هوش مصنوعی | والپیپر | Taswirgar | تصویرگر

PROMPT ENGINEERING MIDJOURNEY – CHEAT SHEET -Part 3- Style References & Image Prompts WHAT IS STYLE REFERENCE? Style reference = using an image URL to copy its style. It transfers the look, mood, and aesthetic to your new image. Works with both Discord and API. SYNTAX --sref [image URL] Example: --sref https://example.com/style.jpg STYLE WEIGHT Controls how strong the style applies. --sw [number 0 to 1000] --sw 100 = default --sw 500 = strong --sw 1000 = maximum QUICK EXAMPLES Style only: cyberpunk city --sref https://example.com/neon.jpg Style + weight: fantasy castle --sref https://example.com/painting.jpg --sw 700 Style + multiple references: portrait --sref url1.jpg url2.jpg url3.jpg WHAT IS IMAGE PROMPT? Image prompt = using an image as reference for content. It tells Midjourney what to generate based on the image. Different from --sref (style only). SYNTAX [image URL] + your prompt Example: https://example.com/photo.jpg a warrior in the same pose IMAGE WEIGHT Controls how much the image affects the result. --iw [number 0 to 3] --iw 0.5 = low influence --iw 1 = default --iw 2 = high influence --iw 3 = maximum DIFFERENCE BETWEEN --sref AND IMAGE PROMPT --sref = copies style (colors, lighting, mood) Image Prompt = copies content (subject, pose, composition) You can use both together! FULL EXAMPLE PROMPT (v8.2) https://example.com/pose.jpg a dark knight with glowing sword in gothic castle --sref https://example.com/painting.jpg --sw 600 --iw 1.5 --ar 16:9 --v 8.2 --no blurry Breakdown: Image URL = pose reference Prompt = describes subject --sref = style reference --sw 600 = strong style --iw 1.5 = medium image influence --ar 16:9 = wide ratio --s 400 = high stylization --c 30 = moderate variation --v 8.2 = latest model --no blurry = remove unwanted PRO TIPS Tip 1: Use high-quality images for better results. Tip 2: Start with --sw 300 and --iw 1. Adjust from there. Tip 3: Multiple --sref images = mix styles. Tip 4: Image prompt works best with clear, simple subjects. Tip 5: Short prompt + strong style = more creative freedom. AVOID THESE MISTAKES Using low-resolution images Forgetting --sw when using --sref Using --iw without an image URL Using too many references (3 max is safe) QUICK REFERENCE --sref = style copy --sw 0-1000 = style strength Image URL + prompt = content copy --iw 0-3 = image strength GOLDEN RULE Style reference for look, image prompt for content. Combine both for maximum control. #AI #PromptEngineering #ImageAI : #AIImageGeneration #MidJourney #CheetSheet Join Us @Taswirgar

High Quality File

Prompt: a minimalist abstract image divided horizontally. the left side is a deep blue field with a thin, glowing light blue horizon line. in the center, this line transforms into a sweeping, curved spectrum of light that bends upward towards the upper right corner. this spectrum transitions through yellow, orange, pink, light blue, and violet. below the horizon line on the right, the colorful light is reflected in a soft, diffused gradient, fading into a dark blue violet shade at the bottom. Model: MidJourney 8.2 #Ai #Phone #Wallpaper : #Abstract #Blue #Orange #Glow #Dark Join us @Taswirgar

High Quality File

Prompt: wavy black and white line art abstraction of sawtooth mountains. --profile ebzanvj --hd Model: MidJourney 8.2 #Ai #Phone #Wallpaper : #Mountain #Mount #Black ##LineArt #Illustration Join us @Taswirgar

