AI generator creates images from your text description. Upload your photo — the AI will edit it by prompt, remove background, or upscale without blur. Opens in any country right in the browser — no Discord, no app to install, no card.
All images below were generated right on this site — with exactly the prompts written under them. Like a result? Click 'Repeat': the prompt and format will be filled into the form, just change the details to fit your needs.
Prompt: White wireless earbuds in an open charging case on a clean white background, soft studio lighting, light shadow under the case, photorealism, product card photo, no text
Prompt: Portrait of a smiling girl with short red hair in a beige sweater, soft window light, neutral gray background, photorealism, 85mm lens, shallow depth of field
Prompt: Laptop and coffee cup on a light wooden table, top view, warm morning sunlight from window, free space on the left for headline, minimalism, photorealism
Prompt: Plate of carbonara pasta with parmesan and black pepper on a dark wooden table, warm side lighting, close-up, shallow depth of field, food photography
Prompt: Bright living room in Scandinavian style: gray sofa, oak floor, large window, potted plants, soft daylight, wide angle, photorealism
Prompt: Summer sale: bright beach accessories — straw hat, sunglasses, flip-flops — on light sand, sunny day, wide shot, free space on the right for text, no text
Prompt: Flat vector illustration: courier on bicycle carrying a box through sunny city street, pastel colors, clean shapes, no text
Prompt: Minimalist coffee shop emblem: cup and leaf, two colors — dark green and beige, flat style, on white background, no text
Prompt: Mountain lake at dawn, fog over water, pine trees on shore, mountain reflection in water, vertical shot for stories, photorealism
Prompt: Perfume bottle of matte glass on black marble, water drops, dramatic side lighting, close-up, premium, photorealism, no text
The most common reason for 'AI draws the wrong thing' is too short a description. The model can't guess: what isn't specified, it makes up on its own. Below is the same task, written in two words versus one sentence.
«beautiful mug»
«Terracotta ceramic mug on light wooden table by window, morning sun, light steam over coffee, blurred background, photorealism»
«product photo sneakers»
«White running sneakers at 45 degree angle on clean white background, studio lighting, light shadow under sole, photorealism, marketplace product card»
The difference isn't in 'magic words,' but in the fact that the second case specifies object, background, lighting, and image purpose. That's the formula — it's below.
The 'Edit photo' tab is an AI photo editor by text. No need to select areas and deal with layers: upload a photo and write what to change. Everything you don't mention stays in place — composition, lighting, faces.
Request: «Make headphones matte black, leave everything else as is»
Request: «Replace wooden table with white marble countertop, don't change food and dishes»
Request: «Turn photo into watercolor illustration, preserve composition and facial features»
The tool finds a person, product, or object and separates it from the background. Hair, fur, semi-transparent edges, and fine details remain — output is PNG with true transparency, not a white square.
Need specifically white background, not transparent? Remove background here, then on the 'Edit photo' tab write 'place on clean white background' — or just put the PNG on a white layer in any editor.
Small supplier photo, screenshot, old archive shot — the tool increases resolution approximately 2x and fills in details so edges stay sharp, not 'blurry' like with regular stretching.
Upscaling doesn't invent what isn't there: very blurry or compressed to artifacts photo will get better, but not perfect. For such cases the combo helps: 'Upscale' → 'Edit photo' with request 'remove noise and sharpen'.
A prompt is a description of an image in words. A good prompt isn't long—it's complete: it answers five questions. Take the formula below, fill in each part, and the result will be predictable on the first try.
Matte glass perfume bottle [subject] on black marble [environment], water droplets, dramatic side light [lighting], close-up [angle], premium, photorealistic [style], no text [exclusion].
| Lighting | soft diffused · harsh studio · golden hour · backlit · neon · window light · overcast day |
|---|---|
| Angle | top-down (flatlay) · close-up · eye level · low angle · 45-degree angle · wide angle |
| Background | clean white · neutral gray · blurred (bokeh) · gradient · natural wood · marble · solid pastel |
| Style | photorealistic · flat vector illustration · watercolor · 3D render · pencil sketch · pixel art · minimalist |
| Mood | cozy · premium · fresh · strict · festive · calm |
| Technique | 85mm lens · shallow depth of field · long exposure · macro |
"Beautiful mug"—what kind, where, in what light? The model will choose for you, and almost certainly not what you imagined.
Cat, sunset, city, space, and retro style in one line makes a mess. One scene—one prompt.
An avatar on a horizontal image will need cropping. Choose square, horizontal, or vertical right away.
If the image will have a headline, write: "empty space on the right", "free area at the top".
Text inside the image is every generator's weak spot. Generate without text, add the words in an editor.
In "Edit photo" describe only the change. The AI will keep everything else.
The first result is a draft. Adjust one word, change the lighting or angle, and run it again: two or three iterations almost always give you what you need.
Below are real tasks that come up most often, and prompts that work for them. Click on a prompt — it will be inserted into the form.
A neural network for product cards is needed when supplier photos are poor or missing. Generate the product on a white background, then remove the background or increase resolution to meet Amazon and eBay requirements.
Avatar for Telegram, social media, or team in unified style: describe person, lighting, and background. Want an avatar from your photo — upload it in 'Edit photo' and describe the style.
Cover for Instagram, Facebook, TikTok and YouTube, book, or track — horizontal or square format, calm background with space for title.
Neural network gives logo idea in seconds: ask for emblem without text, two or three colors and clean background, then refine in vector.
Instead of an hour on stock sites — illustration for specific slide in unified style for entire presentation.
Banners, posts, stories — several variants for A/B testing in a couple minutes. Ready creatives can be run through uniquifier immediately.
The service is free. To prevent bot abuse, there are daily limits on operations — one operation is one image (generation, edit, background, or enlargement). Failed requests don't count.
Opened the page — and you can work. Nothing to enter.
Click 'Login via Telegram', confirm login in bot and subscribe to channel. Takes half a minute.
For those who work with images every day. Set up directly in bot.
Limits reset at 00:00 Moscow time.
Until recently, to get an image for a task, you had to search stock sites, negotiate with a designer, or learn to draw yourself. Now it's enough to describe what you need: an AI image generator creates a picture from a description in half a minute, and a neural network for photo editing modifies an existing image based on a text prompt. This page combines both tools in one window, with background removal and resolution upscaling added — the four most common image operations.
Who needs this. Marketplace sellers — for product cards and replacing poor supplier photos. Marketers and arbitrage specialists — for banners, creatives, and covers for A/B tests. Bloggers and SMM managers — for post covers, stories, and avatars. Website owners — for illustrations for sections and articles. Everyone else — when you need a picture 'like this' but can't find it anywhere.
How it works. Text-to-image generation relies on a model trained on 'description-image' pairs: the more precise the description, the closer the result. That's why so much space on the page is devoted to prompts: formulas, glossary, common mistakes, and live examples with results. Description-based editing uses the same model but takes your photo as a base and only changes what you asked for. Background removal and upscaling are separate specialized models trained specifically for those tasks.
The service opens in any country right in the browser — no Discord, no app to install, no card needed, and requires no registration. Daily limits protect it from bots and reset every night; logging in via Telegram and subscribing to the channel raise the limit, boosting the channel or sending stars make it available for daily use.
