LifeLike AI Playground

Welcome to the LifeLike AI Playground

LifeLike AI is designed to work best when integrated into third-party applications and tools. As an API (Application Programming Interface), LifeLike becomes truly powerful when developers build it into their own apps, websites, and creative tools.

This playground lets users and developers experiment with LifeLike's capabilities and see what the API can do. Think of it as a testing ground to explore the technology before integrating it into your own projects.

API code

Changelog

July 3, 2025

  • NEW: Released LifeLike Remix in beta! This set of models is specifically trained for drawing duplications. Instead of text prompts, you provide an image and LifeLike will create variations of it using Sharpie, Whiteboard, or Chalk models (Crayons coming soon). Unlike other AI tools that convert images to text descriptions first, LifeLike Remix works directly from your input image, better preserving the composition. Note: This current first version is trained to use photographs as input. A version that takes drawings (from impression pads for example) is currently being trained. This is a beta feature and may produce unpredictable results. The quality of the output is greatly dependant on the input image. Your feedback is greatly appreciated!

May 20, 2025

  • Resolved an issue where Glyphs images were printed too small
  • Fixed compatibility issue with Crayons model in Turbo mode

March 12, 2025

  • Added transparent background option that removes the background from images and saves them as PNG files with transparency. For Sharpie and Whiteboard models, the white background is removed. For Chalk model, the black background is removed.

February 26, 2025

  • Improved speed of The Selector
  • Added Turbo mode: if you want ultra-fast images at the potential cost of quality, Turbo mode will give you upto 4 images in 3 seconds. This is particularly useful if you plan on selecting the best candidate(s) yourself. The crayons model does not yet support Turbo mode.
  • Added option to disable The Selector. This generates just 1-4 image(s) at a time as opposed to 10 to select the best ones. This too will improve the speed at the potential cost of quality.

February 20, 2025

  • Added crayons model to create extremely realistic children's doodles. Thanks to Pontus Lindman for supplying his kid's drawings as training data. This model does not use The Selector, as it is so forgiving in the way it portrays subjects. Drawing an extra leg or misplacing an eye is all acceptable in this model.

January 31, 2025

  • Fixed a bug that would disable upscaling.
  • While trying out a newer version of the models, speed is temporarily slowed down to roughly 15s per request.

January 8, 2025

  • Models updated to v5.18
  • Added new landing page at https://lifelike.nextleveleffects.com
  • Added this changelog to Playground
  • Introduced The Selector (automatically pick the best results out of 20 variations)
  • Added upscale option (1 (default) = 512x512, 2 = 1024x1024, 3=1536x1536, 4 = 2048x2048)