# Learning Center Learn Sloosh by watching it work. See how to make images and videos in seconds, then turn what works into a workflow you can run again. Web page: https://sloosh.ai/docs/learn ## Guide ### 1. Make an image from a prompt Open Image. Describe the shot, pick a model, press Generate. 1. **Describe the shot**: Write it like a brief: the subject, the setting and the light. A clear brief gets closer first. 2. **Add a reference**: Press + to upload a photo. It joins the prompt box as Image 1, ready to use. 3. **Pick a model**: Open the model chip. Each model says what it is best at, and its options show up as chips. 4. **Turn on Enhance**: Switch on Enhance and Sloosh rewrites your prompt with more detail before it reaches the model. 5. **Set the shape, size and count**: Open the shape and size chips and pick from their lists. The count is how many you get, up to 4. 6. **Check the price, then Generate**: The number on Generate is what this run costs. You see the price before you spend anything. 7. **Your image lands in My Generations**: Your images land first in My Generations and in Assets. A failed image gives its credits back. 8. **Your turn**: Write a prompt, try every chip, then press Generate. This demo uses example images and credits. Good to know: - **Pull in files with @**: The name joins your prompt, and the picture joins the box. - **Start from a template**: Templates sits next to My Generations. Recreate loads its prompt and model. - **Failed results come back**: If an image fails, its credits return. ### 2. Make a video clip Open Video. Start from a still, describe the motion, set the length. 1. **Create or Edit**: Use the rail left of the prompt box: Create makes a new clip, Edit changes one. On a phone, they are the tabs at the top. 2. **Add a start frame**: Press + and pick Upload frames. The Start frame joins the prompt box and the clip begins there. 3. **Describe the motion**: Say what moves and what the camera does. Keep the subject the same as your start frame. 4. **Pick a video model**: Open the model chip. Each model says what it does best, and its chips change to match. 5. **Set the shape, length and sound**: 16:9 is the shape and 8s the length, changed with - and +. Sound is on; the switch turns it off. 6. **Pick a camera**: The camera chip opens Camera Setup. Turn the camera, lens and aperture wheels for a look. 7. **Check the price, then Generate**: The price sits on Generate before you press it. Video costs credits per second, so a longer clip costs more. 8. **Your turn**: Describe a clip, try the chips and Camera Setup, then press Generate. Credits here are examples. Good to know: - **Camera Setup**: Three wheels, all on Auto: camera, lens, aperture. - **Start and end frames**: Some models take an End frame too. The clip moves between them. ### 3. A Space makes the same thing every time Someone set up the steps once. You fill a short form and press Generate. 1. **Open a Space**: Find a Space on Home or in Spaces and open it. This one puts you into a movie moment. 2. **Fill the form**: Fill in the form. A star means required. Here it asks for one clear photo of you. 3. **Pick a scene**: Some fields are pictures to pick from. Tap a scene card and a yellow ring marks it. This one takes up to three. 4. **Press Generate**: Press Generate to run the Space. The price sits on the button, and until the form is filled the hint under it says what is missing. 5. **Watch the results land**: Your clip lands first in Generations: a vertical 6-second moment, ready for Reels and TikTok. 6. **Look inside**: Overview shows the canvas this Space runs on. Remix gives you your own copy to change and publish. 7. **Your turn**: Drop in a photo, pick a scene, then press Generate. Everything here is an example, nothing real. Good to know: - **Files go in slots**: A Space asks for exactly what it needs: garments, a character, a pose. - **Locked canvases stay closed**: Some Spaces keep their canvas locked. You still run them; Remix is off. ### 4. Build a Space in Spacelab Each node does one job. Wires carry the result to the next node. 1. **Start a new canvas**: Open Spacelab to get an empty canvas and its toolbars. Describe your idea, start from a template, or build it yourself. 2. **Add a node**: Press + to open the node list. Prompt is the first node you need, so add it. 3. **Write the prompt**: Type into the Prompt node. Its text goes down the wire to every node it feeds. 