Explore ImageQuarry

Questions about creating and editing images with AI

Short, direct answers about creating images from a description, editing them by describing the change, keeping what should stay, working with references and versions, choosing a model, sizing images for each channel, and what to check before you publish or upload an image. Each question has its own page with the details, the sources and related questions.

150 questions in 13 topics

About ImageQuarry

What is ImageQuarry?ImageQuarry is an AI image creation and editing workspace where you describe an image, generate it, and then change it by asking in plain words.
How is ImageQuarry different from other AI image tools?ImageQuarry is a conversational AI image workspace: you create an image from a description or an upload, then keep refining it by describing changes in plain words.
Is ImageQuarry a headshot generator?No.
What can I do with ImageQuarry?You can generate images from a description, edit them with plain-language requests, and guide the results with reference images of your own.
Can I start in ImageQuarry from an image I already have?Yes.

Generating images from text

Can AI generate images from text?Yes.
How does text-to-image AI work?A text-to-image model learns from a very large set of images paired with text which visual patterns go with which words.
What makes a good AI image prompt?A good prompt describes the scene in full sentences: the subject, the setting, the light, the camera view and what the image is for.
Why doesn't the AI image follow my prompt?Usually because the prompt is vague, contradicts itself, asks for too many things at once, or names what to leave out.
Can AI generate readable text in images?Yes, current models can render legible text, but not reliably every time.
Why does AI mess up hands?Hands are small in most training photos, appear in countless poses and are often partly hidden, so models see fewer clear examples of them than of faces.
Does AI generate the same image twice from the same prompt?No, not usually.
What is a negative prompt?A negative prompt is a separate instruction that lists what you do not want in the image, such as blur, extra fingers or text.
Should I describe the camera and lens in an image prompt?Yes, when the framing matters.
Can AI generate images in a specific art style?Yes.
Can AI generate photorealistic images?Yes.
Can AI make a poster?Yes.
Can AI make product photos?Yes.
Can AI design an interior from a description?Yes, as a concept.

Editing images with words

Can AI edit an existing image?Yes.
Can you edit a photo by describing the change in words?Yes.
Can AI change the background of a photo?Yes.
Can AI change the lighting in a photo?Yes.
Can AI remove an object from a photo?Yes.
Can AI add an object to a photo?Yes.
Can AI change the clothes in a photo?Yes.
Can AI change the color of one object in a photo?Yes.
Can AI change the composition of an image?Yes, within limits.
Can I edit an AI-generated image after it is made?Yes.
What is instruction-based image editing?Instruction-based image editing means changing an image by writing the change, such as "replace the background with a beach", instead of using editing tools.
How specific should an edit instruction be?Specific enough to name the change, where in the image it happens and what should stay the same, so the model does not have to guess.
Can I ask for several changes in one edit?Yes, but two or three related changes work better than a long list, because each extra change is another chance for the model to miss one.
Is AI image editing a replacement for Photoshop?Not entirely.

Keeping parts of an image unchanged

Can AI edit a photo without changing the face?Usually, yes, if you say so in the instruction.
Why does AI change my whole image when I ask for one change?Because most AI editors redraw the whole picture, not only the part you named.
Can AI edit only part of an image?Yes.
What is inpainting?Inpainting is filling a marked area of an image with new content while keeping the rest.
What is outpainting?Outpainting is extending an image beyond its original edges with AI.
Can AI extend an image beyond its edges?Yes.
How do I tell AI what to keep unchanged in an edit?Name it in the same instruction as the change.
Can AI keep a product exactly the same while changing the scene?Not exactly every time.
Can AI keep text and logos intact when editing?Not reliably.
Does each AI edit reduce image quality?It can, gradually.
What is a mask in AI image editing?A mask is a map that marks which part of an image the AI may change and which part it should keep.
Can AI keep the same character across several images?Often, yes, with reference images and careful checking.

