{"passport":{"unfragile":{"@version":"1.0","version":"2026-05","artifact":{"id":"draw-things","slug":"draw-things","name":"Draw Things","type":"app","url":"https://drawthings.ai","page_url":"https://unfragile.ai/draw-things","categories":["image-generation","model-training"],"tags":[],"pricing":{"model":"free","free":true,"starting_price":null},"status":"active","verified":false},"capabilities":[{"id":"draw-things__cap_0","uri":"capability://image.visual.local.text.to.image.generation.with.metal.accelerated.inference","name":"local text-to-image generation with metal-accelerated inference","description":"Generates images from natural language prompts by executing Stable Diffusion and FLUX models directly on Apple Silicon devices using Metal GPU acceleration, eliminating cloud dependency and network latency. Models are downloaded once and cached locally, enabling offline generation after initial setup. The Metal acceleration framework optimizes tensor operations and memory bandwidth for M-series chips, delivering generation times measured in minutes per image on consumer hardware.","intents":["Generate images locally without sending prompts to cloud services for privacy","Create artwork iteratively without network latency or cloud API rate limits","Run image generation on offline devices or in environments without internet connectivity","Avoid cloud compute costs by leveraging local Apple Silicon hardware"],"best_for":["macOS and iOS users with Apple Silicon devices (M1/M2/M3/M4 chips)","Privacy-conscious creators unwilling to send prompts to cloud services","Individual artists and designers prototyping concepts locally","Users in regions with limited cloud service availability"],"limitations":["Apple Silicon only — no Windows, Linux, or Intel Mac support documented","Generation speed varies significantly by device model and available VRAM; no published benchmarks provided","Models must be downloaded and stored locally before use; typical model sizes 2-7GB per model","Maximum concurrent generations limited by device memory; no batch processing API documented","Image resolution and quality constrained by available device memory; maximum resolution not publicly specified"],"requires":["macOS 11+ or iOS 15+ on Apple Silicon device (M1 or newer)","Minimum 4GB free storage for base model, 8GB+ recommended for multiple models","Internet connection for initial model download only; offline generation supported thereafter"],"input_types":["text (natural language prompts)","numeric parameters (guidance scale, steps, seed)"],"output_types":["image (PNG or JPEG format, resolution varies by model and device memory)"],"categories":["image-visual","local-inference"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"draw-things__cap_1","uri":"capability://image.visual.lora.training.and.inference.on.device","name":"lora training and inference on-device","description":"Enables users to train custom Low-Rank Adaptation (LoRA) modules locally on Apple Silicon devices by fine-tuning base models (Stable Diffusion, FLUX) on user-provided image datasets. Trained LoRAs are stored locally and can be applied during inference to customize model outputs without retraining the full base model. The training process uses gradient descent optimization on-device, with inference applying LoRA weights as low-rank matrix multiplications during the diffusion process.","intents":["Train custom style or character LoRAs on personal image datasets without uploading to cloud services","Apply trained LoRAs to generation prompts for consistent character or style reproduction","Iterate on LoRA training locally with immediate feedback without cloud training queues","Maintain proprietary training data and custom models entirely on-device"],"best_for":["Individual creators building custom character or style models for personal use","Teams with proprietary training data unwilling to upload to cloud services","Artists iterating rapidly on custom model training with immediate feedback","Users in privacy-sensitive jurisdictions or organizations"],"limitations":["LoRA training time and resource requirements not publicly documented; likely hours to days depending on dataset size and device","Training dataset preparation and curation responsibility falls entirely on user; no built-in dataset management tools documented","LoRA inference in free tier limited to local generation only; cloud inference with custom LoRAs requires paid tier","Compatibility matrix between LoRA versions, base models, and app versions not documented","No version control or experiment tracking for trained LoRAs; overwriting previous versions not prevented"],"requires":["macOS 11+ or iOS 15+ on Apple Silicon device with 8GB+ RAM recommended","Training dataset of 10-100+ images (exact requirements not specified)","Sufficient free storage for base model plus LoRA weights (typically <100MB per LoRA)","30+ minutes to several hours of uninterrupted device usage for training"],"input_types":["image dataset (JPEG, PNG format; quantity and resolution requirements unspecified)","text labels or captions for training images (optional; mechanism unclear)","hyperparameters (learning rate, training steps, etc.; defaults