☕ Like my work? Feel free to support me on Ko-fi: https://ko-fi.com/granddyser Thanks! 🙏

Krea 2 Retro Cyberpunk 50s v2

A new Krea 2 version, retrained on a fresh, hand-curated dataset. Atomic-age retro-futurism shot like a 1950s Kodachrome film, with chrome prosthetics, neon diners, tail-finned hover-cars and rain-slick boulevards. Works with ordinary prompts, all showcase images were made without the trigger word.

MiniMax H3 – Retro Cyberpunk 50s

Retro Cyberpunk 50s is now available for MiniMax H3: text-to-video with sound. Atomic-age retro-futurism shot, with chrome prosthetics, neon diners, tail-finned hover-cars and rain-slick boulevards. Works with ordinary prompts, all showcase videos were made without the trigger word and carry their full ComfyUI workflow.

The new MOSS Image version is here!

Trained on a new dataset, this update brings a more realistic look and improved overall image quality to my 1950s retro cyberpunk LoRA. It keeps the blend of vintage aesthetics and futuristic technology, now with a stronger sense of realism.

What if the future was designed in the 1950s?

Cyberpunk 50s drops your generations into a world where chrome meets neon, where Art Deco skyscrapers tower over monorail cities, and where elegant men and women in tailored suits walk alongside machines half human, half polished steel.Transform your generations into a cinematic 1950s retrofuturistic world. This isn't just another style LoRA. It's a complete visual universe.

What you get

• Consistent retrofuturistic aesthetic across portraits, cityscapes, interiors, vehicles, and full scenes

• Cinematic film-quality look with warm Kodachrome-style color grading

• Rich detail: mechanical prosthetics, chrome vehicles, streamline architecture, neon accent lighting, 1950s fashion with sci-fi twist

• Works with people, environments, objects — no one-trick pony

Technical details

• Base model: Z-Image Turbo (Tongyi-MAI/Z-Image-Turbo)

• Trigger word: cy50s

• Shift: 7.0

• Sampler: dpmpp_2m

• Scheduler: sgm_uniform

• Trained on 115 hand-curated, high-quality images with consistent style

• Recommended: guidance_scale 1 | steps 8 | LoRA weight 0.7–1.0

• Format: bf16 safetensors

Note: All example images on this page were generated using version 1.2.