SD
Stable Diffusion (SD) is an open-source latent diffusion model designed to generate high-resolution images from textual prompts, enabling the synthesis of diverse visual content including photorealistic scenes, artistic styles, and conceptual illustrations. Developed initially by the CompVis group at Ludwig Maximilian University of Munich in collaboration with Stability AI and Runway, it compresses image data into a lower-dimensional latent space for efficient training and inference, requiring only modest GPU resources such as 6-10 GB VRAM for operation.[1][2] The model's architecture builds on prior diffusion techniques by applying noise addition and denoising processes within the latent domain via a variational autoencoder and U-Net, conditioned on CLIP text embeddings to align outputs with descriptive inputs; it was trained on subsets of the LAION-5B dataset comprising billions of image-text pairs scraped from the web.[1] This approach yields superior prompt adherence and compositional fidelity compared to earlier proprietary systems, with capabilities extending to image-to-image translation, inpainting, and upscaling through community extensions.[2] Released under the permissive Creative ML OpenRAIL-M license in August 2022, Stable Diffusion democratized advanced generative AI by allowing local deployment and fine-tuning, fostering thousands of specialized variants for niche domains like character design or medical imaging, while its efficiency has accelerated research into scalable multimodal AI.[2] Nonetheless, its unfettered accessibility has ignited disputes over training data provenance—drawing from public internet sources without explicit artist consent—and the potential for misuse in creating deceptive or explicit material, prompting legal actions from stock photo agencies and calls for embedded safeguards, though empirical evidence indicates that output biases largely mirror distributional imbalances in training corpora rather than inherent model flaws.[2][1]Places
United States
The two-letter postal code for South Dakota, a landlocked U.S. state in the northern Great Plains, is SD.[3] The state spans approximately 77,116 square miles and features diverse terrain including prairies east of the Missouri River and more rugged landscapes to the west, with elevations rising to 7,242 feet at Black Elk Peak.[4] Mount Rushmore National Memorial, located in the Black Hills region, depicts the faces of Presidents George Washington, Thomas Jefferson, Theodore Roosevelt, and Abraham Lincoln; construction began after a dedication ceremony on August 10, 1927, presided over by President Calvin Coolidge.[5][6] San Diego, the second-largest city in California by population with over 1.3 million residents as of 2020, is frequently abbreviated as SD in regional and informal contexts..html) Situated on the Pacific coast adjacent to the U.S.-Mexico border, it encompasses key cross-border infrastructure like the San Ysidro Port of Entry, handling over 90,000 daily northbound crossings.[7] The city hosts Naval Base San Diego, the principal U.S. Navy shore installation on the West Coast, supporting more than 50 ships and 20,000 personnel.[8] San Diego County, officially abbreviated as SD in certain California state administrative codes, administers a 4,526-square-mile area including the city and surrounding unincorporated territories.[9]Other locations
Sudan, officially the Republic of the Sudan, is designated by the ISO 3166-1 alpha-2 code SD in international standards for country identification.[10] This code is used in contexts such as internet domain names (.sd), vehicle registration, and data exchange protocols.[11] The country, located in northeastern Africa, borders Egypt to the north, Libya to the northwest, Chad to the west, the Central African Republic to the southwest, South Sudan to the south, Ethiopia to the southeast, Eritrea to the east, and the Red Sea to the northeast, covering an area of 1,886,068 square kilometers.[10] In some administrative or historical mappings, SD has denoted subnational entities, such as the Shandong Province (山东; Shāndōng) in the People's Republic of China, where it serves as an informal or contextual abbreviation in English-language geographic lists and transport codes, though official Chinese standards use numeric or Pinyin-based identifiers. Shandong, situated on the Yellow Sea coast, encompasses 157,100 square kilometers and had a population of approximately 101.7 million as of the 2020 census. This usage contrasts with more standardized codes like ISO 3166-2:CN for Chinese subdivisions, highlighting SD's role in ad hoc international referencing rather than formal nomenclature.Organizations and politics