4. **Wire it to a model**: Drag from a node's output port and let go on the canvas. The Connect to menu lists only the nodes that fit. 5. **Add the next step**: Wire the still into a Video node as its first frame. The Prompt feeds the Video too. Select a node to pick its model. 6. **Press Run**: Press Run. Each node waits for its inputs, then runs and shows its status. 7. **Your turn**: Add a node, draw a wire and press Run. Try a wrong wire too: the canvas only connects ports that fit. Good to know: - **Ports have colours**: Inputs plug in on the left of a node. Results leave on the right. - **Five toolbars**: Name and status, Share and Publish, zoom, the editing tools, then Talk to AI and Run. ### 5. Watch a Space run Press Run and every node takes its turn. Pick a shape and watch it work. 1. **Edit a photo**: The product or the room stays exactly the same. Only the light, backdrop or scene changes. 2. **One brief, many versions**: One input becomes a set of results that belong together: every shade, backdrop or size. 3. **Photo to video**: Get the still right first, then animate it. The still becomes the first frame of the clip. 4. **Before and after**: Give it two keyframes, the before and the after. The model fills in the motion between them. 5. **Talking presenter**: The same person on camera, with captions. A character keeps the face the same, and a script sets what they say. 6. **Short film**: Each shot is its own clip. Video Concat joins them, then music and captions go on top. Good to know: - **Every node shows a status**: Pending, Running, Completed, Failed, Skipped or Awaiting approval. - **Require approval**: Turn it on in a node's settings, and the run waits at that node until you approve it. - **One list, many runs**: An iterator runs everything after it once per item. Bundle gathers the results. - **Run till here**: A node's own Run button runs just that node, on the inputs it already has. ### 6. Publish your canvas as a Space Your input nodes become the form. People fill it in and press Generate. 1. **Each input node is one field**: Each input node becomes one field: an upload for an Image Input, a list for a Select. A Prompt stays on the canvas. 2. **Name fields like questions**: Rename a node and its field's label changes too. Your photo reads better than Image Input. 3. **Make the extras optional**: A wired field is required. Make the ones a run can do without optional: left blank, they use the value set on the canvas. 4. **Run it three times**: Run it three times in a row on real inputs. If all three come out good, the Space is ready. 5. **Open Publish**: Open Publish and press Autofill to draft the details from your canvas. Add a thumbnail and pick runs for the showcase. 6. **Your turn**: Rename a field or make it optional. The form follows every change, and so does the canvas. Good to know: - **The slug is set once**: A Space keeps the web address it gets the first time you publish. - **Workspace or community**: Untick Also publish to community and only your workspace sees the Space. A community Space needs at least one showcase run. - **Autofill is free**: It drafts the details from your canvas, up to 10 times a day per canvas. It never replaces a FAQ you wrote. - **Unpublish keeps the canvas**: Unpublish takes down the Space page, its templates and shared links. Your canvas stays. ### 7. Describe it and get a first draft Talk to AI adds nodes to the canvas from plain words. 1. **Open Talk to AI**: Open Talk to AI, beside Run. It adds and edits nodes from what you type. 2. **Say what you want**: Say what goes in, what should change and what you want out. One short paragraph is enough. 3. **Check the draft**: The nodes land on the canvas, wired up. Check every wire before you run: drafts can be messy. 4. **Goal Mode**: Type /goal and the result you want. Goal Mode builds the canvas, test runs it and checks the results. Test runs use credits. 5. **Your turn**: Describe a Space in one sentence and send it. The demo matches your words to one of six shapes. Good to know: - **Ask about any node**: Not sure what a node does or which one to use? Ask, and the answer comes back in the chat. - **Start from an idea**: On a new canvas, type your idea in Describe Your Idea and Sloosh builds a first version for you. ### 8. Everything you make lands in Assets Your images, videos and Space results, plus the elements you reuse. 1. **Generations**: Generations holds everything your workspace made in Image, Video, Spaces and Spacelab, newest first. 