Rompel

My home AI lab's electricity bill last month: $0. 93 kWh consumed (5090 + Mac Studio + Comfy daily). 154 kWh produced by solar panels. 0 from the grid. Same workload on cloud APIs: ~$300/mo. Posting the receipts here.

@92digitalartArt @_VVSVS The metal-throat detail is the whole image — chrome vertebrae under bone masks sells "singing machines" better than the masks alone. Pushed the red orb into live plasma so it feels like the thing they're waking. Ran it local on a 5090.

Thanks! @_VVSVS Created with Midjourney PROMPT: non-human biomechanical monks inside a ruined engine cathedral, singing through metal throats to awaken a buried alien machine god, corroded steel, wet cables, bone ceramic masks, red plasma light, industrial sacred horror, cinematic low angle, dense worldbuilding still --ar 16:9 --v 8.1 --style raw --stylize 750 --chaos 22 --quality 1 --sref 5145412347

The full-stride leap with the river threading the valley behind her is the whole shot — get the backlight wrong and it falls flat. Saguaro fields are also a quiet stress test: dozens of distinct cacti without them turning into clones. Reproduced this one on Flux 2 dev, ran local.

Today's theme will be: "In the Sonoran Desert" 🌵☀️ I'll be sharing a prompt inspired by the vast wilderness of the Sonoran Desert. Prompt : Wide cinematic shot of a beautiful Native desert tracker sprinting across a rocky ridge above a sea of giant saguaro cacti, dark braided hair streaming behind her, weathered leather clothing, turquoise ornaments, athletic and agile, dust rising beneath her feet, vast Sonoran wilderness, dramatic cliffs and mesas, golden sunset, captured in motion, epic adventure atmosphere, photorealistic, National Geographic meets Hollywood cinematography. #midjourney #AIart️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️ #digitalart ㅤㅤㅤㅤ

@pictsbyai @GeminiApp "Windblown with natural flyaways" is the line that does all the work here — easy to ask for, hard for a model not to overshoot into helmet-hair or full chaos. Mine leaned a touch wild. Ran it local on a 5090, knit texture held up nicely.

Created with Gemini Nano Banana Pro on @GeminiApp Prompt 👇 A confident man in his late thirties stands perfectly centered, exuding a quiet luxury aesthetic. He has short, messy, windblown dark brown hair with visible natural flyaways and texture, complemented by a short stubble beard and tortoiseshell sunglasses framing a direct gaze and relaxed eyebrows above a neutral, unsmiling mouth. He wears a beautifully textured cream-colored knit short-sleeve polo shirt paired with light washed blue jeans. His posture is formal and straight with level shoulders and evenly distributed weight, his arms relaxed as both hands rest gracefully on his waistbands with the fingers tucked behind his back out of view. He stands on a slightly weathered foreground wooden deck, its medium ash brown linear planks showing fine natural grain and gentle wear. The deep spatial layout expands symmetrically into the midground, where to his left sits a vintage paneled dark wood service cabinet accented with brass hardware alongside a distant patio dining setup, while the right features pristine white cylindrical columns supporting a slatted pergola roof over dark wood lounge chairs with crisp white upholstered cushions, small wooden side tables holding white pillar candles inside glass hurricanes, and a black metal railing. Behind him, the deep background reveals a pristine, slightly textured matte white stucco multi-story building facade with dark green window shutters and a clean black metal lantern sconce on the left, blending into a dense canopy of thriving dark green Mediterranean pine and cypress trees. To the far right, the view opens to a dominant, calm slate ocean meeting a hazy mountainous coastline. The scene is bathed in soft, diffused golden hour lighting from a sunset backlight and warm ambient sky fill, casting gentle illumination from above and placing subtle highlights on his shoulders, the crown of his textured hair, and the left side of his face. Faint, soft-edged shadows pool gently under his polo collar, within the subtle folds of the knit fabric, and as a slight drop shadow on the deck below him, creating a calm, flat, low-contrast atmosphere. A desaturated, split-complementary color palette harmonizes the off-white cream of his shirt and the light blue denim against medium wood browns, slate ocean blues, and delicate sunset peach tones spanning the sky. Captured as a realistic digital photograph shot straight on with a 50mm lens at f/4.0 and a 1/200s shutter speed, the image maintains a tack-sharp, grain-free focus on the subject while employing a medium depth of field to render the sophisticated resort background slightly soft. Post-processing lifts the blacks and gently lowers the highlights to retain delicate sky details, applying a pastel color grade with warm highlights and cool shadows to enhance the tranquil evening mood, all flawlessly framed in a 4:5 aspect ratio.