provided but not documented)"],"output_types":["LoRA weights file (proprietary format, stored locally)","generated images using trained LoRA applied to prompts"],"categories":["image-visual","data-processing-analysis"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"draw-things__cap_10","uri":"capability://image.visual.multi.model.support.with.seamless.switching","name":"multi-model support with seamless switching","description":"Supports multiple image generation models (Stable Diffusion, FLUX, and others) with UI-based model selection, enabling users to switch between models for different generation tasks without restarting the app. Each model is downloaded and cached separately, and the app manages model loading and memory allocation. Implementation uses abstraction layer for model inference to support multiple architectures.","intents":["Use different models for different generation tasks (e.g., FLUX for photorealism, Stable Diffusion for stylization)","Compare outputs from different models on the same prompt","Leverage model strengths for specific aesthetics or content types","Experiment with new models as they become available"],"best_for":["Creators using multiple models for different generation tasks","Users comparing model outputs and quality","Researchers and developers experimenting with different architectures","Users wanting flexibility to choose best model for specific prompts"],"limitations":["Supported models not fully documented; Stable Diffusion and FLUX confirmed, others unspecified","Model switching latency not documented; likely requires model unloading and loading","Memory management for multiple models not documented; unclear if models are unloaded when switching","Model compatibility with features (ControlNet, LoRA, inpainting) not documented","No model benchmarking or comparison tools documented","Model versions and update frequency not documented"],"requires":["macOS 11+ or iOS 15+ on Apple Silicon device","Sufficient storage for multiple models (2-7GB per model typical)","Sufficient RAM for model loading (8GB+ recommended for multiple models)"],"input_types":["model selection (UI dropdown or list)"],"output_types":["image (PNG or JPEG, resolution varies by model and device memory)"],"categories":["image-visual","tool-use-integration"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"draw-things__cap_11","uri":"capability://tool.use.integration.native.ios.ipados.macos.unified.interface","name":"native ios/ipados/macos unified interface","description":"Provides native UI implementations across iOS, iPadOS, and macOS using platform-specific frameworks (SwiftUI, UIKit) rather than cross-platform abstractions, enabling optimized UX for each platform. The unified codebase shares inference logic while maintaining platform-specific UI patterns and capabilities. iOS/iPadOS versions leverage touch input and mobile-optimized layouts; macOS version uses keyboard shortcuts and desktop-optimized workflows.","intents":["Use image generation on mobile devices (iPhone, iPad) with touch-optimized interface","Use image generation on desktop (macOS) with keyboard shortcuts and multi-window support","Sync generation history and settings across Apple devices","Access generation capabilities on any Apple device without platform-specific limitations"],"best_for":["Apple ecosystem users wanting native app experience","Mobile creators wanting to generate images on-the-go","Desktop users wanting native macOS integration","Users wanting seamless experience across multiple Apple devices"],"limitations":["iOS/iPadOS limited to App Store distribution (with direct download fallback mentioned)","macOS version requires native installation; no web version available","Cross-device sync not documented; unclear if generation history syncs via iCloud","No Windows or Linux support; Apple Silicon only","No web interface for remote access or sharing","Platform-specific limitations (iOS background processing, memory constraints, etc.) not documented"],"requires":["macOS 11+ (for macOS version) or iOS 15+ (for iOS/iPadOS version)","Apple Silicon device (M1 or newer for macOS; A14 or newer for iOS/iPadOS)","App Store installation (iOS/iPadOS) or direct download (macOS)"],"input_types":["text prompts (keyboard or voice input on iOS/iPadOS)","touch input (iOS/iPadOS) or mouse/trackpad input (macOS)","image uploads (camera, photo library, or file picker)"],"output_types":["images (PNG or JPEG, saved to photo library or file system)"],"categories":["tool-use-integration"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"draw-things__cap_12","uri":"capability://tool.use.integration.free.tier.with.optional.paid.upgrades","name":"free tier with optional paid upgrades","description":"Offers free local image generation on Apple Silicon devices with limited cloud compute hours (Lab Hours), with optional paid tier (Draw Things+) providing higher cloud compute quotas and custom LoRA cloud inference. Free tier enables full local inference without payment; cloud features are optional and quota-based. Pricing model uses monthly Lab Hours allocation rather than per-request billing.","intents":["Access image generation without payment using