Political parties
The Sweden Democrats (Sverigedemokraterna; SD) is a nationalist political party in Sweden, founded on February 6, 1988, by members including former neo-Nazis and skinhead activists who sought to establish a new right-wing alternative amid the fragmentation of earlier nationalist groups.[12] The party has evolved to emphasize immigration restriction, law-and-order policies, and Swedish cultural identity, positioning itself as right-wing populist while distancing from extremism through internal purges and professionalization starting in the early 2010s.[13] In the September 11, 2022, general election, SD received 20.54% of the national vote, securing 73 seats in the 349-member Riksdag and emerging as the second-largest party by vote share, a quadrupling from its 5.7% in 2010 that marked its parliamentary breakthrough.[12] [13] This performance reflected growing voter prioritization of immigration and crime issues, enabling SD to provide external support for a centre-right minority government led by the Moderate Party's Ulf Kristersson from October 2022 onward, despite longstanding establishment reluctance evidenced by prior cordon sanitaire tactics.[14] In Slovakia, Direction – Social Democracy (Smer – sociálna demokracia; Smer-SD) operates as a major political force, established on June 15, 1999, by Robert Fico, blending social democratic rhetoric with left-wing nationalism and populism to appeal to working-class voters disillusioned with neoliberal reforms. Smer-SD has achieved repeated electoral success, forming governments in 2006, 2012, and 2023, with the latter victory on September 30, 2023, yielding 22.94% of the vote and 42 seats in the National Council, allowing a coalition with nationalist parties amid public backlash against EU-driven policies. Unlike traditional social democratic parties, Smer-SD's governance has prioritized state interventionism and Euroscepticism, correlating with higher voter turnout in regions affected by economic migration and corruption scandals.[15] Other entities abbreviated SD include the short-lived Democratic Left (Sinistra Democratica; SD) in Italy, active from 2007 to 2010 as a left-leaning splinter focused on democratic socialism but merging into broader coalitions without significant independent electoral impact. No major historical Sudanese political parties consistently use the SD abbreviation post-independence in 1956, though southern factions like the Sudan African National Union participated in unity efforts without such branding. These cases illustrate SD's varied ideological applications, from right-wing populism in Sweden to hybrid left-nationalism in Slovakia, with empirical gains tied to addressing voter concerns over globalization and identity rather than media-favored narratives.Other organizations
The SD Association is a nonprofit trade organization that develops and promotes standards for Secure Digital (SD) memory cards and related storage technologies.[16] Founded in January 2000 by Panasonic Corporation, SanDisk, and Toshiba, it operates as a global industry consortium focused on ensuring interoperability across devices such as cameras, mobile phones, and automobiles.[17][16] The association maintains approximately 800 member companies, which collaborate through committees and working groups to advance SD specifications, including extensions for higher capacities and speeds.[16] Membership is open to entities involved in SD technology, with annual dues of $4,500 for executive-level participation and $2,500 for general members, headquartered in San Ramon, California.[17] Its activities emphasize technical standardization rather than manufacturing or commercial sales, supporting broader adoption in consumer electronics and embedded systems.[16]Military and defense
Weaponry and equipment
