Google's Imagen models for image generation are now available in GPTunneL

Google's Imagen models for image generation are now available in GPTunneL

GPTunneL has added three image generation models from Google: Imagen 3, Imagen 3 Fast, and Imagen 4. Now, depending on what matters most to you — speed, quality, or image detail — you can choose among the most advanced tools offered by Google.

Google's Imagen models are known for their ability to interpret text prompts to create photorealistic and stylized images, as well as render text correctly.

What the Imagen models can do in GPTunneL

Each model has its own strengths:

Imagen 3 – A quality, balanced model

This model offers a good balance between image quality and performance.

  • Realism and stylization: Imagen 3 handles photorealistic images and images in various artistic styles well. The model better processes details in prompts written in natural language.
  • Fewer artifacts: Compared to previous versions, Imagen 3 produces images with fewer unwanted visual distortions.
  • Good for core visual content: Can be used to create images for social media, promotional materials, article illustrations, and presentations.

Example generation:

The image depicts a futuristic city built into a sheer cliff with cascading waterfalls flowing down. Tall columns of vegetation climb the cliff face, and transparent dome terraces supported by thin metal structures sit on the ledges. In the foreground, a winding pedestrian walkway with railings is crossed by several people: a woman in a long dress holds a tablet, others stroll and admire the views. Glowing points of light resembling fireflies are visible throughout the area. In the background to the right, a sunset over the bay's water surface, with wind turbines and distant mountain silhouettes rising beyond. Below, thin paths and flower beds smoothly transition into terraced structures, creating a sense of harmonious fusion between nature and high technology.

Prompt: A futuristic city built into a vertical cliffside covered with hanging gardens, cascading waterfalls, and solar glass domes. The city integrates nature and technology: skybridges made of transparent material, wind turbines shaped like flowers, bioluminescent plants glowing along the paths. Sunset lighting, warm and surreal. Style: hyper-detailed matte painting, cinematic. -ar 16:9

Imagen 3 Fast – A simplified, fast version

Imagen 3 Fast is built for situations where speed matters.

  • Faster generation: This version runs noticeably faster than the standard Imagen 3.
  • For quick iterations: Well suited for creating draft variants, quickly testing ideas and concepts when you don't need final high quality (though the generation quality is still high!).

Example generation:

The image shows a nighttime city street in a steampunk and late-medieval style. In the center of a cobblestone avenue walks a cloaked figure, back turned to the viewer. On both sides of the street stand tall half-timbered houses with pointed roofs and smoking chimneys. At ground level are market stalls with colorful stretched awnings: orange on the left, green and red on the right. At each stall stand shopkeepers dressed in Victorian-era clothing, lit by the soft yellow glow of lanterns and hanging lamps. On the left in front of the first house stands an elaborate steam apparatus on a cart: metal gears, valves, and pipes surround a central light source, like a steam engine ready to run. Further down the street, silhouettes of other passersby appear, lit by lanterns, while above rise tall towers and spires of gothic buildings disappearing into the evening mist. The sky is thick with clouds, with only a dark blue gap visible at the center, creating contrast with the warm glow of the street lights. Altogether it creates the atmosphere of a mystical city where steam-powered technology intertwines with the architecture of past centuries.

Prompt: A medieval steampunk market square at twilight, filled with brass machinery, gear-driven carts, and merchants in leather aprons. A towering cathedral with pipes and rotating clock mechanisms looms in the background. Steam rises from vents in the cobblestone streets. Atmosphere is dense with fog and warm lantern light. Style: painterly, richly textured, low saturation. -ar 16:9

Imagen 4 – The most powerful model

Imagen 4 is the most advanced model in the lineup, built for tasks that require very high image quality.

  • Improved photorealism: Delivers a high level of realism, allowing you to create images close to photographs.
  • Complex scenes and detail: Handles generating complex scenes with many objects and a high degree of detail well.
  • Legible text: Has the ability to fit text into images cleanly and clearly.
  • For production-ready graphics: Can be used to create covers, commercial visuals, portfolio elements, and other projects that require high image quality.

Example generation:

The image shows a portrait of a middle-aged man in an astronaut helmet decorated on the side with colorful feathers. The man's face is covered with dark blue tattoos in the form of geometric lines and dots that accentuate his facial features. His gaze is focused forward: his right eye emits a bright glowing gleam. Behind him unfolds a cosmic background with a nebula and stars, and around the man's head a semicircular halo is visible, dotted with fanciful symbols resembling runic signs. The contrast between traditional ethnic elements (feathers, tattoo) and the futuristic context of the helmet and space creates a sense of synthesis between ancient beliefs and future technologies.

Prompt: A hyperrealistic portrait of a shaman-astronaut, blending tribal face paint and ceremonial feathers with a scratched, vintage space helmet. The background is a galaxy nebula with floating spirit animals and tribal symbols forming constellations. One eye glows with starlight. High detail, dramatic lighting from below, cinematic realism.

Where can you try the Imagen models?

All three versions – Imagen 3, Imagen 3 Fast, and Imagen 4 – are available in Creative Lab on the GPTunneL platform. You can test and compare the models to choose the best tool for your tasks.