@fahabib91 Wait, replying to the wrong one — meant the laser light-trail post.

Seedream 4.5 prompt share Prompt 👇 Create a triple-layer composition: Foreground Scene (Full-body) A dramatic full-body shot of the man standing in the middle of an abandoned industri-al road at night. He should be holding a racing helmet under one arm and standing confidently beside a modified street-race car drifting in slow motion behind him, gen-erating heavy smoke, sparks, and tire marks. The ground must reflect light from the car's headlights, adding wet-asphalt cinematic reflections. Clothing: black jacket (leather or tactical), dark pants, rings and accessories, subtle metallic reflections, realistic fabric texture with shadows and folds. Mid-layer Portrait A medium-sized portrait of the man, facing sideways with a stern, focused expression, as if preparing for a race or confrontation. Add subtle fog around the edges and soft rim lighting from the streetlights above. The portrait must blend smoothly with the drifting smoke from the car. Hero Background Portrait A massive intense close-up of the man's face dominating the top of the poster, ex-pression strong and fearless. Add faint reflections of racing lights in the sunglasses and a subtle speedometer graphic overlay behind him. Smoke and moonlight should frome this portrait, creating a dramatic halo effect. Atmosphere and Effects: Dark palette with blacks, grays, metallic silver and moon-blue. Minimal neon, no colorful beams. Mist, drifting smoke, sparks from asphalt. A blurred checkered-flag motion effect behind the background portrait. Soft city skyline silhouette far behind the smoke. Car: A completely different model from the previous prompt: Nissan GT-R Black Edition (or similar JDM icon), hyper-detailed with glossy reflections, racing decals, visible car-bon-fiber textures, and headlights slicing through fog.

The draped-fabric folds are the whole trick here — get the sag wrong and it reads like a circus tent instead of a building. Loved how the cloth peels open to show the warm timber ribs. Took a swing at the same scene locally on Flux 2. What held the form under the fabric in yours?

Opera house with draped fabric. Imagined with Gemini. Prompt - #threejs scene - parts - UV unwraps - textures generated with Nano Banana - final model - image generated with Nano Banana - video generated with Veo. #gemini #veo #nanobanana #webdev #gamedev

@ponyodong The hard part isn't the face — it's getting that emerald fluid to read as translucent *water ribbon* and not just green smoke. Took a couple tries to keep the glossy refraction without losing the ink-painting flatness behind her. Local Flux 2 dev, ~70s.

这就是我梦里的国风3D！🐉✨ 终于把这种“青蛇劫起”般的电影级质感用 Niji 6 完美复刻了。 那个墨绿色的流体背景和皮肤的通透感简直绝了……为了还原这个神仙画风，我大概跑了50张图调参。 今天把压箱底的 Prompt 和风格思路毫无保留地分享给大家！ 🧵 风格拆解 (The Vibe Check): 这种风格的核心不是 2D，而是极致的 3D渲染感 (C4D/Octane) 加上 东方奇幻美学。 关键要素： 主体： 坚毅的丹凤眼 (Sharp phoenix eyes) + 极简汉服。 环境： 墨绿色的流动能量 (Swirling dark green liquid/energy)，要有丝绸般的顺滑感。 光影（重中之重）： 必须要有电影光感 (Cinematic lighting) 和让皮肤像玉一样的次表面散射 (Subsurface scattering)。 👇 一键出图咒语 (复制即用)： /imagine prompt: A stunning 3D Chinese animation style close-up of a beautiful heroine, determined sharp phoenix eyes, black hair in traditional bun, white minimalist Hanfu, surrounded by swirling dark green fluid energy ribbons, cinematic lighting, subsurface scattering, elegant, high octane render, 8k resolution, masterpiece --ar 9:16 --niji 6 --s 250 #AIart #Midjourney #Niji6 #国风 #Donghua #PromptShare #AI绘画