The MP5SD submachine gun, developed by Heckler & Koch, incorporates an integral suppressor denoted by the "SD" suffix, standing for Schalldämpfer (German for sound suppressor), reducing muzzle noise to approximately 70 decibels while maintaining ballistic performance comparable to unsuppressed variants.[18] First produced in the late 1970s, it fires 9×19mm Parabellum ammunition at a cyclic rate of 800 rounds per minute and has been adopted by special forces units for covert operations due to its low signature. Self-destruct mechanisms, abbreviated as "-SD" in projectile designations, are incorporated into various military munitions to prevent unexploded ordnance from remaining hazardous post-mission, activating via fuze timers or altimeters after a set range or altitude.[19] For instance, the 20mm M940 MPT-SD round, used in Vulcan air defense systems, features a multi-purpose tracer with self-destruct at around 2,500 meters to minimize ground impact risks.[20] Similarly, Nammo's MP-T/SD 12.7mm ammunition combines tracer, incendiary, and fragmentation effects with self-destruct for enhanced terminal ballistics in machine gun applications.[21] During World War II, the German SD 2 (Sprengbombe Dickkopf 2), nicknamed "Schmetterling" (butterfly) for its winged anti-personnel bomblet design, was deployed in cluster munitions from 1939 onward, scattering up to 23 submunitions per bomb to target infantry with delayed or impact fuzes.[22] Over 12 million were produced, seeing use in the invasions of Poland and the Soviet Union, though their small size complicated clearance efforts postwar.[22] The SD-ROW (Self Defence Remotely Operated Weapon) is a lightweight, remotely controlled turret system mounting a 5.56mm or 7.62mm light machine gun, developed for vehicle or static platform defense, allowing operators to engage threats from cover with stabilized optics and 360-degree traverse.[23] Introduced in the early 2000s, it emphasizes force protection in asymmetric warfare scenarios.[23]Units and operations
The Sicherheitsdienst (SD), or Security Service of the Reichsführer-SS, operated as the primary intelligence and security apparatus of the Nazi SS from its formation in 1931 under Reinhard Heydrich.[24][25] Initially comprising a small cadre of about 300 full-time agents focused on countering internal threats to the Nazi Party through surveillance and informant networks, the SD expanded rapidly after 1933, incorporating part-time operatives and achieving a membership of approximately 3,000 by 1939. Its operational mandate included domestic intelligence (SD-Inland) for monitoring political opponents within Germany and foreign intelligence (SD-Ausland) for espionage abroad, with field units structured into regional offices (SD-Leitabschnitte) and mobile detachments for targeted actions.[24] During World War II, SD units integrated into military campaigns as auxiliary security forces, often embedded with advancing armies to neutralize resistance. In the September 1939 invasion of Poland, SD personnel numbering around 2,700 across six Einsatzgruppen formations conducted "special tasks" (Sonderaufträge), including the arrest and execution of Polish elites, intellectuals, and Jews; Einsatzgruppe IV, for example, reported eliminating over 50,000 individuals in its zone by late 1939 through shootings and roundups.[26][27] These operations extended into the Soviet Union from June 1941, where SD-led subunits within Einsatzgruppen A–D executed systematic killings, documenting over 1 million deaths by 1943 via gas vans, pits, and firing squads in occupied eastern territories.[28] SD efficacy relied on pre-war intelligence files for rapid targeting, though logistical constraints and partisan activity occasionally limited field effectiveness, as evidenced by incomplete reports from SD outposts in Ukraine.[26] In the Free City of Danzig, the Selbstschutz Danzig (SD) functioned as an ethnic German paramilitary militia unit in the pre-war period, formalized in 1937–1939 with roughly 1,500–2,000 volunteers trained in light infantry tactics and sabotage. Activated during the September 1, 1939, German seizure of Danzig, SD forces—coordinated with SS-Heimwehr elements—secured ports, post offices, and police stations, detaining over 100 Polish administrators and suppressing opposition in operations that facilitated the city's bloodless annexation without direct Wehrmacht engagement.[29] Post-annexation, these units transitioned to auxiliary roles in Pomerania, participating in early occupation security sweeps that resulted in the deaths of several thousand Poles by December 1939, though exact SD-attributable casualties remain disputed due to overlapping SS actions.[30]Science, mathematics, and statistics
Mathematics and statistics
