Prompt Share: Retro Solarpunk with Leonardo AI

Combining Lora Elements and Cyberpunk Prompts for beautiful images.

AI images by the author.

This week we’re exploring some combinations of the new Leonardo AI “Soft Retro Futurism” LoRA Element, both standalone and with other elements.

These prompts are also great on their own, and will look incredible in any platform.

Let’s go!

The LoRA Elements

The acronym LoRA stands for “Low Rank Adaption” model, and it allows you to add a particular visual style to your images and fine-tune it. There are LoRAs for everything from coloring book pages, to 2d anime styles, and 3d stylized characters. I’ve covered several of them in previous articles, as well as the back issues of my AI Character Academy Newsletter.

Today, we’re working with two LoRAs: the brand-new “Soft Retro Futurism” and one of my longtime favorites, “Solarpunk”.

Soft Retro Futurism gives images a nice, classic sci-fi feel, with pink and white tones, and polished surfaces. Solarpunk, by contrast, gives images an apocalyptic-yet-lush style — think scrap-metal-meets-Garden-of-Eden. In the images in this article, Solarpunk is what adds the vegetation.

Solarpunk provides lush vegetation and futuristic structures. Soft Retro Futurism provides glossy textures along with futuristic architecture. AI image by the author.

The Prompt & Settings

For these images, I’m using my old standby, the AlbedoBase XL model. You can also use VisionXL or DiffusionXL. Settings are:

• Preset Style: choose Illustration or Environment
• Alchemy V2: ON
• Images Size: 512 x 768
• Upscaler: Alchemy Refiner, Low Normal
• LoRA: Soft Retro Futurism, 80%
• LoRA: Solarpunk, 30%

Prompt: full-body Professional Vector Illustration of a Beautiful Woman, 30 years old, cyberpunk, Japanese woman, short dark hair, future sunglasses, full-body shot, futuristic clothing, illustration, fashion

Negative Prompt: low resolution, mask, text, artist name, username, signature, watermark, blur, blurry, error, cropped, double head, deformations, worst quality, low quality, normal quality, poor quality, jpeg artifacts, black and white, border, frame, poorly drawn, error, abstract, asymmetrical, cropped, cropped feet, disconnected, duplicate, mask, cover face, face cover, mask, extra, missing, bad anatomy, deformed, disproportionate, eye bags, fused, malformed hands, bad hands, extra limbs, missing limb, extra leg, missing leg, missing hand, extra hand, cropped out face, cropped out head, no head, missing head, (((3d model))). (((figurine))), ((((deformed face)))), (((smooth)))

You can also tweak the LoRA settings. In the image below, I’ve toned down both Elements: Soft Retro Futurism at 20%, and Solarpunk at 10%:

Lower the strength of the LoRA Elements for less “glossy” styles. AI image by the author.

Variations

You can change the character’s race, hairstyle, or gender with minimal adjustments to the prompt:

full-body Professional Vector Illustration of a Beautiful Ebony Woman, 30 years old, cyberpunk, white mohawk hair, full-body shot, futuristic clothing, illustration, fashion

Style & Character References

Add a style reference image for additional looks. Here, I’ve used a two-tone fabric texture. I’ve also adjusted the LoRA strengths slightly. Experiment with different values to arrive at your style!

L: a fabric texture used as a Style Reference image; R: The result using the original prompt. AI images by the author.

You can also use hard-surface style images, character references, textures, and more to mix-and-match for endless variety.

[embed]Fabric References, 100 Images to Use as Style References, Input Images, Design - Etsy This Digital Prints item is sold by circuitscreamart. Ships from United States. Listed on Jul 11, 2024circuitscreamart.etsy.com

Upscalers

Leonardo has many built-in upscale options that can refine and enlarge images. I tend to use Alchemy Refiner at a Low Normal setting, because it smooths out rough edges but also generally keeps the character’s look. Higher settings can fix faces and hands, and also will often re-imagine character faces, so experiment to find your preference.

Leonardo AI’s Upscalers. L: Alchemy Refiner at Low Normal. R: Alchemy Refiner at Low Smooth. AI images by the author.

You can also upscale images using the web-based service BigJPG, or a desktop application like Luminar, Topaz Gigapixel Pro, or PaintShop Pro.

For a comparison of Leonardo AI’s Upscalers, see my Medium article here: **https://medium.com/kinomoto-mag/leonardo-ais-upscalers-a-comparison-58f0a5a72897**

I hope you’ve enjoyed this prompt share! For additional prompts and articles, **follow me here on Medium! And, make sure to subscribe to my [AI Character Academy Newsletter](https://ai-character-academy.ck.page/join)** for even more prompts and character design tips.

See you next time!

[embed]Midjourney & Leonardo AI Style Reference Images, 500 Images to Use With Your Prompts for Amazing… This Digital Prints item is sold by circuitscreamart. Ships from United States. Listed on Aug 30, 2024circuitscreamart.etsy.com

I hope this prompt share was fun and helpful! If you want to share your results, follow me on X — I’d love to see what you create!

If you enjoyed this story, clap 50 times and follow me on Medium. I share prompts, tutorials, and workflows several times each week. You can also support my work at my website, **circuitscream.art.**

For additional AI business tips and ideas, check out Stella Sky’s website — she has a ton of great resources for AI Artists looking to succeed online: https://realstellasky.com/