Reference and source images

What is a reference image in AI image generation?A reference image is a picture you give an AI image tool to show what you mean, such as a texture, color, light, shape, style or character.
What is image-to-image AI?Image-to-image AI makes a new image from an input image plus a text prompt.
Can AI copy the style of a reference image?Yes, broadly.
How many reference images should I use?Usually one to three, each with a clear job.
What makes a good reference image?A good reference is clear, well lit and shows one thing you want: a texture, a color, a light, a shape or a character.
Can I use a reference image for color only?Yes.
Can a reference image guide the lighting?Yes.
Can I upload my own photo and edit it with AI?Yes.
What is the difference between a source image and a reference image?The source image is the picture being changed; a reference image only guides the change.
Will AI copy my reference image exactly?No.
Can I use someone else's image as a reference?Sometimes, but be careful.
Can a reference image keep a product accurate?It helps a lot, but it does not guarantee accuracy.
Can AI combine several images into one?Yes.

Versions, variations and re-editing

What is an image version in AI editing?An image version is one saved result in a chain of edits: the first image you make is version 1, and each change you ask for makes a new one.
Can I go back to an earlier version of an AI image?Yes, if the tool keeps a version history: each edit is then saved as a new version, and you can reopen an earlier one when a later edit goes wrong.
Can I edit an earlier version instead of the latest one?Yes, in tools that support branching: you select the earlier version and ask for a different change, and the later versions stay as they were.
What are image variations in AI?Image variations are several alternative results for the same request, generated side by side so you can pick the strongest.
How many variations should I generate before choosing?Two to four at a time is a practical starting point: enough to judge a direction, few enough to compare closely before you pick one to edit.
What is branching in AI image editing?Branching means continuing from an earlier version of an image instead of the latest one, so one image can lead to several separate lines of edits.
How do I compare two versions of an AI image?Put the two versions side by side at the same size and check them against your request: first what you asked to change, then what you asked to keep.
Why keep earlier versions of an image?Because AI edits can change more than you asked for, and a kept version is your fallback, your point of comparison and a place to start a new direction.
Does AI remember earlier edits in a conversation?Yes, within the same conversation: multi-turn image tools use earlier turns as context, so you can say "now make it warmer" without repeating the prompt.
Should I regenerate an image or edit it?Edit when the image is mostly right and one or two things are wrong; regenerate when the composition, subject or style is wrong.
How should I organize AI images across projects?Organize by project first, then keep each final image with the versions, prompt and references that produced it, so nothing is separated from its context.
Can I start a new image from one I made earlier?Yes.
Does ImageQuarry keep every version of an image?Yes.

AI image models and tools

Which AI image model is best?No single model is best for every image.
What is Nano Banana?Nano Banana is Google's name for the native image generation and editing capabilities of its Gemini models, used both to create images and to change them.
What is GPT Image?GPT Image is OpenAI's family of image models for generating new images from text prompts and for editing existing images with instructions.
What is a diffusion model?A diffusion model is a type of generative AI model that learns to create images by reversing a process that gradually turns training images into noise.
Does the model matter for image editing?Yes, partly.
Why do different AI models give different results from the same prompt?Because each model was trained differently, with different data and goals, so each turns the same words into a different picture.
What is the difference between an AI image generator and an AI image editor?An AI image generator creates a new image from a text prompt, while an AI image editor changes an image you already have, from an instruction or a selection.
Can I switch AI models in the middle of a project?Yes, in tools that offer more than one model: an image made by one model can be the input for an edit by another, since current models accept images as input.
Which AI image generator is most realistic?No single generator is the most realistic for every subject, because people, products, interiors and landscapes each fail in different ways.
Which AI image models handle text in images best?No AI image model gets text right every time, so read every word at full size before you use an image, whichever model made it.
Is ChatGPT good at editing images?Yes, for many everyday edits: ChatGPT can change an image you generated or uploaded when you describe the change in chat or select an area first.
What is generative fill?Generative Fill is an Adobe Photoshop feature that adds, removes or changes content inside a selection, using Adobe Firefly models and an optional text prompt.
Which AI models does ImageQuarry use?ImageQuarry works with four image models: the Google image models Nano Banana 2.1, the default, and Nano Banana Pro, plus GPT Image 2.5 Flare and Sunburst.
Which model does ImageQuarry use by default?Nano Banana 2.1, a Google image model, is ImageQuarry's default for generating and editing images, so it is the one in use unless you choose another.

