Living reference
A reference to the tools, companies, laws, and standards behind our coverage. Check the facts, go to the source, and follow what changes.
23 records
Firefly brings image, video, audio and design generation into Adobe's creative workflow. The app now offers both Adobe models and partner models, so choosing Firefly does not identify which model produced an asset. For a studio developing campaign concepts or a brand adapting creative, that distinction belongs in the handoff: record the model, retain the editable work and review the output before delivery.
Adobe's AI Assisted Editor is a beta editing mode inside Photoshop on desktop. Adobe offers it to current Photoshop subscribers who have installed the required update. You can describe changes in a prompt bar and switch back to Pro Editor for Photoshop's full toolset. Adobe distinguishes this mode from the separate AI Assistant. The editor includes Prompt to Edit, Generative Fill and Expand, plus visual markup for communicating edits. Adobe says switching into the mode creates a flattened working copy while preserving the original Pro Editor file and layers. On return, Photoshop places the changes on a new generative layer. Selections or masks can specify the area to edit; visual markup can express position, shape and colour. Features consume different amounts of generative credits, and regional restrictions apply. For a retoucher or art director, the handoff needs a clear brief: preserve the layered source, inspect the returned result, and identify which details need further editing. Keeping the original layers does not establish that each generated change returns as a separate editable component.
Adobe Substance 3D Painter is a painting and texturing application for 3D meshes. Artists work with layers, fills, materials and masks, then export texture maps for use in other software. For fashion production, it can help develop surfaces on an existing footwear, accessory or garment model: a printed graphic, a worn edge or a different finish. These are illustrative uses, not tested product results. The handoff needs more than a convincing viewport. Output templates control file names, channel packing, format and bit depth. Agree those settings with the receiving artist and review the exported maps in the intended renderer before approving product imagery.
Blender is a free, open-source 3D creation suite covering modelling, sculpting, animation, simulation and rendering. It includes compositing and video-editing tools, so an artist can assemble and finish substantial parts of a visual project in one application. For fashion work, it can provide a space for product visualisation, set building and campaign animation around supplied or newly modelled assets. Those are production uses, not evidence that a garment pattern or simulation transfers intact from another tool. A studio still needs a delivery specification: the model and textures to retain, the renderer to use and the files the next collaborator can open.
Browzwear develops software for apparel teams to design, review, and share digital garments. Its products include VStitcher for 3D garment creation and Stylezone for collaboration around those assets. The company lists its headquarters in Singapore and offices across several regions. For a fashion team, the useful distinction is between creating a digital garment and agreeing how it will be used. Pattern, fabric, fit, and approval responsibilities still need to be assigned. Browzwear’s tools connect those stages; claims about speed or sample reduction should be evaluated against the team’s own production process.
Stylezone is the collaboration layer around Browzwear's garment work. Teams can organize collections, review styles, compare versions and keep comments attached to the design. It is useful to merchandisers, designers and production partners who need to discuss the same garment without each working inside the design application. Keep the distinction clear between a presentation variant, a reviewed style and the file approved for production. August 2026 updates add a Chinese interface option and organization-level brand guidance for AI generation. Designers can adjust its strength, and explicit user prompts take priority over the guidance.
VStitcher connects two-dimensional pattern work with a three-dimensional garment, giving designers and technical teams a shared way to examine construction, fit and material choices. It includes avatar fitting, tension and pressure views, and pattern and technical-document exports. For fashion production, the useful question is how closely the digital setup represents the intended garment—not simply how convincing the render looks.
Publishers and creative tools use C2PA to record an asset's origin and edits in a signed Content Credential. A reviewer can inspect the signer and check whether the associated data passes validation. The specification defines how applications bind those assertions to media. A publisher still needs to verify the depicted events and establish permission to use the asset; a valid provenance record cannot settle either question on its own.
The CFTC request for comment on derivatives-market uses of AI and computing. This Index record is not a rule.
Designers use CLO to edit 2D garment patterns and simulate their drape on a 3D avatar. Print Layout supports print placement and fabric-consumption calculations. In AI Studio, designers can develop textures or generate fashion images, including try-on and flat-lay views. Those image workflows serve a different production decision from checking a pattern: a convincing generated garment image does not establish the fit or construction of a physical sample.
ComfyUI makes a generation workflow visible as a graph of connected nodes. A technical artist can retain the steps used for an image or other media task, reuse them and change individual stages. That gives a studio more control over the process than a prompt alone records. A shared graph still needs its matching models, extensions and settings; the workflow file by itself is not the whole production environment.
The EU AI Act regulates AI systems and general-purpose AI models according to their use and risk. For creative businesses, its relevance includes transparency around synthetic content, the responsibilities of model providers and the way AI is used in areas such as employment. A campaign team using a generator is not automatically subject to the same duties as the company providing that system. Establish the actor, use and applicable date before drawing a compliance conclusion.
The EU Batteries Regulation gives the battery passport a defined start date. From 18 February 2027, it covers electric-vehicle batteries, light-means-of-transport batteries and industrial batteries above 2 kWh placed on the EU market or put into service. For a production company evaluating battery-powered equipment, classification comes before procurement language: a supplier's passport claim needs checking against the relevant battery category. The Commission's August 2026 guidance distinguishes required data from conditional or later requirements; it adds no legal duties. The useful handoff connects the purchased battery to its responsible operator and supporting evidence. The law also assigns responsibility for keeping that information accurate.
