready-to-use multi-modal apps for image, video, and text output. The revamped UI is simplified for creative workflows.
Google Translate: Get live conversations translated in real time in over 70 languages with this new feature rolling out to iOS and Android users now.
Ideogram Styles: Choose from a library of styles or create your own with three images for text to image generation on Ideogram.
ElevenLabs Video to Music: Upload a video (up to 10-seconds) to the Studio tab and select “Video to Music” to get a music prompt based on your video analysis. Click “Generate” for a full musical soundtrack for your clip. You can add sound effects, voiceover, and captions.
I used the tool for this video with generated music and sound effects within ElevenLabs:
🖼️ Image Prompts
These character sheets work well with a single input image on image to image editing models , like Nano Banana, Qwen Edit, FLUX Kontext, or GPT Image. I used Qwen Image Edit on Replicate.
Prompt: Expression reference grid of the same character head and shoulders close-up. Six expressions: neutral, smiling, laughing, angry, sad, shocked. Even spacing, clean grid layout, pure white background, consistent character details.
Prompt: Character reference lineup. One full body turnaround row above, one row of close-up headshots below (front, side, back). Clean white background.
I used the input image on the left to generate the poses on the right:
🎥 Video Prompts
Text to video prompt: Inside a grand, dimly lit library, the camera dollies slowly through floating dust motes as a young woman closes a glowing ancient book.
Generated on Seedance Pro:
Text to video prompt: A rain-soaked city alley, neon reflections on the pavement, as a figure in trench coat sprints past flickering signs; the camera whip-pans to follow. Dramatic cinematic action soundtrack
Generated on Veo 3:
Thank you for reading.
Have a creative week!