local device compute","Upgrade to paid tier for faster cloud generation when needed","Control generation costs predictably using monthly quota system","Use custom LoRAs with cloud inference (paid tier only)"],"best_for":["Individual creators wanting free local generation","Users wanting optional cloud acceleration without subscription","Teams wanting predictable monthly costs via quota system","Users wanting to try service before committing to paid tier"],"limitations":["Free tier Lab Hours quota not documented; unclear how many generations per month","Paid tier pricing not documented on website","Custom LoRA cloud inference requires paid tier; not available in free tier","Cloud generation quality or speed not documented; unclear if faster than local","No trial period for paid tier documented","Quota rollover or expiration policy not documented"],"requires":["macOS 11+ or iOS 15+ on Apple Silicon device","Draw Things account (free or paid)","Internet connection for cloud features (optional)"],"input_types":["account creation and authentication"],"output_types":["access to local generation (free) and cloud generation (paid quota)"],"categories":["tool-use-integration"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"draw-things__cap_13","uri":"capability://tool.use.integration.app.store.distribution.with.direct.download.fallback","name":"app store distribution with direct download fallback","description":"Distributes the application through Apple App Store for iOS/iPadOS/macOS with direct download option as fallback when App Store is unavailable or inaccessible. App Store distribution enables automatic updates and seamless installation; direct download provides alternative installation path for users in regions with App Store restrictions or experiencing connectivity issues.","intents":["Install application easily through App Store on iOS/iPadOS/macOS","Receive automatic updates when new versions are released","Access application via direct download when App Store is unavailable","Install application in regions with App Store restrictions"],"best_for":["iOS/iPadOS users wanting App Store installation and automatic updates","macOS users wanting native app installation","Users in regions with App Store restrictions or connectivity issues","Users wanting verified, signed application binaries"],"limitations":["App Store distribution subject to Apple's review process and policies","Direct download requires manual installation and update management","No web-based installation or browser access available","macOS version may require code signing and notarization for security","iOS/iPadOS version limited to App Store or direct download; no sideloading documented"],"requires":["macOS 11+ or iOS 15+ on Apple Silicon device","App Store account (for App Store installation) or direct download capability"],"input_types":["installation method selection (App Store or direct download)"],"output_types":["installed application with access to all features"],"categories":["tool-use-integration"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"draw-things__cap_2","uri":"capability://image.visual.controlnet.guided.image.generation","name":"controlnet-guided image generation","description":"Applies ControlNet conditioning to text-to-image generation, allowing users to guide model outputs using structural constraints (edge maps, pose skeletons, depth maps, etc.) provided as input images. ControlNet modules are loaded alongside base models and inject spatial conditioning into the diffusion process, enabling precise control over composition, pose, or layout without full inpainting. Implementation uses cross-attention mechanisms to blend ControlNet embeddings with text prompt embeddings during denoising steps.","intents":["Generate images with specific poses, compositions, or spatial layouts by providing reference images","Maintain consistent character poses or scene layouts across multiple generated variations","Control image generation output more precisely than text prompts alone allow","Adapt existing images or sketches into new styles while preserving structure"],"best_for":["Character artists and animators needing pose-consistent generation","Concept artists prototyping scene layouts and compositions","Creators adapting existing artwork into new styles while preserving structure","Users requiring fine-grained control over spatial aspects of generated images"],"limitations":["ControlNet types and supported conditioning modes not documented (edge, pose, depth, etc. assumed but unconfirmed)","ControlNet model download and storage requirements not specified; likely adds 1-2GB per ControlNet variant","Conditioning strength/weight parameters not documented; default behavior unclear","Inference latency overhead from ControlNet processing not quantified","No UI documentation for ControlNet input preparation or reference image upload"],"requires":["macOS 11+ or iOS 15+ on Apple Silicon device","Base model (Stable Diffusion or FLUX) downloaded and cached locally","ControlNet model(s) downloaded from Draw Things servers (storage and bandwidth required)","Reference image in appropriate format for conditioning type (edge map, pose skeleton, depth