The EU's Ecodesign for Sustainable Products Regulation establishes a framework for product requirements, including digital product passports. Product-specific delegated acts determine the passport's contents and whether it identifies a model, batch or individual item. For fashion teams, the immediate planning question is who can supply reliable product information and keep it connected to the goods being sold. The Commission describes textile requirements as a future delegated act, with adoption planned for the fourth quarter of 2027. Registry arrangements already have a separate implementing regulation. A brand can organise supplier evidence and ecommerce responsibilities now; the eventual textile act will determine the required information and application date.
Google develops Search, Gemini and creative tools including Flow. Its services touch both video production and the ways publishers reach readers. For creative businesses, those roles create separate decisions about making work and allowing Google to use published material. Publishers can use the Google-Extended robots.txt token to control specified future Gemini training and grounding uses. Google says this choice does not affect inclusion or ranking in ordinary Search. A separate Search Console setting governs appearance and grounding in generative Search features. Google reports worldwide rollout by August 31, 2026, and says sites that opt out lose traffic and impressions from those features. A publisher should assess those choices against the uses it wants to permit and the audience it expects to reach. A company-wide label such as “AI opt-out” obscures the differences between products; each setting needs its own scope check.
Google Flow is Google's video-generation and editing tool. You can generate clips from prompts or supplied images and use a supported model to edit uploaded or generated video. With Gemini Omni Flash 1.1, you can set opening and closing frames on the web, choose clips of four, six, eight or ten seconds, and edit videos up to ten seconds. Google lists extension for that model as forthcoming. Google's August 27 update announced 1080p and 4K export alongside 360p drafting. Its help documentation specifies that Omni 360p generation costs half the credits of 720p; Pro and Ultra subscribers can upscale those drafts to 720p for zero credits. Google documents Omni access for all Flow users, with some features and voice customizations restricted by region. Veo models have different duration, ingredients and editing support; select the active model before planning a shot. For a creative team, these controls separate shot planning from final delivery. Opening and closing images establish endpoints; the team still needs to review movement and product detail between them. An export resolution alone establishes no level of visual fidelity.
GS1 Digital Link gives product identifiers a standard form inside web addresses. For a fashion brand, a QR code can connect a garment's GTIN, with optional batch or serial detail, to online information. A separate resolver specification supports several destinations from the same identity, such as care instructions or a product record; redirects can change without reprinting the code. Brands still need to decide what the identifier represents, who maintains the linked information and how to check its accuracy. Digital Link supplies identification and linking conventions. It neither defines a complete digital product passport nor certifies the sustainability claims attached to one.
IPTC provides structured fields that travel with a photograph: information about the image and the people or processes behind it. Version 2025.1 adds fields for the AI system, its version, prompt information and the prompt writer. These fields help a picture desk or creative team describe an asset consistently. They are declarations, not a substitute for checking provenance or permission to publish.
glTF 2.0 is Khronos's open format for delivering 3D models and scenes to applications and viewers. Its material model and compact asset structure suit product visualisation, including footwear and digital garments prepared for ecommerce. Files can use JSON with separate resources or the binary GLB container. For a brand, glTF belongs near the customer-facing end of a 3D workflow: it leaves out authoring information to reduce runtime work, so a delivery asset should not be the only design archive. Material extensions can describe further surface properties. Review the asset in the intended viewer with its actual lighting before approving the appearance of a fabric or finish.
Luma's Dream Machine documentation covers image and video creation. Its Ray3 Modify guide describes changing an input shot with character references and keyframes, with a strength control for the degree of transformation. That workflow gives a director specific choices to review across the shot. Luma also publishes a newer Luma Agents app guide and separate Ray3.2 API documentation; their controls and account terms need their own scope checks.
OpenUSD, or Universal Scene Description, is an open-source system for describing, composing and exchanging 3D scenes. Artists can combine geometry, materials and lighting while retaining layers and variants for further work. In a digital-fashion workflow, a colourway variant and a campaign-lighting revision can remain distinct contributions to the scene. That is a possible use of USD's composition model, rather than evidence that a particular garment application supports the whole workflow. Export support alone does not establish which shaders or animation will survive a handoff. The production team needs an agreed supported subset and a review of the scene in the receiving application.
Runway's Gen-4 Image References lets a creator guide a new still image with up to three reference images. Creators can save named references, share them with a workspace, or draw over an image to indicate composition and object placement. Runway recommends iteration for complex changes. A creative team can use the workflow to explore treatment frames, then inspect whether the selected result preserves the product details and likenesses the brief requires.
Tilting the Lens is a strategic accessibility consultancy founded by Sinéad Burke in October 2020. Its work combines education, advocacy, and design, with a team based across Europe. The consultancy’s fashion work includes adaptive-design research and collaboration with the Met’s Costume Institute. Its public adaptive-fashion report page describes research with 102 Disabled participants; the full report is provided through an email request form. For brands and studios, this is a reference for bringing people’s lived experience into product decisions, commissioning, and review. The company’s own case studies describe its contribution; they are not independent measures of every client outcome.