PROMPT ENGINEERING MIDJOURNEY – CHEAT SHEET -Part 2- Parameters List --ar or --aspect : aspect ratio of the image --c or --chaos : how different the 4 results are from each other, 0 to 100 --s or --stylize : how creative and artistic the output is, 0 to 1000 --q or --quality : rendering quality and time, values 0.5, 1, or 2 --v or --version : which model version to use, like 6, 6.1, 5.2, or niji 6 --seed : fixed number to get repeatable similar results --no : remove unwanted elements, like --no cars or --no text --iw : image weight, how much the reference image affects the result, 0 to 3 --style : choose specific style mode, like raw or expressive --tile : generate tileable image for seamless patterns and wallpapers --weird or --w : make results unusual and experimental, 0 to 3000 --repeat or --r : how many times to run the same prompt, up to 40 --raw : maximum prompt adherence with minimal AI interpretation --profile or --p : use someone else's saved style profile --sref : style reference from an image URL for style transfer --sw : style weight, how strong the style reference applies, 0 to 1000 --cref : character reference from an image URL for character consistency --cw : character weight, how strong the character reference applies, 0 to 100 --oref : object reference from an image URL for object consistency --ow : object weight, how strong the object reference applies, 0 to 100 --sv : style version for niji model --video : record and download the generation process video --motion : motion intensity for video output --loop : loop the video seamlessly --end : specify which frame to end the video at --fast : fast mode, uses more fast hours, priority speed --relax : relaxed mode, slower but saves fast hours --turbo : turbo mode, fastest generation, consumes more credits --public : make the image visible to everyone on Discord --stealth : keep the image private, only visible to you --draft : draft mode, half the cost, good for quick testing --hd : older high definition mode, not used with v6 --sd : older standard definition mode, not used with v6 --stop : stop the job at a specific percentage, 10 to 100 --time : set timeout for the job in seconds #AI #PromptEngineering #ImageAI : #AIImageGeneration #MidJourney #CheetSheet Join Us @Taswirgar

High Quality File

Prompt: a minimalist composition showing a small island in calm, clear turquoise water. on the island stands a simple light blue wooden shack with a window and an door. to its right, two palm trees leaves lean slightly . a small white boat is beached on the white sand in front of the house. the scene is set against a plain, light blue sky with a warm, desaturated color palette. --profile pfsaaix --hd Model: MidJourney 8.2 #Ai #Phone #Wallpaper : #Blue #Sea #Ocean #Island #Beach Join us @Taswirgar

High Quality File

Prompt: Three waves coming together. Top view. photo by Hasselblad camera. Cyan pink crystal green. --raw --profile 43rgnhu --profile iygsqqg Model: MidJourney 8.2 #Ai #Phone #Wallpaper : #Wave #Sea #Ocean #Green #Beach Join us @Taswirgar

PROMPT ENGINEERING MIDJOURNEY – CHEAT SHEET -Part 1- STRUCTURE: SUBJECT + DETAILS + ENVIRONMENT + STYLE + LIGHTING + MOOD + COMPOSITION First words = most important. Put subject first. REAL PERFECT PROMPT EXAMPLE: "an old weary samurai with a missing left eye, wearing tattered crimson armor and holding a cracked katana, standing in a rain-soaked neon alley in cyberpunk Tokyo, dramatic cinematic style, moody blue and pink lighting, epic and gritty atmosphere, low angle wide shot" Breakdown: Subject = old weary samurai Details = missing left eye, tattered crimson armor, cracked katana Environment = rain-soaked neon alley in cyberpunk Tokyo Style = dramatic cinematic Lighting = moody blue and pink Mood = epic and gritty Composition = low angle wide shot SUBJECT: Be specific. Not "woman" → "old fisherman with grey beard and wrinkles" Not "building" → "abandoned gothic cathedral with broken windows" DETAILS: Age, scars, clothing, posture, movement, expression. "tired samurai holding cracked sword, torn red armor, blood on shoulder" ENVIRONMENT: Where? Indoor/outdoor/fantasy. "neon Tokyo alley", "misty forest", "floating islands" STYLE KEYWORDS: Photorealistic, cinematic, oil painting, watercolor, anime, 3D render, concept art, vintage, minimalist, surreal, Pixar style, ink sketch LIGHTING: Golden hour, neon, studio, moonlight, backlight, dramatic, natural, colored stage lights MOOD: Epic, dark, dreamy, tense, peaceful, mysterious, melancholic, gritty COMPOSITION: Close-up, wide shot, low angle, bird's eye, over-the-shoulder Centered, rule of thirds, shallow depth of field WEAK WORDS TO REMOVE: very, really, quite, somewhat, maybe, like, kind of STRONG WORDS: Use detailed instead of beautiful Sharp instead of good Intricate instead of nice Vibrant instead of colorful COLORS: Be specific. Not blue → navy, cyan, cobalt, midnight Not red → crimson, scarlet, burgundy SHORT PROMPT = more freedom for AI LONG PROMPT = more control for you MIX STYLES: photorealistic + anime cyberpunk + gothic watercolor + sci-fi TEXTURES: rough, smooth, glossy, rusty, cracked, polished, weathered, silky, metallic ENERGY: Dynamic = running, jumping, flying Static = standing, sitting, resting CHECKLIST BEFORE SEND: Subject first? Details included? Environment clear? Style defined? Lighting described? Weak words removed? GOLDEN RULE: No perfect prompt. Generate, learn, adjust, repeat. #AI #PromptEngineering #ImageAI : #AIImageGeneration #MidJourney #CheetSheet Join Us @Taswirgar

