Skip to content
ComfyUI WorldBETA
Sign in
Reference only

Fashion imagery in ComfyUI: separate garment fidelity, model consent and disclosure before publishing

Editing nodes can recolour or restyle garments, but the cited pages do not promise that a product is depicted accurately. This guide covers checking product fidelity, consent for real models and disclosure questions under the EU AI Act, New York model-consent text and FTC guidance, as editorial research and not legal advice.

Updated 2026-10-05Published revision: 3Source baseline: not recorded

Purpose and scope

Fashion teams use ComfyUI for lookbooks, on-model visuals and colourways. This guide separates what editing tools document from what products, models and ads require. Not legal advice; jurisdiction and contracts matter.

Source-supported context

ComfyUI's inpainting page describes a mask input that specifies which areas need modification; it does not say unmasked pixels are guaranteed identical. Source 5 The FLUX.1 Kontext Dev page lists "Character Changes Too Much" and "Composition Position Changes" as common problems and advises stating what should stay unchanged. Source 6 Neither promises exact garment reproduction.

New York's Fashion Workers Act (S9832, Labor Law article 36) defines a model's "digital replica" as a significant computer-generated or AI-enhanced representation of likeness that substantially replicates or replaces appearance or performance, excluding routine edits (section 1031(7)). A client must obtain clear and conspicuous prior written consent for creating or using one, detailing scope, purpose, pay and duration (section 1037(7)); model management companies need separate written consent for AI use (section 1035). Source 4 The text takes effect on the 180th day after it becomes law; the page records signature on 21 December 2024, about 19 June 2025 by calculation.

Article 50(4) of Regulation (EU) 2024/1689 requires deployers of an AI system that generates or manipulates image, audio or video content constituting a deep fake to disclose that it was artificially generated or manipulated. Article 3(60) defines a deep fake as such content resembling existing persons, objects, places, entities or events and falsely appearing authentic. For evidently artistic, creative, satirical, fictional or analogous works the duty is limited to disclosure that does not hamper display or enjoyment. Source 1 Article 113 sets application from 2 August 2026. Regulation 2026/1744 amends Article 50(7) and adds Article 111(4), giving providers of generative systems already on the market until 2 December 2026 for Article 50(2); it does not change Article 50(4). Source 9

The FTC Endorsement Guides Q&A (17 January 2025) says a connection a significant minority of consumers would not expect, and that affects how they weigh an endorsement, should be disclosed clearly and conspicuously. Source 2 The FTC's 14 August 2024 press release says the final rule covers reviews and testimonials by someone who does not exist, such as AI-generated fake reviews, and prohibits creating or selling them. Source 3

The FLUX.1 Kontext [dev] model card lists the FLUX.1 [dev] Non-Commercial License. Source 7 In v1.1.1 a Non-Commercial Purpose excludes revenue-generating activity; the same text says outputs may be used for any purpose including commercial, except as prohibited. Source 8 Ask the licensor how these combine.

Before you start

1. List each deliverable and its use.

2. Gather product specs, colour codes and approved photographs.

3. Gather any real-model agreements.

4. Record model, licence and revision per file (model download records).

Steps and checks

The sequence below is an editorial procedure based on the cited context, not a diagnosis already confirmed for your environment.

Step 1. Classify each output as illustrative or as a representation of a real product.

Step 2. Pick the edit method for the claim. A masked region is repainted by the model, so check recoloured areas as new pixels (inpainting, reference editing, denoise).

Step 3. Compare each output with the physical product: seams, logos, pattern, colour. Treat colourways as unverified until matched to the supplier reference.

Step 4. Before a real person enters a workflow, confirm written consent naming AI use, scope, duration and payment.

Step 5. Decide disclosure per channel and market, including whether the image could pass as a real photograph.

Step 6. Keep synthetic people out of testimonials and reviews.

Step 7. Store workflow, prompts, seeds, model records and consent references per image.

Completion check

Each image has a classification, product comparison, consent evidence, disclosure decision and record.

Limitations and cautions

Not legal advice. The New York text is the bill as printed 3 June 2024; enacted text unchecked. Only the FTC press release was read, not the rule itself. No source tests garment fidelity, and this site did not.

Original sources

Prepared 2026-10-06 from the sources above. No GPU run, quality measurement or runtime guarantee is supplied.

Was this page helpful?

Anonymous. Only a yes/no counter is stored; no account, IP address or device details.

Sources & references

Editorial research from the cited public sources, dated 2026-10-06. This site did not run or test the workflow described.

01EU AI Act, Regulation 2024/1689 (EUR-Lex)Source checked: 2026-10-0602FTC Endorsement Guides Q&ASource checked: 2026-10-0603FTC fake reviews final ruleSource checked: 2026-10-0604NY Assembly page, bill S09832Source checked: 2026-10-0605ComfyUI inpainting docsSource checked: 2026-10-0606ComfyUI FLUX.1 Kontext Dev docsSource checked: 2026-10-0607FLUX.1 Kontext dev model cardSource checked: 2026-10-0608FLUX.1 dev Non-Commercial LicenseSource checked: 2026-10-0609Regulation 2026/1744, Digital Omnibus on AI (EUR-Lex)Source checked: 2026-10-06

Report an issue · 9524287f-6735-5eb3-be11-dc630b223fbb

Keep exploring

Pages linked to this one by the editors.