Back to Home11/18/2025, 7:39:21 AM

Show HN: I built a Nano Banana 2 (Gempix2) playground for 4K AI images

1 points
0 comments

Mood

calm

Sentiment

positive

Category

tech

Key topics

AI image generation

productivity tools

web development

I’ve been building a small playground on top of Nano Banana 2 / Gempix2 and would love feedback from HN.

Quick context: “Nano Banana 2” is the public name of a next‑gen image model; Gempix2 / GEMPIX2 is the internal codename on the Google side. I don’t run the model myself – the app sits on top of the fal-ai/gempix2 endpoint and tries to make the model actually usable for day‑to‑day work.

What it is

- A web UI and workflow layer for Nano Banana 2 / Gempix2: https://www.gempix2.site - Focused on 4K images, 10‑image fusion, and non‑English (especially CJK)text inside the image - Comes with a small prompt library (400+ prompts) for portraits, product shots, infographics and CJK use‑cases: https://www.gempix2.site/prompts

Why it might be interesting

From my tests and research, this model family is good at:

- Native 2K generation with true 4K export (useful for covers, posters,large screens) - Up to 10‑image fusion for collages, product walls, comparison shots, or simple storyboards - Much better text rendering for Chinese / Japanese / Korean, plus mixed‑language prompts - Strong prompt following for practical stuff like “blog cover”, “ecommerce grid”, “simple infographic”

One early case: a 3‑person ecommerce team went from ~8 hours per product (shoot + retouch + layout) to about 45 minutes by moving most of their visuals into Nano Banana 2 workflows: 10‑image fusion for multi‑angle product walls, direct “text‑on” promo posters in Chinese, and simple data infographics.

How it’s built

- Frontend: Next.js + React + TypeScript + Tailwind - Backend: Next.js API routes, Postgres + Drizzle ORM for tasks and credit accounting - Image generation: @fal-ai/client talking to fal-ai/gempix2 (queue API, webhooks, status endpoints) - Extras: Stripe for payments, some basic credit/refund logic around long‑running jobs

What I’m unsure about

- Are the workflows (covers, infographics, product images) actually useful to people who ship things, or am I over‑fitting to my own use cases? - For those who work with Midjourney / DALL·E 3 daily: what controls do you miss here (seed control, finer composition tools, etc.)? - If you regularly need CJK text in images, what would a “dream” Nano Banana 2 / Gempix2 UI look like for you?

Links again:

- App: https://www.gempix2.site - Prompt library: https://www.gempix2.site/prompts

Happy to answer technical / product questions here, and I’d really appreciate blunt feedback on what feels useful vs. unnecessary.

The author built a web UI and workflow layer on top of the Nano Banana 2/Gempix2 AI image model, making it more usable for day-to-day work, and is seeking feedback from the HN community.

Snapshot generated from the HN discussion

Discussion Activity

Light discussion

First comment

4m

Peak period

1

Hour 1

Avg / period

1

Comment distribution1 data points

Based on 1 loaded comments

Key moments

  1. 01Story posted

    11/18/2025, 7:39:21 AM

    14h ago

    Step 01
  2. 02First comment

    11/18/2025, 7:43:28 AM

    4m after posting

    Step 02
  3. 03Peak activity

    1 comments in Hour 1

    Hottest window of the conversation

    Step 03
  4. 04Latest activity

    11/18/2025, 7:43:28 AM

    14h ago

    Step 04

Generating AI Summary...

Analyzing up to 500 comments to identify key contributors and discussion patterns

Discussion (0 comments)

Discussion hasn't started yet.

ID: 45962390Type: storyLast synced: 11/18/2025, 7:41:40 AM

Want the full context?

Jump to the original sources

Read the primary article or dive into the live Hacker News thread when you're ready.