Standard deviation, denoted as \sigma for a population or s for a sample, quantifies the dispersion of a dataset around its mean value.[31] It is calculated as the square root of the variance, providing a measure in the same units as the data, which facilitates intuitive interpretation of variability.[32] For a population of N values, the formula is \sigma = \sqrt{ \frac{ \sum_{i=1}^{N} (x_i - \mu)^2 }{N} }, where \mu is the population mean; for a sample of n values, it uses s = \sqrt{ \frac{ \sum_{i=1}^{n} (x_i - \bar{x})^2 }{n-1} } to provide an unbiased estimate, dividing by n-1 to account for degrees of freedom.[31] [33] The term "standard deviation" was introduced by Karl Pearson in a 1893 lecture and first appeared in print in 1894, replacing earlier terms like "root mean square deviation" for a standardized measure of spread.[34] Unlike variance, which averages the squared deviations from the mean and thus expresses dispersion in squared units, standard deviation scales this back via the square root, yielding a value directly comparable to the mean and better suited for assessing relative variability in empirical distributions.[35] [36] In hypothesis testing, standard deviation enables chi-squared tests to evaluate claims about population variability, assuming normality; for instance, the test statistic \chi^2 = \frac{(n-1)s^2}{\sigma_0^2} compares sample variance s^2 to a hypothesized \sigma_0^2, with rejection of the null if it falls outside critical values from the chi-squared distribution.[37] When population standard deviation is known, z-tests for means incorporate it via the standard error \sigma / \sqrt{n}, standardizing sample means for significance assessment.[38] In quality control, standard deviation measures process variation to set control limits and detect deviations from targets; for example, in Six Sigma, it defines sigma levels where fewer defects occur within \pm 3\sigma or \pm 6\sigma bounds, guiding capability indices like C_p = \frac{USL - LSL}{6\sigma} to quantify how well specifications encompass natural variation.[39] Lower standard deviation indicates tighter process control, enabling predictions of defect rates under normality assumptions.[40]Astronomy
In stellar classification systems, subdwarfs—stars with lower metallicity than typical main-sequence counterparts of the same spectral type—are denoted by the prefix "sd", as in sdO (helium-rich, hot subdwarfs showing strong He II absorption) and sdB (hydrogen-rich, with dominant Balmer lines alongside helium features).[41] These classifications arise from spectroscopic analysis revealing underabundances in metals (e.g., [Fe/H] < -0.5) and peculiar evolutionary states, often as post-red-giant-branch objects with masses around 0.4–0.5 solar masses.[42] Hot subdwarfs, comprising sdO and sdB types, constitute about 1–2% of B-type stars in Galactic halo populations, with effective temperatures exceeding 20,000 K; their identification relies on surveys detecting ultraviolet excess and high proper motions.[43] The Southern Durchmusterung (SD) serves as a historical astronomical catalog designating over 120,000 stars in the southern celestial hemisphere, covering declinations from -1° to -23° using equatorial coordinates (e.g., SD +hh° mmmm for right ascension and declination zones).[44] Compiled by Friedrich Wilhelm Argelander and assistants between 1879 and 1884 from observations at the Bonn Observatory, it extends the Bonner Durchmusterung northward and employs visual magnitude estimates accurate to about 0.5 magnitudes for stars brighter than 9th magnitude.[45] SD designations persist in modern cross-references for proper motion studies and as primary identifiers for faint southern stars lacking brighter catalog entries, with digital versions now integrated into databases like the Strasbourg Data Center for precise astrometry.[44] The Sloan Digital Sky Survey (SDSS), while formally abbreviated SDSS, contributes extensively to "SD"-prefixed object designations through its spectroscopic and photometric mapping of over one-third of the extragalactic sky.[46] Launched in 2000 with a dedicated 2.5-meter telescope at Apache Point Observatory, New Mexico, SDSS has released data on more than 5 million spectra by 2020, enabling discovery of thousands of subdwarfs via uniform ugriz photometry and fiber-fed spectroscopy resolving radial velocities to 1–2 km/s precision.[46] Key findings include sdB stars in globular clusters and field populations, with SDSS Data Release 16 (2020) cataloging 465,000 unique stellar spectra, including low-metallicity halo objects down to g=22.5 magnitude.[47] This survey's legacy datasets facilitate variability studies and binary identifications among subdwarfs, revealing close companions in ~50% of sdB systems via eclipse timing.[42]Medicine
