Cloudflare Workers AI is a very powerful set of APIs, which allows anyone to run machine learning models using serverless GPUs. They have a generous free tier, with which you can rapidly add AI capabilities to your apps.

This is what we will be building:

We will use these APIs through the cloudflare_ai package (co-developed by me 🙂), and generate images according to the prompt the user enters. To start out, create a new Flutter project. Then install the dependencies

$ flutter pub add cloudflare_ai gal

Next, make a new Dart file presentation/home_page.dart and set it as the home in the MaterialApp widget. In the file, initialise a StatefulWidget. In the widget, define an TextToImageModel. You will need an Account ID and API Key, which you can generate from your Cloudflare Dashboard.

import 'package:cloudflare_ai/cloudflare_ai.dart'

TextToImageModel model = TextToImageModel(
    accountId: "Your Account ID",
    apiKey: "Your API Key"
    model: TextToImageModels.DREAMSHAPER_8_LCM,
);

Define 2 other variables,

Uint8List? image; // Image bytes
bool isLoading = false;

Next, make a User Interface, where the user will enter a prompt, and the generated image will be shown.

Scaffold(
  appBar: AppBar(
    title: const Text("Image Generation"),
  ),
  body: Padding(
    padding: const EdgeInsets.all(8.0),
    child: SingleChildScrollView(
      child: Column(
        children: [
          TextField(
            controller: _promptController,
            decoration: InputDecoration(
              label: const Text("Prompt"),
              border: OutlineInputBorder(
                borderRadius: BorderRadius.circular(10),
                borderSide: const BorderSide(
                  color: Colors.white10,
                ),
              ),
            ),
            maxLines: null,
          ),
          ElevatedButton(
            onPressed: () {
              generateImage(_promptController.text);
            },
            child: const Text("Generate Image"),
          ),
          image != null
              ? Image.memory(image!)
              : isLoading
                  ? const Padding(
                      padding: EdgeInsets.all(8.0),
                      child: CircularProgressIndicator(),
                    )
                  : Container(),
          image != null
              ? ElevatedButton(
                  onPressed: () {
                    saveToGallery(image!);
                  },
                  child: const Text("Save to Gallery"),
                )
              : Container(),
        ],
      ),
    ),
  ),
);

Now, create the functions generateImage() and saveToGallery()

import 'package:gal/gal.dart';

// This goes above the build function
void generateImage(String prompt) async {
  if (prompt != "") {
    try {
      setState(() {
        image = null;
        isLoading = true;
      });
      Uint8List imageRes = await model.generateImage(prompt);
      setState(() {
        image = imageRes;
        isLoading = false;
      });
    } catch (e) {
      ScaffoldMessenger.of(context).showSnackBar(
        SnackBar(
          content: Text(e.toString()),
        ),
      );
    }
  }
}

void saveToGallery(Uint8List image) async {
  await Gal.putImageBytes(image);
  ScaffoldMessenger.of(context).showSnackBar(
    const SnackBar(
      content: Text("Image saved to gallery"),
    ),
  );
}

All code in this article is available on this GitHub repository - https://github.com/MananGandhi1810/cloudflare-ai-image-gen

Thank you for reading. Do leave a feedback, and any doubts, I'll try to answer as many as I can.