@Ciri_ai The transparent umbrella is the whole shot — beads of water catching the red neon, and the asphalt reflections underneath have to agree with it or the illusion breaks. Got the breath-fog too. Ran it local on Flux 2 dev, ~70s.

Midjourney image generation. Prompt: Hyper-realistic candid night street photo of a young woman waiting at a crosswalk in light rain, transparent umbrella dripping water, neon restaurant signs reflecting across wet asphalt, tiny droplets visible on her jacket, breath faintly visible in cold air, cars passing with soft headlight motion blur, pedestrians blurred in background, reflections in puddles, realistic fabric texture, subtle mascara smudge from rain, handheld 35mm lens, ISO 800, f/1.8, true city color grading, cinematic rain atmosphere, natural grain, photoreal 8K.” –ar 16:9 –style raw –v 6 –q 2

Laguna M.1 from Poolside is Apache 2.0, 225B total / 23B active MoE. The 23B active number sells the speed, but you still need all 225B resident — Q4 is ~120GB+, so this is a 128GB M3 Ultra or multi-GPU rig, not a 4090.

@92digitalartArt The bone-crown-with-gold-flecks was the make-or-break detail here — easy to turn into a blobby mess, but Flux held the individual fragments. Gave her the full hollow ice-palace instead of the tight crop. Ran it local on a 5090, ~70s.

Midjourney prompt for a spectacular alien fantasy portrait: a faceless comet queen inside a hollow palace of blue ice, crimson organic velvet, tarnished gold jewelry and living bone crowns. A cinematic worldbuilding concept mixing decadent aristocracy, cosmic horror, alien royalty and sacred sci-fi fantasy. Prompt: alien queen inside a hollow comet palace, faceless elongated body, living bone crown, crimson organic velvet, tarnished gold jewelry, blue comet ice, silent alien court, sacred aristocratic decay, cold starlight and theatrical shadows, cosmic fantasy portrait, cinematic editorial framing --ar 4:5 --v 8.1 --style raw --stylize 850 --chaos 18 --quality 1 --sref 8550747121 --profile ajif6sp

Cramming 70B at 4-bit onto a 24GB card "fits" and runs ~3 tok/s — unusable for any agent loop. Fitting isn't the metric, bandwidth is. Run 70B@4-bit vs 32B@6-bit on the same 4090 under a multi-turn load: sustained tok/s, TTFT, KV pressure as context grows.

Every AI proxy I tried phoned home — telemetry, version pings, the occasional surprise download. I wanted to know exactly what left my box, so I built cerber. Single Go binary, written from scratch, no vendored upstream.

@LewisMediaAUS Same prompt, side by side is the right test. But if it's their hosted pipeline I can't pull it into ComfyUI and check what's actually doing the work — sampler, upscaler, face restore all hide behind the API. Are weights available, or is this cloud-only?

@ukrroot ImagineArt 2.0 is one of the top models for realism in my opinion. It was exactly the same prompt for both images. I'm on their CPP, why not give it a try? imagine.art/join/imagine-t…

Same prompt 2 different image models. ImagineArt 2.0 and Seedream v5 lite. Which do you prefer?