In medicine, "SD" most frequently denotes standard deviation, a biostatistical measure quantifying the dispersion of data points around the mean in clinical datasets, such as patient response variability in randomized controlled trials.[48] This metric is essential for assessing the reliability of efficacy endpoints, like mean changes in biomarkers or symptom scores, where lower SD values indicate tighter clustering and stronger evidence of treatment effects amid biological variability.[49] For instance, trial reports often present results as "mean ± SD" to enable power calculations and confidence intervals, guiding sample size determinations to detect clinically meaningful differences with statistical rigor.[48] Shoulder disarticulation (SD) refers to a high-level upper limb amputation performed through the glenohumeral joint, removing the humerus head while preserving the scapula and glenoid fossa for potential prosthetic fitting.[50] This procedure, indicated for severe trauma, tumors, or infections untreatable by lesser resections, classifies as a proximal amputation level in orthopedic prosthetics, where socket designs must accommodate scapular movement for functional myoelectric control.[51] Postoperative rehabilitation emphasizes residual limb shaping and phantom pain management, with outcomes depending on preserved pectoralis and latissimus dorsi innervation.[52] Less commonly, "SD" abbreviates streptodornase, an enzyme derived from streptococci used historically in mixtures with streptokinase for debridement of necrotic tissue in wound care, though its clinical application has declined with modern antimicrobials.[53] In oncology staging under RECIST criteria, SD signifies stable disease, denoting neither progression nor sufficient response in tumor measurements during therapy evaluation.[54] No major syndromes are uniquely codified as SD in ICD-11, though archaic references link it to Sydenham's chorea (a post-streptococcal neurological disorder with involuntary movements, ICD-11 code 8A08.1).Other scientific applications
Sustainable development (SD), a core concept in environmental science, originated from the 1987 World Commission on Environment and Development's Brundtland Report, which defined it as "development that meets the needs of the present without compromising the ability of future generations to meet their own needs." This framework integrates empirical assessments of ecological limits, resource flows, and human impacts, influencing models for planetary boundaries, such as those quantifying safe operating spaces for humanity based on nine biophysical processes like biodiversity loss and nitrogen cycles. Applications include case studies like the 1992 Earth Summit's Agenda 21, which applied SD to local land-use planning through data-driven evaluations of soil degradation rates exceeding 24 billion tons annually in vulnerable regions. In practice, SD has guided interdisciplinary experiments in resilience modeling, such as simulations of coupled human-natural systems where variables like fishery yields are tested against overexploitation thresholds, revealing collapse risks when harvest rates exceed 20-30% of maximum sustainable yield, as demonstrated in Baltic Sea cod stock analyses from 1970-2000. However, empirical reviews indicate limitations, with global indicators under UN Sustainable Development Goals (adopted 2015) showing persistent deterioration in areas like freshwater use, where per capita availability declined 20% from 1990-2020 despite policy efforts, attributed to mismatched incentives favoring short-term extraction over causal feedback loops in hydrological cycles.[55] These outcomes highlight challenges in translating SD's first-principles emphasis on intergenerational equity into verifiable, localized metrics amid institutional biases toward aggregated, top-down targets.Technology and computing
Electronics and hardware