Image quality and resolution

What resolution are AI-generated images?It depends on the model and the size you ask for.
Why are AI images blurry?Usually because the image is too small for where it is shown, it was compressed when saved or shared, or the model left part of the scene soft.
What is AI upscaling?AI upscaling enlarges an image and fills in new detail with a model, so it looks sharper than a simple resize.
Can AI images be printed?Yes.
What resolution do I need to print an image?About 300 pixels per inch at the final print size for prints seen up close.
Why do AI images look fake?Mostly because they are too perfect: smooth skin, flawless lighting, glossy surfaces and centered subjects, plus small details that don't add up, such as hands, reflections or lettering.
What file format are AI images saved in?Usually PNG, JPEG or WebP, depending on the tool and its settings.
Should I save an AI image as PNG or JPEG?Choose PNG for images with text, logos, sharp graphics or transparency, and for anything you will edit again.
Are AI images good enough for ecommerce listings?Often yes for lifestyle and scene images, as long as the product is shown accurately.
What image size do I need for a website hero image?Wide enough for the largest screen it fills, times the screen's pixel density.
Why does AI get small details wrong?Because small things get few pixels and little of the model's attention.
Are AI images sharp enough for print advertising?Yes, if the file has enough pixels for the printed size and the details hold up close.
What should I check before using an AI image?Check it at full size, not as a thumbnail: hands, faces, text, logos, reflections and shadows.

Aspect ratios and sizes

What is an aspect ratio?An aspect ratio is the shape of an image: its width compared to its height, written as two numbers such as 16:9 or 4:5.
What aspect ratio should I use for Instagram posts?Use 4:5 portrait, 1080 x 1350 pixels, for most feed posts: it fills the most screen of the classic feed shapes.
What aspect ratio do Instagram Stories and Reels use?9:16 vertical, full screen.
What aspect ratio does TikTok use?9:16 vertical.
What size is a YouTube thumbnail?16:9 for regular videos.
What aspect ratio works best on Pinterest?2:3, such as 1000 x 1500 pixels.
What image size works for LinkedIn posts?1.91:1 landscape at 1200 x 628 pixels or 4:5 vertical at 720 x 900 are LinkedIn's recommended ratios for single image ads, and it also lists 1:1 square at 1200 x 1200.
What image aspect ratios do Facebook and Instagram ads use?4:5 for Facebook and Instagram feeds, at 1440 x 1800 pixels, and 9:16 for Stories and Reels, at 1440 x 2560.
What is the difference between 16:9 and 9:16?They are the same shape turned on its side.
What is a 4:5 aspect ratio used for?4:5 is a slightly tall portrait shape.
Can AI generate vertical images?Yes.
Can AI change the aspect ratio of an existing image?Yes, in two ways: by cropping, which removes part of the image, or by extending it, where AI draws new area beyond the original edges.
Should I crop or extend an image to change its aspect ratio?Crop when there is enough space around the subject, because cropping changes nothing that remains.
How do I leave room for a headline in an AI image?Say where the empty space goes and roughly how much: "subject in the lower left, empty sky across the top third for a headline".
What aspect ratio should product images use?1:1 square for most online stores.