map, etc.)"],"input_types":["text prompt (natural language)","reference image (JPEG, PNG; format and resolution requirements unspecified)","conditioning strength parameter (numeric, range unspecified)"],"output_types":["image (PNG or JPEG, resolution varies by device memory and model)"],"categories":["image-visual"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"draw-things__cap_3","uri":"capability://image.visual.inpainting.and.selective.region.image.editing","name":"inpainting and selective region image editing","description":"Enables users to edit specific regions of images by masking areas and regenerating only masked regions using the diffusion model, preserving unmasked content. The infinite canvas feature allows expanding the image boundaries and filling new regions with model-generated content. Inpainting uses masked diffusion, where the model only denoises masked pixels while keeping unmasked pixels fixed, enabling seamless blending of edited and original content.","intents":["Remove or replace unwanted objects in images by masking and regenerating","Extend image boundaries and fill new areas with contextually appropriate content","Modify specific regions of images without affecting surrounding areas","Iteratively refine generated images by editing unsatisfactory regions"],"best_for":["Photo editors and retouchers working with generated or existing images","Concept artists extending compositions beyond original boundaries","Users refining generated images through iterative editing","Creators removing unwanted elements from images locally"],"limitations":["Mask creation and editing UI not documented; unclear if manual drawing, automatic segmentation, or both supported","Inpainting quality and seamlessness not quantified; blending artifacts possible at mask boundaries","Infinite canvas expansion limits not documented; maximum image size constrained by device memory","Inpainting inference latency not specified; likely slower than text-to-image due to additional masking operations","No batch inpainting or multi-region editing in single pass documented"],"requires":["macOS 11+ or iOS 15+ on Apple Silicon device","Base model (Stable Diffusion or FLUX) downloaded and cached locally","Source image (JPEG, PNG; resolution limited by device memory)","Mask definition (manual drawing, selection tool, or automatic segmentation)"],"input_types":["image (JPEG, PNG; resolution varies by device memory)","mask (binary or grayscale image indicating regions to regenerate)","text prompt (optional; guides regeneration of masked regions)","inpainting strength parameter (numeric, range unspecified)"],"output_types":["edited image (PNG or JPEG, same resolution as input)"],"categories":["image-visual"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"draw-things__cap_4","uri":"capability://image.visual.style.transfer.and.image.to.image.transformation","name":"style transfer and image-to-image transformation","description":"Transforms images into different artistic styles (animation, painting, sketch, etc.) by using the source image as a conditioning input to the diffusion model. The process encodes the source image into latent space and applies style guidance through text prompts or style presets, regenerating the image in the target style while preserving composition and content. Implementation uses image encoding followed by conditional diffusion with style-specific prompts or LoRA weights.","intents":["Convert photographs into animated or illustrated styles","Apply consistent artistic styles to multiple images","Transform images into specific visual aesthetics (oil painting, watercolor, sketch, etc.)","Create variations of images in different artistic directions"],"best_for":["Illustrators and concept artists creating stylized versions of reference images","Content creators converting photos into animated or illustrated formats","Users applying consistent visual styles across image collections","Creators exploring artistic variations of source images"],"limitations":["Style presets and available transformations not documented; unclear which styles are built-in vs. LoRA-based","Style transfer quality and fidelity not quantified; potential for loss of fine details or artifacts","Source image resolution and aspect ratio constraints not specified","Inference latency for style transfer not documented; likely similar to text-to-image but with additional encoding step","No fine-grained control over style intensity or blending parameters documented"],"requires":["macOS 11+ or iOS 15+ on Apple Silicon device","Base model (Stable Diffusion or FLUX) downloaded and cached locally","Source image (JPEG, PNG; resolution limited by device memory)","Optional: LoRA weights for specific styles (downloaded from Draw Things servers)"],"input_types":["image (JPEG, PNG; resolution varies by device memory)","style preset or text prompt describing target style","style intensity parameter (numeric, range unspecified)"],"output_types":["transformed image (PNG or JPEG, resolution varies by device memory)"],"categories":["image-visual"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"draw-things__cap_5","uri":"capability://image.visual.image.to.video.animation.generation","name":"image-to-video