High Quality File

Prompt: center view of a bluestone-paved chinese Yihe Lane. Yihe Lane is a bustling street alive with the hum of daily life, flanked by weathered old residential buildings;old Chinese neighborhood, the ground floors house various snack shops,everyday life, power lines crisscross,Utility pole, no-entry sign. mottled walls, clothes hanging out to dry，Warm, blue sky white clouds, illustrations,china,book cover art, brushstroke,watercolour,childhood,4K resolution. --sref 1564222235 --sref 194990917 --sref 2457848750 Model: MidJourney 8.2 #Ai #Phone #Wallpaper : #Colorful #Street #Animation #DownTown #City #Urban Join us @Taswirgar

High Quality File

Prompt: a tiny person making a single dark line with paintbrush on their huge canvas among thousands of other small lines and dashes to make up the full painting, trendy scratchy minimal illustration combining influences from jean jullien drawing style. --ar 9:20 Model: MidJourney 7 #Ai #Phone #Wallpaper : #Cream #Choas #Man Join us @Taswirgar

High Quality File

Prompt: 1985 low-budget European dark fantasy film still, [NOTICE SIZE HERE] cinematic shot. Wide long shot from behind: a cloaked wizard walks down a muddy path through a dense foggy forest, using a gnarled wooden staff as a walking stick. A crumbling stone castle looms in the misty mid-ground. Theatrical stage lighting with deep chiaroscuro shadows, soft optical bloom on highlights, peeling plaster walls on ruins, practical creature suits and hand-built costumes visible in frame, no CGI, grainy 35mm film texture, muted earthy palette with cold blue-grey fog, 4K high detail, photorealistic vintage film grain notice size of image while using prompt Model: GPT image 2 #Ai #Phone #Wallpaper : #Wazard #Old #Movie #Dark #Castle Join us @Taswirgar

Reduce Token Usage with TOON Format Every time you work with language models, you pay per token. JSON has lots of redundant data that just burns tokens. TOON displays the same data with a more compact structure. Without losing any information, it cuts token usage by 30-60%. JSON example: {"users":[{"id":1,"name":"Alice"},{"id":2,"name":"Bob"}]} Same data in TOON: users[2]{id,name}: 1,Alice 2,Bob This isn't just for text models. Image generation models like GPT Image 2 also price by token. So TOON works there too. JSON example for image: {"prompt":"a cyberpunk city with neon lights","width":1024,"height":1024,"quality":"high"} Same in TOON: image{prompt,width,height,quality}: "a cyberpunk city with neon lights",1024,1024,high Online Converters: jsonformat.org/json-to-toon toonconverter.com json2toon.de toonifyit.com All free, no data sent to server. Turning Regular Prompts into TOON If you're looking for ways to convert normal prompts to TOON, there are three alternatives: 1. Text to Prompt Browser Extension This Chrome/Firefox extension grabs text from any page and automatically converts it to optimized prompts for AI . It supports direct conversion to TOON (spec v1.5) and is optimized for LLM use. 2. toonify Library (Professional Approach) For developers, the toonify Python package has a generate_structure() function that creates TOON templates from a simple schema. Instead of writing example data, you just define the requested structure and it handles the conversion: from toon import generate_structure schema = { "name": "person's name", "age": "person's age" } structure = generate_structure(schema) # Output: name: <person's name> # age: <person's age> This saves tokens and keeps your prompts cleaner. 3. Convert with AI Itself (Easiest Way) If you don't want to use any of the above, just tell the model: "Convert this prompt to TOON format" Show it the structure. Models handle this well with a simple system prompt like "You are an intelligent model specialized in converting natural language text into valid TOON format". Smaller prompts = lower costs + faster responses #AI #PromptEngineering #LLM : #AIImageGeneration #JSONtoTOON #toonify #TOON #TokenReduction Join Us @Taswirgar

AI OCR + Humanization + Tone Analysis How this skill works A single-step skill for any image or raw text — delivers clean, humanized output with tone analysis: · Image or Text Input → upload an image with text, or paste raw/messy text · OCR Extraction → extracts all text from images with high accuracy, auto-detecting language and script · Humanization → transforms text into natural, fluent language (fixes OCR errors, punctuation, spacing, awkward phrasing — meaning stays intact) · Tone Analysis → short summary in English covering formality, emotion, and intent · Multi-Language → works with any language or mixed-language content No installation, no account, no extra app. Just this conversation. Prompt: TXT or Markdown file Model: Any LLM with image support (ChatGPT, Claude, etc.) #AI #LLM #Skill : #OCR #Humanization #ToneAnalysis #Multilingual Join us @Taswirgar

High Quality File