Show pageOld revisionsBacklinksFold/unfold allBack to top This page is read only. You can view the source, but not change it. Ask your administrator if you think this is wrong. ====== AI Tools ====== ===== Model DNA ===== Generative AI models differ based on: *Training distribution (dataset it was trained on) *Internal aesthetic priors (different creative directors) *Prompt parsing philosophy (composition keywords vs materials / lighting vs narrative metaphors) ==== Atlas ==== === Base Prompt === <code txt> an old worn book with a thick leather cover, slightly weathered edges and visible page layers, simple embossed details on the cover, resting on a flat surface, closed, low angle shot, realistic materials, </code> <code txt> soft ambient light, clean composition, balanced and readable structure, neutral background, </code> Then build modular prompts (base structure): ''[subject] + [composition] + [lighting] + [style intent] + [constraints]'' * Keep 70% identical across tests. * Run cross-model batches. Run on all models & Do NOT tweak per model === Styles === * **Fantasy :** stylized fantasy art * **Cinematographic :** cinematic lighting, dramatic shadows, high contrast * **Hyper Realistic :** ultra realistic, photorealistic, natural lighting, high detail textures * **Illustration :** stylized illustration, hand-painted look, simplified shapes * **Bright / whimsical :** vibrant colors, soft light, whimsical atmosphere * **Graphic / Design :** minimalistic, flat lighting, clean shapes, product design aesthetic * **Painterly / Fine Art :** oil painting style, visible brush strokes, artistic texture, classical painting * **Technical / Scan-Like :** neutral lighting, flat illumination, orthographic feel, no shadows, documentation style * Alternatives : Abstract / symbolic and Product-like / clean === Light === * **Soft :** soft natural lighting * **High Contrast :** strong directional light, deep shadows, high contrast lighting * **Volumetric / fog :** volumetric lighting, light rays, subtle fog, atmospheric depth * **Neon / emissive :** neon lighting, emissive glow, colored light accents, subtle bloom * **Low key / dark / moody :** very low light, dark environment, minimal illumination, moody shadows * **Overexposed / bright :** bright lighting, soft overexposure, high key, washed highlights === Complexity === Keep the core part : <code text> an old worn book with a thick leather cover, slightly weathered edges and visible page layers, simple embossed details on the cover, closed, low angle shot, realistic materials, cinematographic, soft natural light, </code> * **Isolated** * **Structured composition :** carefully arranged objects, balanced composition, clear spacing, readable layout * **Multi-object interaction :** natural arrangement, slight overlap, coherent composition, readable subject * **Dense environment :** rich environment, layered elements, moderate visual density, subject remains identifiable, depth * **Controlled clutter :** cluttered desk, uneven arrangement, overlapping objects, semi-chaotic but still readable * **Chaos / fragmentation :** chaotic scene, scattered objects, fragmented layout, unbalanced composition, visual disorder, reduced readability Keep a single style (cinematographic, neutral) and probe through the different layers of complexity. Look for : * **Subject priority :** Does the book stay the focus? Or does it get lost? * **Spatial intelligence :** believable placement? gravity respected? overlaps make sense? * **Composition control :** does it still “frame” the image? or just fills space randomly? * **Noise management :** clean vs messy ? intentional vs accidental clutter ? === Conflict tests === * **Structural conflict :** closed book, visible open pages * //Tests whether the model protects object logic.