tonik’s Post

The sketch to render has been one of the most popular AI workflows using stable diffusion. A newer method using IPadapters (add-on for using images as prompts), you can use it to copy the style, composition, or a face in the reference image. Great to convert a sketch into a specific style render you have a reference of and more creative flexibility. Tutorial on how this is done: https://lnkd.in/gGr8qMFp Instagram account: https://lnkd.in/enjnZ5ep #stablediffusion #ai #architecture #styles #design #sketchtoimage #urban #apps

Obscured Orange is Abstract. Made in #ROI with Rubbrband Prompt Parts: + Abstract + Bold Brush strokes + Orange Color with Grayscale + Plain bright background Prompt Part: Subject + Cup of Tea + Sky-Scrapper + Sports car + A wall Inspiration + Since Autumn is coming, many leaves have started to turn Yellow and Orange. + + The countryside in Ireland is very green which becomes orange in Autumn. Thoughts: + The style is versatile. It produces many possibilities across portrait, landscape and close up photos. + Orange alone doesn't do well, but when mixed with grayscale produces a stark contrast, which catches our attention. Note: + All images generated using one or another AI image generation technique. Images are for research and entertainment purposes. + Respect to all the artists who did the hard work for years creating the actual art. Their work got sucked into the machine. PS: Autumn is here in Ireland. PPS: Went to see some developer friends. PPPS: The field of AI is just getting started. It is hard to get hold of good servers to run a AI project. Random Question:  Is Orange the new Blue? — Generic social media plug: 👯 Follow 🗣 Comment ♻️ Repost #madebyai #obscure #orange

Taking an older fashion prompt (Naboowear) from a few months back and tweaking it with MJ personal code and several --sref styles that I generated and saved over the past weeks. I dialed the --sw weight and --s weights up and down to push the results. This explanation sounds like gibberish if you don't use Midjourney, and I'm still frustrated that we have to enter code values into the prompts instead of having UI with numerical sliders. It's just lazy. This applies to all of the AI text to image generation tools. #aiart #aiartwork #aicc #aicccreators #midjourney #midjourneyart #midjourneyai #illustration #nijijourney #NIJI #characterdesign #characterart #generativeart #generativeai

Uizard AI transforms you from clueless to Picasso without a single line of code, all free like digital magic. 🎩✨🖱️ #AI #ArtificialIntelligence #design #mockup #prototypes #Uizard #wireframe

👉 Midjourney has introduced 'Style Tuner', which allows users to create their own styles and apply them to new image generations by changing various aspects such as colors and character details. 👉 To use the Style Tuner, users type "/tune" followed by a prompt, select the number of base styles, and upon completion are presented with a custom web page where they can create their personalized style. 👉 Additional features of the Style Tuner allow users to create random style codes, combine multiple style codes, and control the strength of the style code. 👇 How to use the new Style Tuner for Midjourney #ai #artificialintelligence #midjourney

Designing with the Colors You Love. Khroma uses AI to learn which colors you like and creates limitless palettes for you to discover, search, and save. A Personalized Algorithm: By choosing a set of colors, you'll train a neural network powered algorithm to generate colors you like and block ones you don't, right in your browser. Creates Infinite Combos: Khroma has learned from thousands of the most popular human-made palettes across the internet to produce great combos which you can view as typography, gradient, palette, or custom image. Which You Can Search: Got something in mind? Khroma's search allows you to search and filter the generator by hue, tint, value, color, as well as hex and RGB values. And Save to Your Collection: Create an unlimited library of your favorite combos to reference later. Get color names, hex codes, RGB values, CSS code, and WCAG accessibility rating of each pair. @khromacolors Check out the Khroma Instagram for inspiration curated by and featuring the work of Lucie Bajgart. Make Colors Work for You You're just a few minutes away from an endless color combination generator designed by you. Generate Get Started https://www.khroma.co/

Here's a 22 page golden ticket to leveraging MidJourney like a pro. MJ is a powerful AI Image Generation tool. But 99% of users don't know how to get the best results. As with most LLMs, prompt engineering is key. Here are 16 prompts that, when combined, will help you produce visionary outputs: ➟ Prompt Structure: Subject:[subject description] ➟ ImageType: [visual descriptor, colour grading] ➟ Environment Focus: [depth, camera focus] ➟ Subject Focus: [depth, camera focus] ➟ Environment Description: [details] ➟ Subject Wearing: [clothing attire] ➟ Subject Position: [body position] ➟ Foreground: [foreground focus] ➟ Phototype: [shot type, details] ➟ Lighting: [lighting details] ➟ Textures:[image textures] ➟ Parameters: [parameters] ➟ Environment: [location] ➟ Time of Day: [details] ➟ Filmstock: [film type] ➟ Emotion: [tone] #genai #creativeai #midjourney ---- P.S. If you'd like a web version of this guide that you can easy bookmark, comment and I'll send you a link.

How I created 4 consistent environments from a 3d render (made with Blender) using Depth Anything ControlNet and IP Adapter. Challenge was simple: Can I create compositionally consistent environment designs with differing textures using AI? Workflow was: - Create a 3d environment render in Blender and export the viewport. - Import to Invoke. - Use the render as a Depth Anything ControlNet. - Add in another fully rendered environment concept as an IP adapter. - Adjust with the text prompt. - Generate multiple images. Some thoughts here: - IP Adapter (Image Prompts) is powerful at capturing style and may be sufficient for certain use cases (rather than needing to develop a custom trained model for every use case). - Depth Anything is incredible! It even correctly identified a lamp post in the resulting depth map… Text prompts are still a major driver in overall image quality. Let me know if there are any other workflows you’d like me to try! #stablediffusion #sdxl #invoke #AICC #midjourney #blender

llama.ttf is a font file which is also a large language model and an inference engine. Yes, really. Some font rendering engines are Turing-complete, and this font abuses that to its logical extreme to embed AI into your text. There's a video on that page showing it in action writing a fairy tale. It does this not by changing the text you type, but by changing how the text is rendered; if you change the font or copy and paste it into another application the story disappears, because it only exists as dynamically modified glyphs in LLamaSans. This is utterly useless but technically astounding.

New Open-Source AI Image Generator Beats Midjourney? TL;DR: Black Forest Labs, creators of Stable Diffusion, unveil Flux, a 12 billion parameter open-source text-to-image model. Flux delivers visuals rivaling top models like Midjourney, offering three versions: Flux Dev, Flux Schnell, and Flux Pro. Available on Hugging Face and free to test on Replicate.com. Highlights: • Flux: 12 billion parameter text-to-image model. • Outperforms Midjourney v6.0, Dall-E 3 (HD), SD3 Ultra. • Three versions: Dev (non-commercial), Schnell (Apache 2 license), Pro (API-based). • Best in visual quality, prompt accuracy, size variability, typography, output diversity. • Flux Dev and Flux Schnell downloadable on Hugging Face. #AIArt #ImageGeneration #BlackForestLabs #OpenSourceAI #FluxModel