{
 "version": "https://jsonfeed.org/version/1.1",
 "title": "AI Pics blog",
 "home_page_url": "https://aipics.geenius.io/blog/",
 "feed_url": "https://aipics.geenius.io/blog/feed.json",
 "description": "Practical posts on describing pictures, seeds and variations, keeping a history of runs and being honest about simulated previews, from the team building AI Pics.",
 "language": "en",
 "items": [
  {
   "id": "https://aipics.geenius.io/blog/say-what-a-preview-is/",
   "url": "https://aipics.geenius.io/blog/say-what-a-preview-is/",
   "title": "Say what a preview is",
   "summary": "Why AI Pics names its Preview as simulated, what the Preview actually does, and why that is the honest starting point for pictures from a sentence.",
   "date_published": "2026-10-18T09:00:00Z",
   "authors": [
    {
     "name": "The AI Pics team"
    }
   ],
   "tags": [
    "preview",
    "honesty"
   ],
   "content_text": "Most picture tools show a finished image and let you assume the rest. We want to show you the machine, so we start by saying what ours is.\nWhat the Preview is\nWhen you press Generate in AI Pics today, a Preview makes the pictures. It runs in your browser and paints on a canvas. From your words and a seed it picks a palette that leans with the words, chooses a motif, and sets your sentence in type. The same words and the same seed always give the same picture. If you start from another picture, the Preview paints over it.\nWhat it is not\nIt is not an image model. It does not understand your sentence and it does not draw what you described. If you write a lighthouse at dusk, you will not get a lighthouse. You will get a picture with colours that feel a bit like dusk and your sentence on it.\nWhy ship it\nBecause everything around the maker can be built, tested and used without a vendor, and that is most of the product. The composer, the shapes, the counts, the seeds, the history of runs with their settings, variations, collections, favourites, downloads and saving to Geenius OS Media with provenance all work with the Preview. A real model plugs into the same place, and nothing else has to change.\nWhat is connected\nA Gemini image model is connected in code. It runs on the server and works only where its key is set, and whether it is switched on for the hosted app is something we say we have to confirm. The composer names the model that will make each run, and every picture keeps the name of the model that made it. The Imagen models were retired by Google, so that provider is kept only so older pictures can name it.\nHow to use the Preview well\nUse it for placeholders and layout mock-ups, where a stand-in picture of the right shape is what you need. Replace it before anything goes out. Do not use it as a picture of the thing you described.\nWhy this matters\nIf a tool lets you believe a picture is something it is not, you stop trusting the rest of what it says. We would rather you trust the history and the provenance.\nSee what makes the pictures."
  },
  {
   "id": "https://aipics.geenius.io/blog/seeds-and-repeatable-pictures/",
   "url": "https://aipics.geenius.io/blog/seeds-and-repeatable-pictures/",
   "title": "Seeds make a picture repeatable",
   "summary": "What a seed is in AI Pics, how it makes a result repeatable, how variations change it, and why keeping the seed in the history matters.",
   "date_published": "2026-10-18T09:00:00Z",
   "authors": [
    {
     "name": "The AI Pics team"
    }
   ],
   "tags": [
    "seeds",
    "craft"
   ],
   "content_text": "A seed is a number. It is the smallest idea in picture-making and one of the most useful.\nWhat it does\nA picture is made from your words plus some randomness. The seed fixes the randomness. With the same words, the same shape and the same seed, you get the same result. Change only the seed, and you get another take of the same idea.\nIn AI Pics\nThe composer lets you set a seed, or leaves it to chance. Either way, every run keeps its seed in the history, together with the prompt, the shape, the count and the model. So you can come back to a run from last week and make it again, or change one thing, such as the shape, and keep the rest.\nVariations\nMaking variations of a picture takes its settings and gives new seeds. That is why a variation looks related: the words and the settings are the same, and only the randomness differs. You can make one to four at a time.\nSeeds and the Preview\nWith today's Preview, the seed also decides the palette and the motif. That makes the Preview a good place to see how seeds behave: change one digit and the picture changes, change nothing and it is identical.\nSeeds and a real model\nWith a real model the seed is a request, not always a promise. Some models honour it exactly, some roughly. The history shows the seed used either way, which is why it is worth keeping.\nA habit\nWhen you like a picture, look at its seed in the history. When you make a variation, note which seed you liked. When you hand a picture on, hand the seed and the prompt with it: saving to Geenius OS Media does that for you, as its provenance.\nWhat it does not do\nA seed does not make a picture better. It makes it repeatable, and repeatability is what lets you improve a picture on purpose.\nOne more reason to keep them\nWhen you hand a picture to a colleague, the seed and the prompt are the recipe. With them they can make the same picture again, or change one thing. Without them they have a file and a guess. The history and the saved provenance make sure the recipe never gets lost.\nSee history and variations."