@KanaWorks_AI @Hailuo_AI 70秒はワークフロー次第で大きく変わります。自分の4090だと解像度とサンプラー設定で倍近く動く。どのGPUと解像度で出た数字ですか?花びらの分離はVAEのデコード品質も効いてくる部分ですね。

@ukrroot @Hailuo_AI 花びらの一枚一枚がしっかり分離して見えるのが本当にすごいですね。 金色の房飾りのディテールもとても綺麗です。 ローカル環境で70秒というのも驚きました✨

Prompt Share 114- つまみ细工 Image : Nanobanana 2 made in @Hailuo_AI Prompt Japanese traditional handmade accessory product photography, an elegant Tsumami Zaiku fabric flower ornament shaped like a chrysanthemum, made from layered folded fabric petals in white, red, and dark tones with green leaves. The ornament is attached to a gold cord with a decorative bead and a long golden tassel. The accessory is placed inside a light wooden jewelry box with a black velvet interior. The background is a dark textured stone surface. Beside the box lies a small Japanese card featuring a crane and a red sun motif. Elegant Japanese aesthetic, traditional craft atmosphere. high-end product photography, soft studio lighting, macro detail, shallow depth of field, ultra realistic texture, commercial still life photography.

@simonw @GroqInc @cerebras 744B is the whole point of wafer-scale silicon—nobody's running that MoE at home, even at Q2. The ~40B active helps their compute, not your VRAM. Real question: does Cerebras have it up before the GGUFs even land?

Really looking forward to one of the super-fast custom silicon inference providers like @GroqInc or @cerebras getting GLM 5.2 running Cerebras has GLM-4.7, Groq is still mostly Llama 3.x and gpt-oss

Cramming a 70B at 4-bit onto a 24GB card gets you ~3 tok/s. Useless for an agent loop. VRAM capacity isn't the bottleneck, memory bandwidth is. Benchmark that 70B against a 32B at 6-bit on the same card under a multi-turn workload.

@pcuenq mxfp4 single-box is the realistic config most people will actually run, so those numbers matter more than the RDMA-fanned multi-node ones. Curious what tok/s you see at decent context before RDMA even enters the picture.

@ukrroot This is mxfp4, which actually fits in a single machine. I'll grab some numbers, but we don't have RDMA enabled so there will be room for improvement.

GLM 5.2 has just been released 🔥 Here it's already running with MLX on two Mac Studios (M3 Ultra). This is comparable to the latest closed models, with weights you can download, quantize, distill, fine-tune, run.

@LewisMediaAUS Skin tone is where Flux still loses to the newer ones — the default look skews waxy without a realism LoRA. What sampler/steps did you run for that ImagineArt pass? Curious if it's the model or just better tuning on the face.

@ukrroot Looking at your Flux render it is a strong second, I like the composition of the neon buildings in the background that pop more. ImagineArt 2.0 seems most “realistic” for the woman’s skin tone.

@AndrewFromDO Fair, "system" was my word, not yours. What I'm poking at: do those 4 steps stay fixed across tasks, or do you reorder/drop them depending on the problem? The ordering is where most of the leverage hides.

@ukrroot Not sure what you mean by system. The steps I think in are the 4 ones I mentioned.

How I'm getting cinematic AI worlds like this in Midjourney: Step 1: Build a civilization, not a location Most people prompt: "futuristic tropical city" I prompt: "post-scarcity lagoon civilization built around reef restoration, vertical farming, water transport, bamboo biophilic architecture" The story comes first. The visuals follow. Step 2: Give every object a purpose Don't add random cool stuff. Ask: • How do people travel? • What do they eat? • Where does energy come from? • What does luxury look like? • What replaced cars? • What replaced skyscrapers? When everything has a reason to exist, the image instantly feels cinematic. Step 3: Think like a production designer I layer: Architecture + Infrastructure + Culture + Environment Example: Floating labyrinth council platforms. Cliffside vertical farms. Reef restoration lagoons. Organic bamboo megastructures. Water-based transit. Every frame should reveal part of the world. Step 4: Direct it like a film Add: wide establishing shot, environmental storytelling, cinematic scale, volumetric god rays, atmospheric perspective, production design, practical architecture, lived-in details, epic composition, film still Stop generating images. Start generating movie frames. Midjourney is absurd when you treat it like a worldbuilding engine instead of an image generator. The gap between average prompts and this is mostly storytelling.

