Describe a picture.
Watch it appear.
A free studio for posters, photos, illustrations, anime and icons. Words on a poster come out spelled right.
Open the studioThis is the real render, step by step.
Your scroll is the sampler.
Six rooms.
One for each kind of picture.
Every tab picks the open model that is best at it, shows only the settings that matter, and has its own guide and AI helper.
Each print opens its tab with the same prompt loaded. Change a word and make your own.
Open the studioDraw a box.
Change one. Keep the rest.
In Layout, every text block and object gets a box that says what goes there. Keep the seed, change one box, and the picture changes only where you asked.
"label": "object", "caption": "A worn red fabric couch with two cushions", "bbox": [41, -420, 389, 123]
Same seed (1000), same layout, one caption edited. Both pictures are real renders.
How it goes.
- Pick the kind. Text & Posters, Layout, Photo, Illustration, Anime or Icons. The tab sets the model.
- Describe it. A sentence is enough. Improve with AI rewrites it the way that model reads best. In Layout, AI plans the boxes as JSON.
- Make it, then remix. Watch the real step count. Keep the seed and change one detail, upscale, cut out the background, or share a link.
- First picture
- No account needed.
- Free account
- 20 credits a day once your email is confirmed.
- Turbo poster or fast photo
- 1 credit, about a minute and a half.
- Best-quality modes
- 3 or 4 credits, up to about 10 minutes.
- Where it runs
- On one GPU in our own rack, shared with other apps. You see your real place in line.
- Paid plans
- None.
Questions
Is it free?
Yes. Your first picture needs no account. A free account gets 20 credits a day once you confirm your email. A Turbo or fast picture costs 1 credit, the slow best-quality modes cost 3 or 4. There are no paid plans.
Which models does it use?
MOSS-Image (Apache 2.0) for Text & Posters, Layout and Anime, FLUX.1 for photos, DreamShaper 8 for illustrations, and a game-icon SDXL model with background removal for icons. Every tab says which model it runs and why.
Why does the text come out right?
MOSS-Image reads your prompt with a 9-billion-parameter language model and was trained to render text. Put the exact words in quotes, or draw a box for them in the Layout tab.
What is a JSON layout?
A description of the picture plus a box for each text block, object or character, saying what goes where. You draw the boxes (or let AI plan them) and the model places things there. You can view and paste the JSON too.
How long does a picture take?
Turbo MOSS pictures and FLUX photos take about a minute and a half, illustrations about 25 seconds. Standard MOSS quality takes around 10 minutes. Everything runs on one shared GPU, and the studio shows your real place in line.
Can I use the pictures commercially?
MOSS-Image, FLUX schnell and DreamShaper are released under licences that allow commercial use. FLUX dev (the Max detail option in Photo) is licensed for non-commercial use only. Check that the subject itself is yours to use.
Can it edit my own photo?
Not yet. MOSS-Image only makes new pictures from text. The closest thing is Remix: keep the seed, change one detail or one layout box, and most of the picture stays the same.
What happens to my pictures?
Pictures from an account stay in your library until you delete them, and you can download everything as a ZIP. Pictures made without an account are deleted after 7 days. Share links are off unless you switch them on.