The Secure Digital (SD) card format represents a proprietary standard for non-volatile flash memory cards used in portable electronics, digital cameras, and mobile devices. Jointly announced by SanDisk, Panasonic, and Toshiba in August 1999, the format was standardized through the SD Association, established in January 2000 by these founding members to ensure interoperability and promote adoption across consumer and embedded hardware.[56] [16] The full-size SD card measures 32 mm × 24 mm × 2.1 mm with 9 pins (two ground pins) for high-speed and UHS-I interfaces, supporting backward compatibility with earlier MultiMediaCard standards while enabling secure data storage via optional encryption.[57] Smaller variants include miniSD cards at 21.5 mm × 20 mm × 1.4 mm, introduced in 2003 for compact devices like early mobile phones, and microSD cards at 15 mm × 11 mm × 1.0 mm with 8 pins (one ground pin), launched in 2005 for ultra-portable applications such as smartphones and action cameras.[58] [57] Capacity specifications delineate four categories: original SD cards up to 2 GB using FAT12/16 file systems; SDHC from 2 GB to 32 GB with FAT32; SDXC from 32 GB to 2 TB employing exFAT for extended addressing; and SDUC from 2 TB to 128 TB, ratified in 2018 to accommodate future high-density NAND flash in enterprise and archival hardware.[59] [60] These hardware limits stem from 128-bit block addressing in the physical layer protocol, with SDUC extending it via 274 GB blocks to support theoretical densities beyond petabyte-scale aggregation in multi-card arrays.[57] In video hardware, "SD" refers to Standard Definition signals, standardized under ITU-R Recommendation BT.601 for digital studio encoding in 525-line (NTSC) and 625-line (PAL/SECAM) systems, using 13.5 MHz sampling for luminance (Y) and 6.75 MHz for chrominance (Cb/Cr) components at 4:2:2 or 4:4:4 ratios.[61] This specification governs hardware interfaces in legacy broadcast equipment, DVD players, and analog-to-digital converters, distinguishing it from high-definition formats by lower resolution (typically 720×480 or 720×576 active pixels) and bandwidth requirements under 270 Mb/s serial digital interface (SDI).[61]Software and computing
Stable Diffusion is an open-source latent diffusion model for generating high-resolution images from textual prompts, developed by Stability AI in collaboration with researchers from LMU Munich and Runway ML.[2] Released publicly on August 22, 2022, it marked a significant advancement in accessible AI-driven image synthesis, enabling users to run the model locally on consumer hardware without reliance on cloud services or proprietary restrictions.[62] The initial version, Stable Diffusion 1.4, and subsequent iterations like 1.5 were trained on a filtered subset of the LAION-5B dataset, which contains approximately 5.85 billion image-text pairs derived from web-scraped data using CLIP embeddings for relevance.[63] This training approach, while drawing from vast uncurated internet sources, has empowered decentralized fine-tuning and extensions by the open-source community, contrasting with closed models that incorporate built-in content filters to limit outputs deemed sensitive.[64] The model's permissive licensing under the CreativeML Open RAIL-M license facilitates broad adoption for applications in art, design, and research, with community-hosted variants on platforms like Hugging Face achieving widespread use for tasks such as inpainting, outpainting, and style transfer.[65] By October 2024, updates like Stable Diffusion 3.5 improved prompt adherence and image quality, maintaining the emphasis on open weights to support innovation outside corporate oversight.[64] This has spurred algorithmic advancements, including efficient inference optimizations like distilled variants that reduce computational demands while preserving generative capabilities. In Unix-like operating systems, particularly HP-UX, SD denotes the Software Distributor (SD-UX), a suite of commands and tools for packaging, installing, updating, and removing software depots on HP 9000 Series systems.[66] Introduced with HP-UX 10.x and refined in 11i versions, SD-UX supports depot-based distribution via commands likeswinstall for graphical or command-line management, ensuring dependency resolution and media compatibility for enterprise software deployment.[67]
SD also serves as an abbreviation for Software Development in computing methodologies, referring to structured processes or roles in the software lifecycle, such as in SA/SD (Structured Analysis and Structured Design) frameworks used for requirements modeling and modular implementation in legacy system design.[68] This usage appears in documentation for phased development, distinct from modern agile practices but foundational in enterprise computing standards.