Whisk - labs.google/fx

Một công cụ thử nghiệm mới cho phép bạn dùng câu lệnh ở dạng hình ảnh để khơi gợi ý tưởng và kể câu chuyện của mình.

labs.google/fx

Whisk Tool from Google Labs

Whisk is an experimental image generation tool from Google Labs, designed to allow users to create and remix images using other images as prompts instead of relying solely on text prompts. This approach aims to simplify the creative process, making it more accessible and intuitive for users who may not be familiar with detailing prompts through text.

How Whisk Works

Whisk employs Google's Gemini AI for analyzing and generating captions of uploaded images, which are then processed by Imagen 3, a cutting-edge image generation model. This process involves transforming images into text descriptions (I2T) and then back into images (T2I), capturing the essence of the input images rather than exact replicas. The workflow typically involves three main stages:

1. Prepare: Upload images to define subjects, scenes, and styles. These are analyzed by the Gemini AI to create captions.
2. Explore: Users can mix these components to generate new images. Users have the option to provide additional guidance, such as preferred color schemes or actions within the scene.
3. Refine: Users can make smaller adjustments to the output, ensuring the image remains close to their vision.

Key Features

• Image-Based Prompts: Users drag and drop images instead of crafting lengthy text prompts, defining key aspects like subjects, scenes, and styles.
• Gemini AI and Imagen 3 Integration: Delivers a seamless transition from image to text narration and back to a visual outcome, focusing on creativity.
• User-Friendly Interface: Includes features like drag-and-drop functionality, 'Inspire Me,' and 'Roll the Dice' to encourage experimentation without detailed input from users.

Applications and Use Cases

Whisk is intended for rapid visual exploration and is particularly suited for hobbyists and creative professionals looking to experiment with visual ideas. It can be used for designing digital content like plushies, enamel pins, and stickers, and is also useful in crafting unique visual representations for creative projects.

Limitations

While Whisk facilitates creativity, it has some limitations:

• The final images may deviate in appearance from the original input, due to its essence-capturing design.
• It is experimental and may require further development to improve user experience.
• It is not intended for high-fidelity, professional-grade image editing.

Availability

Whisk is currently accessible in the United States and several other countries through Google Labs, but it is not available in the UK. Users can interact with the tool via Google's website and are encouraged to share feedback and creations as part of the tool's ongoing development.

Comparison with Other Tools

Whisk differentiates itself by focusing on image-based prompts, unlike text-heavy models such as OpenAI's DALL-E or Adobe Firefly. This allows for a more visual-centric creative process, though it may limit some aspects of creative diversity attainable through open-ended text inputs. Whisk’s user-oriented design fosters creativity by simplifying the method of image generation, directing it more towards artistry and exploration than precise control.