Pricing and credits

How much does AI image generation cost?AI image generation ranges from a few cents per image through a developer API to monthly plans of roughly $10 to $120, by model, size and volume.
How do credits work in AI image tools?A plan gives you a balance of credits, and each generation or edit uses some.
Why do AI image tools use credits?Because each image request has a real compute cost, and that cost changes with the model, the image size and the kind of work.
Does editing an image cost the same as generating one?Often, yes.
Why is AI image generation expensive to run?Every AI image runs a very large model on specialized hardware, paid for on each request.
Is a subscription or pay-per-image better for AI images?A subscription is usually better if you make images every week.
What is metered overage on a subscription?Metered overage means you pay for each extra unit you use after your plan's included amount is used up, instead of being stopped or moved to a bigger plan.
Do higher-resolution AI images cost more?Often, yes.
How much does ImageQuarry cost?ImageQuarry has two monthly plans: Starter at $9 with 100 image credits, and Pro at $29 with 300 image credits.
What is the difference between ImageQuarry Starter and Pro?Price and monthly usage.
What happens when I run out of ImageQuarry credits?Turn on optional metered overage to keep creating: $0.12 per extra credit on Starter, $0.09 on Pro.

Usage rights and ownership

Can I use AI images commercially?Usually, yes, if the tool's terms allow commercial use and the image does not copy someone else's protected work, likeness or trademark.
Who owns AI-generated images?The tool's terms decide who holds the rights in an output.
Can AI images be copyrighted?Only partly.
Can I sell AI-generated images?Often, yes, if the tool's terms allow it, the image infringes no rights and the marketplace accepts AI content, which often must be labeled.
Do I need to disclose that an image is AI-generated?Sometimes.
Can AI images be used in ads?Yes, if the ad follows platform policies and advertising law and the image infringes no rights.
Can I use AI to edit a photo of a real person?Yes, for many ordinary edits, but get permission before you publish.
Can AI images include brand logos or trademarks?Your own logo, yes.
Does editing a photo with AI change who owns it?No.

Privacy and uploaded images

Is it safe to upload photos to an AI image tool?It depends on the tool's policy.
Are uploaded images used to train AI models?Sometimes.
Do AI image tools keep my photos?Most keep them for some time, and the period differs by provider and setting.
Are AI image generators private?Not always.
What should I check in an AI image tool's privacy policy?Check five things: training use, retention, human review, sharing and deletion, for the plan you use, since defaults can differ.
Should I upload photos of other people to an AI tool?Only with their permission, and never sensitive images such as children, IDs or private moments.
What are Content Credentials on AI images?Content Credentials are provenance information attached to an image under the C2PA open standard, recording where it came from and how it was edited.

Creative workflow problems

What can I do when an AI image comes out wrong?Name the one thing that is wrong, ask for that change on its own, and say in the same request which parts of the image should stay as they are.
Why does AI ignore part of my instruction?Usually because the request asks for several things at once, or because parts of it conflict with each other or with the image.
Why does AI add things I didn't ask for?Because a model fills every gap your description leaves with what is typical for that kind of image, from props and people to stray text.
Why do my images drift after many rounds of edits?Because many AI editors redraw the whole image on every edit, not only the part you meant, and each round starts from the previous output.
When should I start over instead of editing again?Start over when the problem is the idea rather than a detail, such as the wrong subject, layout, viewpoint or style of the whole image.
Why does text in AI images come out misspelled?Because image models draw letters as shapes instead of typing them, so precise spelling and placement are still a known weak spot.
How do I get consistent images across a campaign?Fix what should repeat and reuse it in every request: the same reference images and the same described palette, light and framing.
Can AI match my brand colors exactly?Not reliably to the exact value: AI image models get close to a color you describe or show them, but they do not guarantee a hex or Pantone match.
Why does AI change the shape or label of my product?Because the model redraws your product instead of copying it, and fine details such as label text, logos and proportions are where redraws drift most.
How should I phrase feedback to an AI image tool?Name one concrete change, where it is, and what must stay the same, for example: "Make the light warmer from the left.

Create an image. Edit it with words. Keep refining.

Describe an image, ask for the change you want, and keep every version as you go.

Start creating Already have an account? Sign in