  },
  {
   "id": "https://aipics.geenius.io/blog/a-history-beats-a-gallery/",
   "url": "https://aipics.geenius.io/blog/a-history-beats-a-gallery/",
   "title": "A history beats a gallery",
   "summary": "Why AI Pics keeps every run with its settings instead of only the best pictures, and how that helps you repeat, vary and explain a result.",
   "date_published": "2026-10-18T09:00:00Z",
   "authors": [
    {
     "name": "The AI Pics team"
    }
   ],
   "tags": [
    "history",
    "product"
   ],
   "content_text": "Most picture tools show you a gallery of results. A gallery is a good place to look. It is a poor place to work, because it forgets how each picture was made.\nWhat a run is\nIn AI Pics, one press of Generate is a run. A run is kept with its prompt, shape, number of pictures, seed, provider and model, and the state of each picture. The pictures appear one by one, and the run stays in the history.\nWhat you can do with that\nReuse the settings of a run for a new one without retyping. Retry a run that failed. Compare two runs that differ in one setting. Delete a run and its pictures when you are done with it. And when someone asks how a picture was made, you can say exactly.\nFailed is information\nA run can fail: a model is not ready, a key is missing, the page was closed at the wrong moment. The history keeps the failure and its reason instead of hiding it, and a failed picture is not counted against your month.\nThe library is separate\nPictures you want to keep go into collections and favourites, and you can search them by the words in their prompt. The library is on every plan, including Pro, which does not make pictures at all. That is useful by itself if you bring pictures in from elsewhere or from Geenius OS Media.\nProvenance leaves with the picture\nWhen you save a picture to Media, the prompt, model, seed and settings go with it. A picture in your library that cannot tell you where it came from is a small liability. One that can is a small asset.\nWhat it does not do\nIt does not judge which picture is good. It does not group similar pictures for you. It remembers.\nA small example\nImagine you made four pictures last Tuesday and liked the second. In a gallery you have the picture. In the history you have the picture, the exact words, the shape, the seed and the model, so on Friday you can make four more like it, or the same one in a different shape, without guessing what you typed. That is the whole difference, and it is quietly the most useful thing the product does.\nSee history and Geenius OS Media."
  },
  {
   "id": "https://aipics.geenius.io/blog/queue-it-let-a-page-make-it/",
   "url": "https://aipics.geenius.io/blog/queue-it-let-a-page-make-it/",
   "title": "Queue it, and let a page make it",
   "summary": "How an AI assistant queues AI Pics runs through Geenius OS and why a browser Preview run is made when somebody has the app open.",
   "date_published": "2026-10-18T09:00:00Z",
   "authors": [
    {
     "name": "The AI Pics team"
    }
   ],
   "tags": [
    "ai",
    "mcp"
   ],
   "content_text": "If you work with an AI assistant, you can ask it to make pictures in AI Pics. There is one detail that is easy to miss, and we would rather you heard it from us.\nWhat your AI can do\nThrough Geenius OS your AI gets a picture tool and a collection tool. For pictures it can check which providers are ready, queue a run with a prompt, a shape, a count and a seed, list runs and pictures, read one, check usage and delete a run with a confirmation. For collections it can list, create and add or remove pictures.\nWhat happens when it queues a run\nThe run goes into the history as queued. The Preview, today's maker, runs in a browser. So a queued run is made as soon as somebody in the workspace has AI Pics open. Your AI cannot make it on its own; it should tell you to open the app and give you the address.\nWhy a browser\nThe Preview paints on a canvas and needs no server and no key. That keeps the whole product runnable without a vendor. A real model runs on the server instead, and once its key is set, a queued run is made without anyone watching.\nWhat your AI should say\nThat it queued a run, that the pictures come from a simulated Preview unless a real model is ready, and which provider is ready. It should not promise a result. Asking for the providers first gets that answer.\nPlans\nMaking pictures needs Premium or Max, whoever asks. The door is on every plan, and a Pro workspace gets the library and nothing queued.\nA good first ask\nAsk your AI to check which providers are ready, queue one run of two square pictures with the Preview and give you the run id. Then open the app and watch the pictures arrive.\nA note on waiting\nA queued run that nobody opens just waits. That is not a failure, and nothing is lost: the run stays in the history as queued until a page claims it, and a page that is closed hands the run on to another. If it matters that a run finishes tonight, open the app yourself or leave a tab open.\nSee use with your AI."