2. **Open, copy, download**: Click a tile to open it full size. Hover it to copy it, download it or use it in Studio. 3. **Elements**: Elements are the things you reuse: a character, a pose, a background. Each one gets a tag. 4. **Use @ in a prompt**: Type @ in a prompt and pick a tag. The name joins the text, the picture slides out above. 5. **Your turn**: Type @ in the prompt and pick an element. Its name joins the text and its picture the tray. Good to know: - **Four categories**: Character, Pose, Background or Others. Filter Elements by any of them. ### 9. You see the price before you press Credits pay for every run. The cost shows on the button first. 1. **The price is on the button**: Every run shows what it costs on its button, before you press. In Image and Video, on Generate. 2. **A bigger model costs more**: Pick another model and the number on the button changes with it. Bigger models cost more. 3. **Press, and the cost lands in Credit history**: Sloosh holds the price while it runs, then charges what it really cost, line by line on Usage & billing. 4. **A failed image comes back**: If an image fails, a refund line puts its credits straight back. You pay only for results. 5. **Your turn**: Pick a model and press Generate, then watch the line land in Credit history. These credits are examples. Good to know: - **Where the price shows**: On Generate in Image, Video and a Space. On Run in Spacelab. ### 10. Let an agent use Sloosh for you Connect ChatGPT, Claude, Cursor or Claude Code. They can find a Space, check the price and run it. 1. **Open MCP**: MCP sits in the top bar. Pick your agent from the tabs: ChatGPT, Claude, Cursor or Claude Code. 2. **Add the server**: Follow your agent's short steps. In Claude Code it is one command, then /mcp to sign in. 3. **Sign in**: A browser window opens. Sign in with your Sloosh account and pick the workspace to use. 4. **Set your limits**: Nothing paid runs until a daily limit is set. Above Ask before, the agent asks you in chat first. 5. **Ask for something**: Ask in plain words, as you would ask a person. Try: list my Spaces and what each one needs. 6. **Your turn**: Slide the daily limit, then send a request. Ask it to list your Spaces or check a price. Good to know: - **Pro and Max**: Agent MCP access comes with Pro, and Max has everything in Pro. ## Tutorials - [Your first image](https://sloosh.ai/docs/learn.md#image): One image from a prompt, saved in Assets. - [Your first video clip](https://sloosh.ai/docs/learn.md#video): A clip from one still, saved in Assets. - [Run your first Space](https://sloosh.ai/docs/learn.md#spaces): Results from a Space someone else built. - [A tour of Spacelab](https://sloosh.ai/docs/learn.md#spacelab): You know where to add, wire and run nodes. - [Publish your first Space](https://sloosh.ai/docs/learn.md#publish): A Space with a clear form that people can run. - [Draft a Space with Talk to AI](https://sloosh.ai/docs/learn.md#talk-to-ai): A first draft on the canvas from one paragraph. - [Find and reuse your assets](https://sloosh.ai/docs/learn.md#assets): You pull a saved element into a prompt with @. - [Read the price before you spend](https://sloosh.ai/docs/learn.md#credits): You know what a run costs and why. - [Connect Sloosh to Claude](https://sloosh.ai/docs/learn.md#agents): Claude can list your Spaces and check a price. - [Build your first Space](https://sloosh.ai/docs/learn/tutorials/first-build): A Space that changes the backdrop and keeps the product. - [Read the canvas behind a Space](https://sloosh.ai/docs/learn/tutorials/read-space): You can point at the node behind every field on the form. - [Wire the right ports](https://sloosh.ai/docs/learn/tutorials/wire-types): You can tell a good wire from a bad one before you drag. - [Paste photos onto the canvas](https://sloosh.ai/docs/learn/tutorials/paste-drop): Three product photos become cutouts in one run. - [Day to dusk listing photos](https://sloosh.ai/docs/learn/tutorials/day-to-dusk): A blue hour version of one listing photo, resized for the listing. - [Tidy room photos](https://sloosh.ai/docs/learn/tutorials/clean-room): The same room with the clutter gone and every fixture kept. - [Four backdrops, one product](https://sloosh.ai/docs/learn/tutorials/backdrops): The same sunglasses on four backdrop colours. - [One polish, every shade](https://sloosh.ai/docs/learn/tutorials/shade-range): A full shade range from one product photo. - [Catalog shots from one photo](https://sloosh.ai/docs/learn/tutorials/catalog-shots): Front, back, side