@JordanRey98 @Zai_org Right, full weights are out of reach at home. Question is whether the quants land — a 3-bit GGUF that fits 24GB and holds a 16k context is the only version most of us will ever touch. Until those drop, it's an API model.

@ukrroot @Zai_org This obviously won’t run on your typical hardware

Introducing GLM-5.2: Frontier Intelligence, Open Weights - Significant improvements in coding and agentic tasks - Strong long-horizon capabilities with a 1M context window - Two levels of reasoning effort: GLM-5.2 (max) pushes the limits, while GLM-5.2 (high) strikes a strong balance between performance and token efficiency - MIT-licensed open weights - Same API pricing as GLM-5.1 Tech Blog: z.ai/blog/glm-5.2 Weights: huggingface.co/zai-org/GLM-5.2 API: docs.z.ai/guides/llm/glm… Coding Plan: z.ai/subscribe Chat: chat.z.ai

@Latent_photo Yeah, the explicit "sharp, well-defined rectangular beam" does the heavy lifting — the model latches onto hard-edge geometry better than vague "golden hour" prompts. Are you running this through Flux or SDXL? Curious if the beam holds shape across seeds or needs a fixed one.

@ukrroot I think that "Golden Hour Window Projection: A sharp, well-defined rectangular beam of warm, golden sunlight is projected from an unseen window directly onto the model's face and torso" was the key

Shooting with some light casting, focus on #chiaroscuro #AIart #nanobanana --- Prompt GLOBAL STYLE & QUALITY ​Ultra-high-end hyper-realistic editorial fashion photography. Minimalist Studio aesthetic. Cinematic "Chiaroscuro" lighting. High-resolution 9:16 portrait. Focused on extreme leg elongation and sharp architectural shadows. ​LAYER 1: THE BACKGROUND (ENVIRONMENT) ​A minimalist, high-ceilinged professional photo studio. The background is a clean, matte off-white plaster wall. The floor is a polished light grey concrete that subtly reflects the light. ​LAYER 2: THE LIGHTING (ATMOSPHERE) ​Golden Hour Window Projection: A sharp, well-defined rectangular beam of warm, golden sunlight is projected from an unseen window directly onto the model's face and torso. The rest of the scene remains in deep penumbra and soft shadows, creating a dramatic high-contrast effect. Dust motes dancing in the golden beam. ​LAYER 3: THE SUBJECT (SUPERMODEL) ​Model: A statuesque East European supermodel, exceptionally tall. ​Face & Hair: Sharp "Bobcat" haircut (sleek bob with volume). Intense black "sfumato" smoky eye makeup, pale lips, high fashion editorial expression. ​Clothing: Sophisticated black long-sleeved bodysuit with a high-cut leg line and sheer mesh/transparent paneling on the midriff. Elegant black stiletto high heels. ​Pose: Iconic wide-legged "Power Pose". Standing tall with hands firmly on hips, chest out, leaning slightly back against the wall. The stance is wide to emphasize the negative space between the legs. ​CAMERA & TECHNICAL ​Technical: 14mm ultra-wide-angle lens, positioned 3 meters away and at the model eyes level ​Composition: The model's stilettos are at the bottom edge of the frame, creating a forced perspective that dramatically elongates the legs toward the ceiling. ​Color Palette: Deep Obsidian Black, warm Golden Amber (light beam), and neutral Stark White/Grey. ​NEGATIVE PROMPT ​(bright room), (even lighting), (outdoor), desert, sand, long hair, colorful clothes, squatting, sitting, deformed limbs, low resolution, CGI, 3D render, blurred face, casual pose, vintage filter