// * **Compositional conflict :** chaotic scene, clean composition * //Tests whether layout instructions override entropy.// * **Visual hierarchy conflict :** dense environment, subject remains dominant, reduced emphasis on subject * **[[ai:atlas-prompts|Atlas Prompts]]** ==== Model List ==== * **[[ai:nanobanana|nanoBanana 2]]** * **[[ai:flux2|Flux Pro 2]]** * **[[ai:firefly5|Firefly 5]]** * **[[ai:runway4|Runway Gen4]]** * **[[ai:gptimage1p5|GPT Image 1.5]]** * **[[ai:ideogram3|Ideogram 3]]** * **[[ai:midjourne|MidJourney]]** ===== Model specific Workflows ===== ***[[ai:midjourney|MidJourney]]** ===== Other Models ===== ***[[https://runwayml.com/|Runway]]** : Video generation and Manipulation ***[[https://www.latted.com/|Latted]]** : pay-as-you-go video models. Unified timeline and tools ***[[https://www.openai.fm/|OpenAI FM]]** : Text to speech+ ===== Image Generating ===== * **Scene:** Describe the environment, main objects, and their relationships. * **Focus:** Mention what should be most visible or important in the frame. * **Style:** Specify realism, lighting, camera angle, color tone, etc. * **Purpose:** Explain the context or use (e.g., documentation, tutorial, UI illustration). * **Reference Images:** Mention if you have any uploaded files to guide style, layout, or color. ===== Ratio References ===== * **Ultra wide Displays:** 3440x1440: 43:18 * **S23 Ultra:** 1440x3088: 9:19.5 or 9:20 ===== Camera Views ===== ^ **Eye-Level Shot** ^ **Low-Angle Shot** ^ **High-Angle Shot** ^ **Bird's Eye View** ^ | {{:ai:eye_level_shot_-_old_wise_wizard.png?direct&200|}} | {{:ai:low_angle_shot_-_victorious_sorceress.png?direct&200|}} | {{:ai:high_angle_shot_-_lost_fairy.png?direct&200|}} | {{:ai:bird_eye_view_-_ancient_ritual_circle.png?direct&200|}} | | //**Neutral, natural** perspective. Great baseline// | //Makes subject feel **powerful** or imposing// | //Makes subject feel small or **vulnerable**// | //Strong composition, **look down** on subject// | ^ ^ ^ ^ ^ ^ **Worm's Eye View** ^ **Over-the-Shoulder Shot** ^ **Close-up** ^ **Extreme Close-up** ^ | {{:ai:worm_eye_view_-_majestic_wizard_academy.png?direct&200|}} | {{:ai:over_shoulder_shot_-_wizard_and_balrog.png?direct&200|}} | {{:ai:close_up_-_elven_lady.png?direct&200|}} | {{:ai:**extreme_close_up_-_ancient_gem_in_hilt_of_sword.png?direct&200|}} | | //More extreme than low angle, **very stylized**// | //Great for **storytelling** and interaction// | //**Emotion**, detail, texture// | //Hyper focus on a **detail**// | ^ ^ ^ ^ ^ ^ **Medium shot** ^ **Full shot (full body)** ^ **Wide shot** ^ **Extreme wide shot (establishing shot)** ^ | {{:ai:medium_shot_-_tavern_keeper_pouring_beer.png?direct&200|}} | {{:ai:full_shot_-_female_viking_warrior.png?direct&200|}} | {{:ai:wide_shot_-_lone_wizard_in_desert.png?direct&200|}} | {{:ai:extreme_wide_shot_-_village_in_the_mountains.png?direct&200|}} | | //Waist-up framing, **very stable** composition// | //Shows **character + posture**// | //Subject in **environment**// | //Environment **dominates**// | ^ ^ ^ ^ ^ ^ **Dutch angle (tilted shot)** ^ **Isometric view** ^ **POV shot (first-person)** ^ **Macro shot** ^ | {{:ai:dutch_angle_-_gothic_manor.png?direct&200|}} | {{:ai:isometric_view_-_medieval_town.png?direct&200|}} | {{:ai:pov_shot_-_torch_and_battle_axe.png?direct&200|}} | {{:ai:macro_shot_-_dewdrop_pixie_fairy.png?direct&200|}} | | //Adds tension, **instability**// | //Very popular in **generative AI**, super clean results// | //Immersive, **game-like**// | //**Tiny subjects**, strong depth of field effects// | ^ ^ ^ ^ ^ ^ **Aerial Shot** ^ **Tracking shot (implied motion)** ^ ^ ^ | {{:ai:aerial_shot_-_hobbit_path_to_lonely_mountain.png?direct&200|}} | {{:ai:tracking_shot_-_runner_in_medieval_city.png?direct&200|}} | | | | //Less vertical than bird’s-eye, more **cinematic landscape**// | //Works **surprisingly** well even in stills// | | | ===== Style ===== * Photographic / Photo realistic * Anime * Neon Punk * Oil Paint * Watercolor * Fantasy Art * Pencil Sketch * Chalk Pastel * Digital Art ---- ==== Subject ==== Describe the subject in detail. Shape, size, how things are positioned. ---- ==== Colors ==== Specify the desired dominant colors or mood board. ---- ==== Materials ==== Describe materials. ---- ==== Viewing Distance ==== * Extreme Long Shot * Very Long Shot * Long Shot/Wide Shot * Full Shot * Medium Long Shot * Medium Shot * Medium Close-up Shot * Close-up Shot * Extreme Close-up Shot ---- ==== Viewing Angle ==== * Top-Down View (90°) * High Angle View (45°) * Eye-level View (0°) * Low-angle View/Hero Shot (-15°) * Extreme Low-angle View/Worm's eye View (-60°) ---- ==== Focal Length ==== * **18mm :** Super wide, capture a lot in the frame. Good for big landscapes or tight spaces. Adds depth and slight warping. * **24mm :** Versatile wide-angle view, great for portraits, street scenes, and landscapes. More background than a standard lens, less distortion and less warping. * **35mm :** Natural perspective closer to the human eye. Ideal for street photography or environmental portraits, where subjects interacts with their surroundings. * **50mm :** Standard and versatile. Very close to the human eye - natural and familiar feel. * **85mm :** Zooms in a bit, narrower view - flattering perspective for portraits. Minimizes distortion - popular choice for capturing people and food close-ups without warping the subject. * **105mm :** Telephoto - great for portrait photography, flattering look. Macro photography - shallow depth of field (blurry background) to isolate the subject. * **135mm :** Very narrow angle of view, nice magnification. last focal length for portrait photography - headshots with smooth blurry background. * **200mm :** Perfect for isolating subjects from a distance. Ideal for wildlife, sports, and astrophotography. ---- ==== Special Lenses ==== * **Fisheye Lens :** Super wide-angle, up to 180 degrees, bending the scene with a curved bubble effect. Trippy, distorted look. * **Tilt-Shift Lens :** Blur parts of the image, creating a “miniature effect” that makes giant buildings look tiny. * **Lensbaby :** Creative blur on most of the image and focus on a specific spot. Dreamy, magical effect. * **Infrared Lens :** Reveal things invisible to the naked eye. Alien-looking landscapes. Mysterious and otherworldly vibes. * **Pinhole Lens :** Soft photos with and a dreamlike quality. Experimental and back-to-basics approach to images. ---- ==== Various ==== * **Short/Fast Shutter Speed :** No blur, quick camera blink. Less light in, darker images. Good for catching splashes, jumps, or fast movements. * **Slow/Long Shutter Speed :** Blurry waterfalls, streaky city lights. Shutter stays open longer, capturing movement and light over time. Good for showing motion or starry night skies. * **Wide Aperture :** (low f-number between f/1.4 and f/5.6). Lots of light in with soft blur effect in the background, perfect for portraits or product pics. * **Narrow Aperture :** (high f-number, like f/11 or f/32). Less light in but everything in focus, from front to back. Perfect for landscapes, buildings, or scenes with lots of details. * **Low ISO :** (ISO 100 or 200) Perfect for bright scenes. Keep photo crisp and clear, no graininess. * **High ISO :** (ISO 800, 1600, or beyond) Capture images in darker environments, sometimes adds noise. ---- ai/start.txt Last modified: 2026/04/09 23:46by mh