and close-up shots of the same model. - [A lookbook that matches](https://sloosh.ai/docs/learn/tutorials/lookbook-grade): Every shot in the set shares one light and one grade. - [Product still to a short clip](https://sloosh.ai/docs/learn/tutorials/still-to-motion): A short clip with the product sharp in every frame. - [Splash loop for a drink ad](https://sloosh.ai/docs/learn/tutorials/splash-loop): A slow motion splash clip from one still. - [From one line to a short clip](https://sloosh.ai/docs/learn/tutorials/writer-film): A still and a clip from a single sentence. - [Before and after reveal](https://sloosh.ai/docs/learn/tutorials/before-after): A clip that moves from the bare room to the finished one. - [Pay for video only when the still is right](https://sloosh.ai/docs/learn/tutorials/checkpoint-video): The run stops after the still and waits for your yes. - [Redo one shot with Run till here](https://sloosh.ai/docs/learn/tutorials/run-till): You redo one shot without running the whole film again. - [A presenter who talks about your product](https://sloosh.ai/docs/learn/tutorials/presenter): A captioned shot of the same presenter talking about your product. - [Neighbourhood reel](https://sloosh.ai/docs/learn/tutorials/neighbourhood): A scored, captioned reel of the area around a listing. - [Ad matrix: four looks, three sizes](https://sloosh.ai/docs/learn/tutorials/ad-matrix): Twelve ad images, every look in every size. - [Have an agent build a Space](https://sloosh.ai/docs/learn/tutorials/agent-build): Your agent builds this canvas in Spacelab for you. ## Words - **Image**: The page where you make a picture from a prompt. Pick a model, press Generate. - **Video**: The page where you make a clip from a prompt or a still. - **Model**: The AI that makes the result. Each one says what it is best at. - **Prompt**: What you type to describe the result. - **Enhance**: Rewrites your prompt with more detail before it reaches the model. - **Generation**: One result, an image or a clip. It lands in Generations. - **Space**: A set of steps that runs the same way every time. People fill a short form and press Generate. - **Spacelab**: Where you build a Space: nodes joined by wires on one canvas. It sits in the top bar. - **Node**: One step on the canvas. Each node does one job. - **Port**: The round plug on the side of a node. Wires join two ports. - **Wire**: Carries a result from one node to the next. - **Run**: Do the work. A run costs credits. - **Publish**: Turn your canvas into a Space other people can run. - **Remix**: Make your own copy of a Space to change. - **Assets**: Everything you made, in Generations, plus your Elements. - **Element**: An image or video you keep to reuse, with a tag you can @mention. - **Credits**: What a run costs. You see the number before you press. - **MCP**: The connector that lets an agent use Sloosh. - **Daily limit**: The most credits your agents can spend in a day. No agent can raise it. ## Nodes - [Prompt](https://sloosh.ai/docs/nodes-input.md#prompt): Text you write on the node, passed on to a model. - [Text Input](https://sloosh.ai/docs/nodes-input.md#text-input): Plain text with no variables. - [Image Input](https://sloosh.ai/docs/nodes-input.md#image-input): One image. - [Reference Image Input](https://sloosh.ai/docs/nodes-input.md#reference-image-input): Wires exactly like Image Input. - [Multi Image Input](https://sloosh.ai/docs/nodes-input.md#multi-image-input): Several images on one node, sent down one wire as a set. - [Background Input](https://sloosh.ai/docs/nodes-input.md#background-input-characters-and-poses): Pickers for backgrounds, people and poses. - [Characters](https://sloosh.ai/docs/nodes-input.md#background-input-characters-and-poses): Pick one or more characters from the Spaces library. - [Poses](https://sloosh.ai/docs/nodes-input.md#background-input-characters-and-poses): Pick one or more poses from the Spaces library. - [Image Pool](https://sloosh.ai/docs/nodes-input.md#image-pool): A whole folder of images, with no limit on how many. - [Video Input](https://sloosh.ai/docs/nodes-input.md#video-input): One video clip. - [Document](https://sloosh.ai/docs/nodes-input.md#document): A file for a model to read: a brief, a deck, a spec or a sheet. - [URL Input](https://sloosh.ai/docs/nodes-input.md#url-input): A web page read from its link: a product page, a brand's about page, a public Google Doc. - [Pinterest Scraper](https://sloosh.ai/docs/nodes-input.md#pinterest-scraper): Searches Pinterest