One real workflow
Every creative route uses the same two-image upload, optional subscriber-triggered analysis, prompt, generation, and result flow. Uploading alone never starts analysis.
About the product
A focused workspace for two-image creation
AI Image Combiner is built around a narrow promise: let two references guide one new composition, and make the active model, input roles, output choices, task state, and retention rules visible before a user commits.
Operator and public identity
Independently operated, without a fictional company or team
AI Image Combiner is a brand used and maintained by one independent individual operator. The brand is not presented as a registered company, a multi-person team, or an affiliation with any model or infrastructure provider.
“AI Image Combiner Editorial” is a site-maintained editorial label for articles. It is not a separate legal entity, a professional credential, a certification, or a claim that multiple people wrote or reviewed the content.
The operator maintains the product, policies, billing support, and correction process. The public contact channel is support@imagecombiner.app. A personal name, home address, and private location are not published on this website.
Temporary private assets
Public copy distinguishes temporary image files from longer-lived task, payment, and credit records. The privacy policy explains the current 24-hour access window.
Observable product states
The interface uses queued, processing, success, and failure states instead of inventing a provider percentage or pretending an unavailable model is live.
Editorial standard
How public claims are checked
The goal is not to sound certain about every model. It is to separate what the current product does, what official documentation states, what a project-owned example demonstrates, and what has not been tested.
Product behavior
Claims about accepted inputs, prompts, state transitions, retention, and downloads should match the implemented application. Planned behavior is labeled as planned rather than presented as live.
Provider capabilities
Model-specific articles link to current official provider documentation. Provider names, endpoints, limits, and lifecycle details can change, so dated documentation remains the source of truth.
Examples and demonstrations
Published gallery images are project-owned synthetic demonstrations. They are not customer testimonials, controlled benchmarks, or evidence that another run will reproduce the same output.
Comparisons
A capability table based on documentation is not a live quality test. Speed, cost, failure rate, and visual preference require the same inputs, settings, account conditions, date, and review method.
Example provenance
The current gallery uses synthetic images generated with OpenAI for this project. The register records the file, method, creation date, purpose, prompt summary, and modifications. Public pages identify these assets as demonstrations and never relabel them as user work.
Open the asset registerAuthorship and corrections
Articles use the brand editorial byline “AI Image Combiner Editorial.” As explained in the operator disclosure, that label identifies site-maintained content and does not claim an individual credential, professional certification, company, or team. Material revisions update the article date.
To report an inaccurate product claim, outdated provider link, rights concern, or privacy issue, email support@imagecombiner.app with the page URL and the correction requested.
Independent and source-aware
GPT Image 2 is provided by OpenAI; Nano Banana 2 and Gemini 3 Pro Image are provided by Google. Analysis and generation may independently use our DMX/New API gateway, KIE, or Cloudflare's REST AI services. The selected relay and upstream model provider process submitted references and prompts; Cloudflare requests go to api.cloudflare.com rather than a Workers AI binding. AI Image Combiner is independently operated and is not affiliated with or endorsed by those services or providers. Only a capability marked Available can be submitted because availability depends on its complete transport, request, failure, storage, and billing path—not a catalog label alone. We do not claim a live three-model or three-transport benchmark.