animation generation","description":"Converts static images into short video clips by generating frame sequences that extend or animate the source image content. The process likely uses frame interpolation or latent space animation to create smooth transitions between generated frames. Implementation details are undocumented, but the feature appears to create basic animations rather than full video generation from scratch.","intents":["Create short animated clips from static images for social media or presentations","Generate looping animations from still artwork or photographs","Extend image content into short video sequences","Create engaging visual content from generated or existing images"],"best_for":["Content creators generating short-form video clips for social media","Animators creating simple animations from static artwork","Users creating looping background animations","Creators extending image content into video format"],"limitations":["Video generation method not documented (frame interpolation, latent animation, or other approach unclear)","Output video codec, resolution, frame rate, and duration not specified","Animation quality and smoothness not quantified; potential for jittery or unrealistic motion","Inference latency for video generation not documented; likely significantly longer than image generation","No control over animation direction, speed, or motion type documented","Video export format and compatibility not specified"],"requires":["macOS 11+ or iOS 15+ on Apple Silicon device with 8GB+ RAM recommended","Base model (Stable Diffusion or FLUX) downloaded and cached locally","Source image (JPEG, PNG; resolution limited by device memory)","Sufficient free storage for video output (codec and resolution dependent)"],"input_types":["image (JPEG, PNG; resolution varies by device memory)","animation parameters (duration, speed, direction; specifics undocumented)"],"output_types":["video file (codec, resolution, frame rate unspecified; likely MP4 or MOV)"],"categories":["image-visual"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"draw-things__cap_6","uri":"capability://data.processing.analysis.model.download.and.local.caching.management","name":"model download and local caching management","description":"Manages the download, storage, and lifecycle of image generation models (Stable Diffusion, FLUX, ControlNets, LoRAs) from Draw Things servers to local device storage. Models are downloaded once and cached locally, enabling offline inference after initial setup. The system tracks model versions, manages storage quotas, and provides UI for model selection and deletion. Implementation uses HTTP downloads with resume capability and local filesystem storage with version tracking.","intents":["Download and cache models locally for offline use without repeated downloads","Manage limited device storage by selectively downloading and removing models","Switch between different model versions or architectures for different generation tasks","Track which models are available locally and their storage consumption"],"best_for":["Users with limited internet bandwidth wanting to download models once","Users with limited device storage needing to manage model versions","Creators using multiple models for different generation tasks","Users in regions with unreliable internet connectivity"],"limitations":["Model sizes not documented; typical Stable Diffusion models 2-7GB, FLUX models likely larger","Download resume capability not documented; interrupted downloads may require restart","Model version management not documented; unclear if multiple versions can coexist or if updates overwrite","Storage quota management not documented; no automatic cleanup or storage warnings mentioned","No support for custom model sources (Hugging Face, Civitai, etc.); limited to Draw Things servers","Model compatibility matrix not documented; unclear which models work with which features (ControlNet, LoRA, etc.)"],"requires":["macOS 11+ or iOS 15+ on Apple Silicon device","Internet connection for initial model download (bandwidth varies by model size)","Free storage space equal to model size plus 20% overhead (e.g., 8GB+ for typical setup)","Draw Things account for model access (free tier available)"],"input_types":["model selection (UI-based selection of available models from Draw Things catalog)"],"output_types":["local model files (proprietary format, stored in app sandbox or Documents folder)"],"categories":["data-processing-analysis","memory-knowledge"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"draw-things__cap_7","uri":"capability://tool.use.integration.optional.cloud.compute.offload.with.quota.based.billing","name":"optional cloud compute offload with quota-based billing","description":"Provides optional cloud inference capability for users who want faster generation or higher quality outputs than local device allows, with quota-based billing using 'Lab Hours' currency. Free tier includes limited monthly Lab Hours; paid tiers (Draw Things+) increase quota. Cloud inference uses Draw Things' managed servers and supports custom LoRA inference, which is not available in free