01
Say what you see
Name the subject, setting and mood: "product photo of a sneaker on a pastel background, soft studio light" beats "nice shoes". Details in, details out.
How to use ModelPixLab.
Three steps to your image — plus tips for better prompts.
No installs, no Photoshop. Everything happens on the model pages: describe what you want, pick a model, generate, download. This page walks through each step and shows how to write prompts that get better results.
Getting started
Three steps to your image
STEP 1
Describe your image
Open a model page and type what you want to see in the prompt box. Plain sentences work: "a cozy cabin in snowy mountains at dusk". Not sure where to start? Tap one of the example chips to fill the box, then edit it.
STEP 2
Pick model & settings
Each page is one model. Choose the image size (square, landscape or portrait) and, where offered, the style or mode. Editing a photo? Switch to Edit mode and upload your picture first.
STEP 3
Generate & download
Press Generate. Most images take seconds. Happy with it? Download up to 4K. Want another take? Adjust one detail in the prompt and generate again — small changes, better odds.
Prompt tips
Write prompts that work
Four habits that make the biggest difference.
02
Add a style word
Finish with the look you want: photo, anime, illustration, 3D render, watercolor. One style word anchors the whole image.
03
Put text in quotes
Words inside the image go in quotation marks: a neon sign that says "OPEN". Short text works best — a few words, spelled exactly.
04
Change one thing at a time
Regenerating with the same prompt gives variations. To fix a result, change one detail per round instead of rewriting everything.
Prompt → result
Read an image like a prompt
Every strong result can be read backwards: subject + place + light + style. When your image misses, add the missing piece instead of starting over.
Subject: Tokyo street · Setting: rainy dusk · Details: neon reflections, transparent umbrellas
Scene + mood + detailsTry this style →
Subject: frosted perfume bottle · Setting: stone pedestal, sage backdrop · Style: minimalist product photo
Product prompts need stagingMake a product shot →
Subject: young explorer · Setting: mountain sunset · Style: stylized 3D animated portrait
Style words steer the whole imageMake a character →
Editing photos
Edit by describing the change
UPLOAD
Add your photo
On the Nano Banana 2 page, switch to Edit mode and upload the photo you want to change.
DESCRIBE
Say what changes
"Make it golden hour", "remove the people in the back", "turn this into a watercolor painting". Faces and details stay consistent.
REFINE
Go round by round
Each edit builds on the last. Fix one thing per message and the result stays under your control.
Which model?
Pick by the job
LiveOpenAI
GPT Image 2.5
Posters, covers and product shots — especially anything with text in the image or step-by-step instructions to follow.
Open generator →LiveGoogle
Nano Banana 2
Editing real photos: background swaps, relighting, outfit changes — while keeping faces consistent.
Open generator →Guide
Not sure yet?
Start on the homepage, look at what each model does, and try the free generations on both pages — same account, same credits.
Browse models →FAQ
Using the lab
Do I need an account to try it?
No. Every model page gives you free generations every day without signing up. An account keeps your credits and history in one place.
How many free generations do I get?
3 free generations per day on the free plan. Paid plans add monthly credits; one-time credit packs never expire.
What size should I choose?
Square for avatars and product shots, landscape for banners and covers, portrait for posters and phone wallpapers. You can always regenerate in another size.
Can I use the images commercially?
Paid plans include a commercial license. Free-plan images are for personal use.
The result is not what I wanted. Now what?
Change one detail in the prompt and generate again — subject, setting, style, in that order. The prompt tips above cover the fixes that work most often.