Stefan Tippelt

@spogelsemaskine @anloremi Love this, make cables a feature 🔥

A little tech behind the screens of my LED physicals 🤌 🤓 for my series "Cathode Rain"

Back in the terminal after setting things up. Starting with fresh ideas. Experiments incoming 🦾 First Latent Locus waves loading soon. #AIArt #StableDiffusion #GenerativeArt #LatentLocus #NFT

@dangreenheck Amazing! 🤩

Finally fixed the refraction in Three.js Water Pro. I wasn't properly scaling the refraction strength based on distance from the water surface. The parts of the boat closest to the water should be refracted less.

@Pixel0Symphony Love it, reminds me of my speakers

@beeple finally some real art in Berlin 🐶💩🖼️

the dogs are headed to BERLIN this spring!!🦾

@carolletta Usually I start small and then keep adding more detail. Later I start consolidating and try to keep only the most relevant parts ✨

Hey creatives! What type of prompts do you prefer? Shorter and open to possibilities or longer and more descriptive?

@kimasendorf Congrats @kimasendorf you are setting the bar 🔥

PXL POD has minted out. I am sorry there was nothing left for public minting. I shouldn't have announced it in retrospect. My fault.. Demand was high and I couldn't live up to the expectations of many. But eventually it is my art and I have to sell it the way I want. Apologize again.. And thanks again to all collectors.

@r0bp Super nice

insight

@thepropgallery Count me in 🙏

Art donation!!! I’m giving away pieces to people who’d appreciate them Comment below if you’d like some 🙏

@sares_at Congrats, looks amazing 🤩

GM🪷 Anima from my Aura Series exhibited at Max Center in Spain, curated by SMTH.

Flux textures on fire 🦾✨ Molten black-gold luxury vs. deep navy-teal cosmic flow—both gorgeous, but composition still fights me (high CFG makes it rigid). Which vibe grabs you more? Gold or teal? 👇 #AIArt #FluxAI”

New Flux batch — chasing the full panoramic evolution 🦾 From Curiosity spark → Understanding build → Mastery emergence. Still iterating, but these feel closer. Which one hits you hardest? 👇 #AIArt #FluxAI #GenerativeArt

@tinoch Hell yes — making art is pure badass energy 🦾 Especially when you lose yourself in the flow. Keep creating! @tinoch

Making art is just so badass😈

@RogerDickerman Best subscription you’re gonna get this year — buckle up! 🦾🔥 Minted and loving it already 🙏 @24_Hours_Art

TWO HOURS LEFT! The Sunday morning surge has been amazing to see 🔥 ART Mint (327/380): transient.xyz/mint/24-hours-…

@RaoulGMI @NedRyersonBing Haha, same — almost bailed too, but glad I didn't! 🦾 Late mint gang strong 💪 @RaoulGMI @NedRyersonBing

@NedRyersonBing Better late than never I guess...

Minted 🫡

@Inspector_9 @haydclay Wow, that’s incredible! 🤩

Sunset through the eyes of @haydclay

Date night at Time Out Market — bottle of sparkling + bottle lid = physical Latent Locus prototype 🦾 Even relaxing, the renders don't stop. Built this 2-layer model with fingernails as tools 😉 Latent ideas in real life. #AIArt #GenerativeArt

@VanArman @DagieDee Thanks Pindar — loved Part 1 🦾 Funny parallel: it took centuries for landscapes to stand alone without figures... now screens are doing the same, digitally. (Still treasure that postcard you sent last year 🙏) @VanArman

Excellent read from @DagieDee - recommended

@clairesilver @degens @VanArman @hollyherndon @genekogan @quasimondo @sashastiles Gene's 2017 live CNN demo was mind-blowing 🦾 It directly inspired my own Deep Dream experiments shortly after (proof from back then below). x.com/steftip/status… Early AI art had real soul. @clairesilver @genekogan

@degens Partial to mine, but also @VanArman, Robbie Barrat, @hollyherndon, @genekogan, @quasimondo, @sashastiles for 2018/early 2020s Tons of AI-Collaborative artists 2022-now as well. Still will be early for the rest of this decade, big picture/relatively speaking.

Random normies talking about “I asked gpt” “I vibe-coded with Claude” and you still think early AI art isn’t valuable They’re Paleolithic artifacts and this is completely obvious if you have any foresight whatsoever

Pro tip: Front-load critical subject/style in first 77 tokens for CLIP. Better prompts = better renders. More experiments coming—share your tips! 👇

Optimization trick: Pass negative prompts separately. They don't consume your 512-token budget for the positive prompt. This lets you use the full 512 tokens for what you WANT, while still having rich negative guidance. ✨

🧵 Prompt engineering tips for FLUX.1-dev — hard-earned lessons 🦾 FLUX uses dual encoders: • T5 (primary): 512 token limit • CLIP (secondary): only sees first 77 tokens Key insight below 👇 #FluxAI #AIArt