  },
  {
   "id": "https://aipics.geenius.io/blog/a-library-for-pictures/",
   "url": "https://aipics.geenius.io/blog/a-library-for-pictures/",
   "title": "A library for pictures, on every plan",
   "summary": "Why the AI Pics library of collections, favourites, search and Media is on every plan including Pro, and how to use it without making pictures.",
   "date_published": "2026-10-18T09:00:00Z",
   "authors": [
    {
     "name": "The AI Pics team"
    }
   ],
   "tags": [
    "library",
    "plans"
   ],
   "content_text": "Pro does not make pictures. That sounds like a limit, and it is a decision.\nTwo jobs\nAI Pics does two jobs. One is making pictures, which is the AI feature and needs Premium or Max. The other is keeping them: collections, favourites, searching by the words in the prompt, downloads and Geenius OS Media. The second job is useful even if you never press Generate, so it is on every plan.\nWhat Pro is good for\nA place to keep pictures that came from somewhere else, with names and collections. A way to bring pictures in from your Geenius OS Media library and save them back. A searchable history of what you made on a bigger plan if you change plans. And a door for your own AI to keep your collections tidy.\nWhat Pro is not\nIt is not a trial of making pictures. If you want to make them, choose Premium or Max. We do not offer a free plan or a trial period, and a promo key from us can grant a short trial to people we invite.\nSearching\nSearch looks at the words in the prompts. If you name your pictures by what they are for, in the prompt, you will find them later. Collections give them a place.\nProvenance\nSaving to Media puts the prompt, model, seed and settings with the picture. That is part of the library too, and it works on Pro.\nCounting\nMaking pictures is counted per month: 2,000 on Premium and 5,000 on Max, and a failed picture is not counted. Pro counts none because it makes none.\nBe honest with yourself\nToday the maker is a simulated Preview. A library is the part of AI Pics that is fully real today.\nA fair question\nWhy pay for a library if you cannot make pictures? If you have pictures already, or you work from Geenius OS Media, or your team needs one tidy place with search and collections and downloads, the library alone is enough. If you want to make pictures, the bigger plans are there. We would rather offer the plain choice than hide the library behind the maker.\nSee collections and plans and limits."
  },
  {
   "id": "https://aipics.geenius.io/blog/shapes-are-a-decision/",
   "url": "https://aipics.geenius.io/blog/shapes-are-a-decision/",
   "title": "Shapes are a decision, not an afterthought",
   "summary": "How to choose among AI Pics' eight shapes, from square to cinema, so that a picture fits where it will be used before it is made.",
   "date_published": "2026-10-18T09:00:00Z",
   "authors": [
    {
     "name": "The AI Pics team"
    }
   ],
   "tags": [
    "shapes",
    "craft"
   ],
   "content_text": "A picture that is the wrong shape for its place gets cropped, and the crop is rarely the part you wanted. AI Pics asks for the shape before it makes anything.\nThe eight\nSquare, 1:1, for posts and avatars. Classic, 4:3, for slides and photos. Portrait, 3:4, for prints and profiles. Photo, 3:2, for a camera frame. Poster, 2:3, for posters and covers. Wide, 16:9, for video and screens. Tall, 9:16, for stories and reels. Cinema, 21:9, for banners.\nChoose by the slot\nStart from where the picture will sit. A hero banner is wide or cinema. A phone story is tall. A profile picture is square. When you know the slot, you know the shape, and you do not have to crop later.\nA model may support fewer\nEach provider says which shapes and how many pictures per run it supports. The Preview supports all eight and up to four pictures. A real model may support a subset, and the composer shows the choices a model can make.\nSize\nThe longest side is limited by the model. The Preview paints at up to 1,024 pixels on the long side. That is fine for placeholders and previews on a screen, and not for print.\nVariations keep the shape\nVariations take the settings of a run, so they keep its shape. If you want the same picture in another shape, make a new run and change the shape, keeping the seed.\nFor layouts\nIf you are building a layout, make the stand-ins in the shapes of the slots, and keep them in a collection named for the project. When the real pictures come, you swap them in place and nothing reflows.\nA small rule\nDecide the shape first, the words second. A good sentence in the wrong shape is still the wrong picture.\nAn example\nSuppose a page needs a wide banner, a square avatar and a tall story card. Make three runs with the same words and the same seed, one per shape, and keep them in one collection. You get three pictures that clearly belong together, in the right shapes, and the history shows how you made them.\nSee the composer."
  }
 ]
}