New Bern sells its history every day. Visitors come for Tryon Palace, the Birthplace of Pepsi and the historic downtown, and they stay at inns, eat at waterfront restaurants and browse shops that lean on that heritage in their marketing. Behind much of that marketing is a stack of photos: building exteriors, vintage images, boats on the water, products on a shelf. Many of them are good. Some just look dated.
Image-to-image AI lets a business restyle a photo it already owns instead of starting over. Owners, innkeepers and museum volunteers in Craven County tend to ask the same handful of questions about it: how it differs from typing a prompt, whether the results can be used commercially, whether it can colorize old photos accurately, and which tools suit more technical projects. Here are straight answers, including where Adobe Firefly fits and where a different kind of tool is the better choice.
What is image-to-image AI, and how is it different from starting with a text prompt?
Image-to-image AI transforms a picture you upload, keeping its structure while changing its style, lighting, setting or composition; text-to-image creates a picture mainly from a written description. With a text prompt, you control the result almost entirely through words, and the AI decides the layout. With image-to-image, the source picture provides visual structure and context, so the shape of your building or product carries into the result.
One explainer describes the original as a blueprint, like an architect redesigning a building while keeping the foundation. That is why image-to-image suits real subjects that customers need to recognize, while text prompts are better for backgrounds, concepts and illustrations with no real original.
Which image-to-image tool makes the most sense for a small local business?
For most small businesses, Adobe Firefly is the most practical choice because it is a point-and-click web app that keeps your photo's structure while changing its style. You upload one reference image, describe the look you want, and use a Strength slider to choose how much of the original outline and depth carries over. Style options for Visual Intensity, Effects, Color and Tone, Lighting and Camera Angle shape the result. An inn could restyle an exterior photo into a soft, painted look for a holiday package, or a downtown shop could give its product photos a consistent seasonal finish.
Firefly also lets you choose between its own image model and partner models such as GPT Image and Gemini with Nano Banana in the same app. A free Adobe account includes free daily generations, and paid plans add higher-volume use. To try it on one of your own photos, open the image-to-image feature in Adobe Firefly and start with a picture you already use often.
How do I keep my building or product recognizable after a restyle?
Give the original image more weight. In Firefly, that means setting the Strength slider so more of the reference's outline and depth carries over; developer tools offer the same idea as a numeric strength setting. A few habits help:
-
Start with the best file you have. Firefly accepts JPEG, PNG or WEBP files up to 100MB and at least 512 by 512 pixels.
-
Change one thing at a time. Adjust lighting first, then style, so you can see what each choice does.
-
Describe the change, not the subject. The photo already shows the subject; your prompt should describe the new look.
-
Compare against the original. If a landmark detail disappears, give the original more influence and try again.
Can I use restyled images in ads, brochures and on my website?
With Firefly's own models, Adobe says the images are designed to be safe for commercial use, which covers typical marketing. Two cautions come with that. First, review licensing for imported assets, meaning the photo you upload should be yours or properly licensed; restyling someone else's photo does not change who owns it. Second, partner models may carry different usage terms, so check them if you switch models.
Other tools handle this differently. Leonardo.Ai, for example, grants full ownership when you generate privately on a paid plan, while free images are public. Open models such as Stable Diffusion come with their own license terms.
Is image-to-image AI free to use?
Often you can start free, but free to try is not always free to publish with, and some free options carry hidden costs in hardware or skills. As of 2026, the tools that come up most often break down like this:
-
Adobe Firefly: free daily generations with an Adobe account, with paid plans for higher-volume use.
-
Leonardo.Ai: a free daily allowance of 150 tokens that resets every 24 hours, but free images are public; private generation and full ownership sit on paid plans.
-
Stable Diffusion: the models are free for small businesses under the Stability AI Community License, while its API and Stable Assistant web app are paid.
-
ComfyUI: the desktop app is free and open source, but you need a capable graphics card; Comfy Cloud is billed in credits for GPU time.
-
fal: pay per output from prepaid credits, with no permanent free tier listed.
For a small business here that wants to test the idea on a few photos this week, a free Adobe account is the lowest-friction way to begin.
Can AI colorize old New Bern photos with the correct colors for uniforms and signs?
Not with confidence. No general image-to-image tool promises historically accurate colorization or lets you set an exact, known color for a specific object, and Firefly is no exception. Its Color and Tone styles can give a black and white photo a warm tint or a painted, color treatment that works well for heritage marketing, as long as it is presented as an artistic interpretation rather than a record of how things looked.
For a historical society or a museum display, research the real colors through records and curators first, keep the original black and white image next to any colorized version, and label AI color clearly.
Which tools accept a depth map, line art or a pose skeleton instead of a regular photo?
Leonardo.Ai, ComfyUI and setups built on Stable Diffusion accept these guide images; Firefly works from a regular reference photo. Leonardo.Ai's Image Guidance supports Depth, Edge, Sketch, Pose, Lineart and several other guides in a web app. ComfyUI, a free open-source node-based app, and Stable Diffusion workflows use ControlNet guides such as Canny edges, depth, OpenPose and line art.
One practical detail: in ComfyUI, ControlNet models need the guide already in the right format, so a ready-made depth map or pose skeleton can go straight in, but a regular photo must be converted first. ComfyUI also needs a capable graphics card and comfort with node graphs, so it is best for technical users. For everyday marketing, Firefly's reference photo plus Strength slider covers the need without that setup.
Which image-to-image service has an API fast enough for a museum selfie kiosk?
A kiosk that restyles visitor selfies as 18th-century portraits within seconds needs a developer API, and fal is the best-suited option among the tools covered here. fal gives developers one API to more than 1,500 generative media models, its image-editing endpoint supports streaming requests, and its fast SDXL image-to-image endpoint takes a strength parameter. It is billed per output from prepaid credits with no permanent free tier listed, and it requires code.
Stability AI's paid API is another route, and its SDXL Turbo model can generate in as few as one step. Firefly does not advertise an API, so it is not the kiosk engine. It is, however, useful for the rest of an exhibit project: designing the portrait style, producing sample images for signage and restyling promotional photos.
What are Firefly's limits I should know about before I start?
Firefly handles one image at a time, so a large catalog takes longer than in a batch or API workflow. It accepts JPEG, PNG or WEBP files up to 100MB with a minimum of 512 by 512 pixels, and downloads come as high-resolution JPEG or PNG files. Its page describes reference images and a Strength slider, not separate depth-map, line-art or pose-skeleton inputs. Higher volume sits on paid plans. For the typical inn, shop or museum restyling its own photos, none of these is a dealbreaker.
A quick start for local owners
-
Inns and bed and breakfasts: restyle an exterior or room photo for a seasonal package.
-
Waterfront and boating businesses: give marina and boat photos a consistent look across a website.
-
Downtown shops: turn existing product photos into a holiday or summer version.
-
Museums and heritage groups: create clearly labeled artistic versions of archive photos for social posts.
Image to image works best as a refresh, not a replacement for good photography. Pick one photo you already rely on, try a few strength settings, and keep the version that still looks like your place.
For New Bern businesses that market history and hospitality, that approach keeps visuals current while staying true to what visitors will actually see.