Nano Banana 2｜Selfie to Portrait
Upload one photo—or just type a sentence—and get a finished 4K image in seconds. One selfie is enough.
Nano Banana 2 (Gemini 3.1 Flash Image) online: text to image, photo edits and multi-image fusion in up to 4K. One selfie makes a studio portrait—try free now.
Features
- Renders in Seconds——The speed of an everyday default model: the image arrives almost as fast as you finish typing. No waiting on inspiration.
- Native 4K Output——Top native resolution up to 4K—product textures and poster headlines stay clean even zoomed in.
- Stays On-Character——A selfie outfit-change keeps your face; fused photos and comic panels keep the same character throughout.
- Accurate In-Image Text——Headlines, dates and sign carvings come out spelled right. Tiny body copy can occasionally blur—retry with new wording.
- Text or Photo Input——No photo? One sentence is enough. Have photos? Upload to edit or fuse—even a single selfie works.
- Invisible AI Watermark——Outputs carry Google's SynthID invisible watermark. Label AI-made work when you publish it.
How it works
- Upload a Photo (Optional) A phone selfie, a product shot, a blank sign—anything goes. No photo? Skip it and just type a sentence.
- Pick a Style, Describe It Tap a preset (portrait, product shot, infographic…), then spell out the scene, the text and the ratio.
- Generate and Download AI renders in seconds: 4K clarity, text laid out. Not happy? Change a line and generate again.
FAQ
- Can I do it with just one selfie?
- Yes. One sharp selfie is enough—Nano Banana 2 keeps faces strongly consistent, so it can change your outfit, background, or age-style while keeping your features. Upload the photo, pick “Edit my photo,” and describe the result. Front-facing, well-lit, unfiltered shots work best.
- Hands and faces keep warping—how to fix?
- Start with a sharp, front-facing photo where the face fills a quarter of the frame, and keep poses simple (hands down, holding a cup). Lock identity in words: “keep the face 100% unchanged.” Do it in two passes—change the scene first, then fix details—and retry with different wording if a hand warps.
- Is it free? Any card required?
- You can try it free right on this page after signing in: new users get trial credits and each image costs a few. No card binding needed to start—generate a couple, and only top up if you want more.
- How is it different from Pro and 2.1?
- One line: 2 is the fast everyday default, Pro is maximum-fidelity research grade, 2.1 is the newest refresh. This page runs 2 (Gemini 3.1 Flash Image)—plenty for daily images. If you need extreme documentary accuracy, reach for Pro; if you chase the newest, watch for 2.1.
- How many reference photos per try?
- Officially, one workflow can hold up to 5 people plus 14 objects consistent. This page's upload box takes one image at a time—for multi-image fusion, send all your images plus the description to the Agent together and it fuses them under the consistency rules.
- Can it do 4K? Which ratios?
- Yes—pick “4K ultra” clarity for native top resolution. Ratios cover 1:1, 16:9, 9:16, 3:4 and 4:3, enough for feeds, covers and posters.
- How to tell if a pic is AI-made?
- Zoom to 100% and check fingers, teeth, earrings and background signage—warped small text and bent straight lines are giveaways. Images from this tool also carry Google's SynthID invisible watermark. For news or money matters, always verify through multiple sources.
- Beginner: what should I type?
- Use the formula: subject + style + scene composition + light and lens + detail locks + what must not change. Example: “same woman, black evening gown, city-night bokeh, cinematic portrait, face 100% unchanged.” Or simply tap a preset below the input box and edit the filled-in text.
Try it free