for a keyword and returns the pins it finds. - [Aspect ratio](https://sloosh.ai/docs/nodes-input.md#aspect-ratio): Pick a ratio and wire it into the Aspect ratio port of Image Model or a video node. - [Number](https://sloosh.ai/docs/nodes-input.md#number): A number, with an optional min and max. - [Select](https://sloosh.ai/docs/nodes-input.md#select): Your own list of choices. - [Any LLM](https://sloosh.ai/docs/nodes-generate.md): A text model. - [Image Model](https://sloosh.ai/docs/nodes-generate.md): Makes an image from a prompt and, optionally, reference images. - [Video](https://sloosh.ai/docs/nodes-generate.md): Makes a clip from a prompt, optionally starting and ending on images you give it. - [Video (references)](https://sloosh.ai/docs/nodes-generate.md): Makes a clip that follows reference images, and for some models reference videos and audio. - [Prompt Enhancer](https://sloosh.ai/docs/nodes-edit.md#prompt-enhancer): Turns a short description into a detailed prompt. - [BG Remove](https://sloosh.ai/docs/nodes-edit.md#bg-remove): Cuts the subject out and returns a PNG with a transparent background. - [Auto Mask](https://sloosh.ai/docs/nodes-edit.md#auto-mask): Selects the thing you name, such as the bag or the shoes, and returns a cut-out that keeps only that thing on a transparent background, plus a black-and-white mask (white is selected). - [Crop](https://sloosh.ai/docs/nodes-edit.md#crop): Crops to a width and height in pixels, centred unless you set X and Y offsets. - [Rotate & Flip](https://sloosh.ai/docs/nodes-edit.md#rotate-flip): Rotates an image or video by 90, 180 or 270 degrees clockwise, and flips it horizontally or vertically. - [Colour Grade](https://sloosh.ai/docs/nodes-edit.md#colour-grade): Adjusts exposure, contrast, highlights, shadows, temperature, tint, saturation and hue, on an image or a video. - [Upscaler](https://sloosh.ai/docs/nodes-edit.md#upscaler): Enlarges an image 2x, 4x or 6x with Topaz. - [Image Resize](https://sloosh.ai/docs/nodes-edit.md#image-resize): Resizes an image to exact pixels. - [Video edit](https://sloosh.ai/docs/nodes-edit.md#video-edit): Changes an existing video from a prompt: restyle it, add or remove something, change the light. - **Video Upscale**: Upscale a video up to 4K with FLUX or Topaz. - [Extract Frame](https://sloosh.ai/docs/nodes-edit.md#extract-frame): Pulls one frame out of a video as a JPEG. - [Trim video](https://sloosh.ai/docs/nodes-edit.md#trim-video): Keeps the part of a video between Start and End, in seconds. - [Video Concat](https://sloosh.ai/docs/nodes-edit.md#video-concat): Joins clips end to end, in the order of its Clip ports. - [Veed Subtitles](https://sloosh.ai/docs/nodes-edit.md#veed-subtitles): Transcribes the speech and burns styled subtitles into the video. - [Text to speech](https://sloosh.ai/docs/nodes-generate.md): Reads a script aloud in a stock ElevenLabs voice. - [Lipsync](https://sloosh.ai/docs/nodes-generate.md): Makes the person in a video say an audio track, with Sync Lipsync 2. - [Lyria 3](https://sloosh.ai/docs/nodes-generate.md): Makes music from a prompt. - [Audio Input](https://sloosh.ai/docs/nodes-input.md#audio-input): One audio file. - [Merge audio + video](https://sloosh.ai/docs/nodes-edit.md#merge-audio-video): Puts an audio track on a video. - [Download](https://sloosh.ai/docs/nodes-tools.md#download): Wire images and videos into it and it gives you a download button on the canvas. - [PDF Export](https://sloosh.ai/docs/nodes-tools.md#pdf-export): Puts images into one PDF, one page per image: a lookbook, a catalogue, a contact sheet. - [Video Analyzer](https://sloosh.ai/docs/nodes-tools.md#video-analyzer): Watches a video and answers a prompt about it. - [Garment Classifier](https://sloosh.ai/docs/nodes-tools.md#garment-classifier): Sorts images into groups you describe. - **If / Else**: Send the flow one way or another depending on a condition. - [Text Iterator](https://sloosh.ai/docs/nodes-tools.md#text-iterator): A list of text values. - [Image Randomizer](https://sloosh.ai/docs/nodes-tools.md#image-randomizer): Picks one image at random from the ones wired in, each run. - [Bundle](https://sloosh.ai/docs/nodes-tools.md#bundle): Collects the results of an iterator back into one set, which you can wire into a port that takes several images. - [Image Iterator](https://sloosh.ai/docs/nodes-tools.md#image-iterator): The same for images: everything after it runs once per image. - **Sticky Note**: Add a free-form note to your workflow.