One-stop AI Image Generation Platform

How Nanomelon turns a prompt into a still image

Nanomelon is a browser studio for still images. You type a description, optionally attach a reference photo, pick a model and an aspect ratio, and we call an image model. The public gallery is not decoration: each card keeps the prompt, the engine name, and the pixel size so you can see why a picture looks the way it does. If you are new, read this section once, then open the prompt guide before you spend the starter credits.

Who runs this site, how to contact us, and how money works

Nanomelon (nanomelon.top) is an independent AI image studio. Support, privacy, and takedown requests go to support@flyingmob.com. Details live on About and Contact. We are not affiliated with the upstream model vendors named in the model list—those labels describe the engine we call.

Credits are not “unlimited free generation.” New accounts receive a small starter balance (currently five credits on registration). Each successful image job costs credits shown on the Generate button (for example two credits for one image on the default model). Failed and policy-blocked jobs do not deduct credits. When the starter balance is gone, you buy more through the pricing modal or stop. Browsing Learn guides and Public Images does not cost credits. High-resolution packages may require a paid plan as stated in pricing.

AI-generated images have limits

Outputs can invent details, warp text, misrepresent products, or resemble people or brands you did not intend. Treat every file as synthetic unless you verified it. Do not present Nanomelon results as documentary photos of real events or as endorsed by celebrities. Commercial use remains your responsibility under our Terms; see also AI image copyright for commercial use.

Text-to-image is a brief, not a wish

In text-to-image mode the model must invent everything you did not name. “Sunshine” can mean weather, a color grade, or a backlight. A useful brief names a subject, a place, a light direction, and a medium (photo, watercolor, comic). The form’s ratio control is part of that brief: 3:4 for a character poster, 16:9 for a street, 1:1 for an avatar. Generating five 16:9 portraits and cropping them later wastes both composition and credits.

Image-to-image is for a change, not a new universe

If you already have a product shot or a character you like, switch to image-to-image and describe only the edit: time of day, wardrobe, background. Restating the whole scene tells the model it may replace the identity you were trying to keep. Quantity greater than one is for a stable prompt, not for the first test. Details: text-to-image vs image-to-image and models and credits.

Case paths beyond button labels

Event posters and store banners fail when the prompt invents slogans—lock the plate here, composite type offline (commercial posters). Neon looks need props, not the lone word “cyberpunk” (case study). Tabletop and indie-game art needs a fixed character sheet before outfits change (character art workflow).

A first session that does not burn the starter credits

Write the brief before Generate. One model, one ratio, quantity 1. Change a single clause. Prefer image-to-image once a face or product is good. Public gallery pages with long object lists teach more than mood fragments. Go deeper: portraits, anime vs photo, ecommerce, templates, policy blocks, community moderation.

Prompt guide Copyright Credits About Contact Privacy