local tier. Implementation uses account-based authentication and cloud API for offloading generation requests.","intents":["Generate images faster using cloud compute when local device is too slow","Use custom LoRAs with cloud inference for higher quality or faster results","Exceed local device memory constraints by offloading to cloud servers","Batch generate multiple images efficiently using cloud compute"],"best_for":["Users with slower Apple Silicon devices (M1) wanting faster generation","Creators needing to use custom LoRAs with cloud inference (free tier limitation)","Users generating high-volume images and wanting to preserve device battery","Teams or organizations needing higher throughput than local devices provide"],"limitations":["Lab Hours quota not documented; unclear how many generations per hour or month are included","Cloud API rate limits not documented; unclear if concurrent requests are supported","Data retention and privacy guarantees for cloud-generated images not documented","Custom LoRA cloud inference requires paid tier; free tier limited to local generation only","Cloud generation latency not documented; unclear if faster than local or just higher quality","No batch API or programmatic access documented; cloud offload appears UI-only","Privacy Pass feature mentioned but mechanism unclear; suggests some cloud operations not fully private by default"],"requires":["Draw Things account (free or paid tier)","Internet connection for cloud inference","Lab Hours quota (free tier includes limited monthly quota; paid tiers increase quota)","For custom LoRA cloud inference: Draw Things+ paid subscription"],"input_types":["text prompt (natural language)","model selection (local or cloud-available models)","optional: custom LoRA selection (paid tier only)","generation parameters (guidance scale, steps, seed, etc.)"],"output_types":["image (PNG or JPEG, resolution varies by cloud model)"],"categories":["tool-use-integration","automation-workflow"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"draw-things__cap_8","uri":"capability://image.visual.try.it.on.virtual.fitting.for.apparel.and.characters","name":"try it on virtual fitting for apparel and characters","description":"Enables users to visualize apparel or character designs on virtual models or in different contexts by applying generated or uploaded designs to reference images. The mechanism is undocumented but likely uses image-to-image transformation or ControlNet-guided generation to adapt designs to different poses, body types, or character models. Allows rapid prototyping of apparel concepts without physical samples.","intents":["Prototype apparel designs on virtual models before physical production","Visualize character designs in different poses or contexts","Test apparel fit and appearance on diverse body types","Iterate rapidly on design concepts without physical samples"],"best_for":["Fashion designers prototyping apparel concepts","Character designers visualizing designs in different contexts","E-commerce businesses previewing products on models","Game developers prototyping character outfits"],"limitations":["Try It On mechanism not documented; unclear if uses image-to-image, ControlNet, or other approach","Available models and body types not documented","Apparel design input format not documented (image, description, etc.)","Fit accuracy and realism not quantified; potential for unrealistic or poorly-fitted results","No control over model pose, lighting, or background documented","Inference latency not documented; likely slower than basic image generation"],"requires":["macOS 11+ or iOS 15+ on Apple Silicon device","Base model (Stable Diffusion or FLUX) downloaded and cached locally","Apparel design or reference image (format unspecified)","Optional: reference model image or selection from available models"],"input_types":["apparel design (image or description; format unspecified)","model selection (reference image or preset model)","optional: pose or context parameters"],"output_types":["image showing apparel on model (PNG or JPEG)"],"categories":["image-visual"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"draw-things__cap_9","uri":"capability://text.generation.language.prompt.engineering.and.generation.parameter.control","name":"prompt engineering and generation parameter control","description":"Provides UI controls for fine-tuning text-to-image generation through prompt input and numerical parameters (guidance scale, sampling steps, seed, etc.). Guidance scale controls how strongly the model follows the text prompt; sampling steps control diffusion quality and speed; seed enables reproducible generation. Implementation uses standard diffusion parameter exposure through UI sliders and text input fields.","intents":["Refine generated images by adjusting guidance scale and sampling steps","Reproduce specific generated images using seed values","Experiment with different prompts and parameters iteratively","Control trade-offs between generation speed and quality"],"best_for":["Users learning to write effective prompts for image generation","Creators fine-tuning generation parameters for specific aesthetics","Users reproducing specific generated images for iteration","Developers and researchers experimenting with diffusion parameters"],"limitations":["Parameter ranges and defaults not documented (guidance scale typical range 7-15, steps typical range 20-50)","Prompt syntax and special tokens not documented; unclear if supports Stable Diffusion prompt weighting or other syntax","Seed reproducibility not guaranteed across model versions or app updates","No prompt history or saved presets documented","No advanced prompt engineering tools (prompt weighting, syntax highlighting, etc.) documented"],"requires":["macOS 11+ or iOS 15+ on Apple Silicon device","Base model (Stable Diffusion or FLUX) downloaded and cached locally"],"input_types":["text prompt (natural language, optional special syntax)","guidance scale (numeric, typical range 1-20)","sampling steps (numeric, typical range 1-100)","seed (numeric, for reproducibility)","optional: negative prompt (text describing what to avoid)"],"output_types":["image (PNG or JPEG, resolution varies by device memory)"],"categories":["text-generation-language","image-visual"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"draw-things__headline","uri":"capability://image.visual.local.ai.image.generation.app.for.apple.devices","name":"local ai image generation app for apple devices","description":"Draw Things is a native macOS and iOS application that enables users to run image generation models like Stable Diffusion locally on Apple Silicon, offering optimized performance and privacy without cloud dependencies.","intents":["best local image generation app","AI image generator for macOS","offline image generation tool for iOS","image generation software for Apple Silicon","best app for Stable Diffusion on Mac"],"best_for":["creative professionals","hobbyists"],"limitations":["only available on Apple devices"],"requires":["Apple Silicon"],"input_types":["text prompts"],"output_types":["generated images"],"categories":["image-visual"],"confidence":0.5,"matches":0,"success_rate":0}],"trust":{"score":56,"verified":false,"data_access_risk":"high","permissions":["macOS 11+ or iOS 15+ on Apple Silicon device (M1 or newer)","Minimum 4GB free storage for base model, 8GB+ recommended for multiple models","Internet connection for initial model download only; offline generation supported thereafter","macOS 11+ or iOS 15+ on Apple Silicon device with 8GB+ RAM recommended","Training dataset of 10-100+ images (exact requirements not specified)","Sufficient free storage for base model plus LoRA weights (typically <100MB per LoRA)","30+ minutes to several hours of uninterrupted device usage for training","macOS 11+ or iOS 15+ on Apple Silicon device","Sufficient storage for multiple models (2-7GB per model typical)","Sufficient RAM for model loading (8GB+ recommended for multiple models)"],"failure_modes":["Apple Silicon only — no Windows, Linux, or Intel Mac support documented","Generation speed varies significantly by device model and available VRAM; no published benchmarks provided","Models must be downloaded and stored locally before use; typical model sizes 2-7GB per model","Maximum concurrent generations limited by device memory; no batch processing API documented","Image resolution and quality constrained by available device memory; maximum resolution not publicly specified","LoRA training time and resource requirements not publicly documented; likely hours to days depending on dataset size and device","Training dataset preparation and curation responsibility falls entirely on user; no built-in dataset management tools documented","LoRA inference in free tier limited to local generation only; cloud inference with custom LoRAs requires paid tier","Compatibility matrix between LoRA versions, base models, and app versions not documented","No version control or experiment tracking for trained LoRAs; overwriting previous versions not prevented","builder identity is not verified yet","no observed match outcomes yet"],"rank_breakdown":{"adoption":0.7,"quality":0.9,"ecosystem":0.39999999999999997,"match_graph":0.25,"freshness":0.75,"weights":{"adoption":0.25,"quality":0.25,"ecosystem":0.1,"match_graph":0.35,"freshness":0.05}},"observed_outcomes":{"matches":0,"success_rate":0,"avg_confidence":0,"top_intents":[],"last_matched_at":null},"maintenance":{"status":"active","updated_at":"2026-05-24T12:16:21.548Z","last_scraped_at":null,"last_commit":null},"community":{"stars":null,"forks":null,"weekly_downloads":null,"model_downloads":null,"model_likes":null}},"distribution":{"claim_url":"https://unfragile.ai/submit?claim=draw-things","compare_url":"https://unfragile.ai/compare?artifact=draw-things"}},"signature":"+lMN5gIc73hiq4Wd7qVWcEo0GbupuO6GnRmNsWxgM6Pb5dLWQz4JRZ7cTL3Z9f1oHCcQTsujvN8GSh/KsXHCDA==","signedAt":"2026-06-23T04:07:23.069Z","signedBy":"unfragile.ai","version":1},"_links":{"self":"https://unfragile.ai/api/v1/passport/draw-things","artifact":"https://unfragile.ai/draw-things","verify":"https://unfragile.ai/api/v1/verify?slug=draw-things","publicKey":"https://unfragile.ai/api/v1/trust-passport-public-key","spec":"https://unfragile.ai/trust","schema":"https://unfragile.ai/schema.json","docs":"https://unfragile.ai/docs"}}