stable-diffusion-webui项目地址
具体部署教程请去B站寻找或者直接使用整合包
这里直接编写工具类
public class StableDiffusionUtil {private static final String BASE_URL = "http://127.0.0.1:7860";private static final OkHttpClient CLIENT = new OkHttpClient();private static final ObjectMapper OBJECT_MAPPER = new ObjectMapper();/*** 获取可用模型列表** @return 模型列表的 JSON 字符串* @throws IOException 如果请求失败*/public static List<ModelInfo> listModels() throws IOException {Request request = new Request.Builder().url(BASE_URL + "/sdapi/v1/sd-models").get().build();try (Response response = CLIENT.newCall(request).execute()) {if (response.isSuccessful()) {String jsonResponse = response.body().string();// 将 JSON 字符串解析为 ModelInfo 对象列表return OBJECT_MAPPER.readValue(jsonResponse, new TypeReference<List<ModelInfo>>() {});} else {throw new IOException("请求失败: " + response.code());}}}/*** 生成图片并保存** @param prompt 生成图片的描述* @param negativePrompt 负面描述* @param steps 生成步骤* @param cfgScale CFG 参数* @param width 图片宽度* @param height 图片高度* @param samplerIndex 采样器* @param modelCheckpoint 模型名称* @param outputFilePath 保存图片的文件路径* @throws IOException 如果请求失败或保存图片失败*/public static byte[] generateImage(String prompt, String negativePrompt, int steps, int cfgScale,int width, int height, String samplerIndex, String modelCheckpoint) throws IOException {// 请求参数String json = "{"+ "\"prompt\": \"" + prompt + "\","+ "\"negative_prompt\": \"" + negativePrompt + "\","+ "\"steps\": " + steps + ","+ "\"cfg_scale\": " + cfgScale + ","+ "\"width\": " + width + ","+ "\"height\": " + height + ","+ "\"sampler_index\": \"" + samplerIndex + "\","+ "\"sd_model_checkpoint\": \"" + modelCheckpoint + "\""+ "}";RequestBody body = RequestBody.create(json, MediaType.parse("application/json; charset=utf-8"));Request request = new Request.Builder().url(BASE_URL + "/sdapi/v1/txt2img").post(body).build();try (Response response = CLIENT.newCall(request).execute()) {if (response.isSuccessful()) {String jsonResponse = response.body().string();String imageBase64 = jsonResponse.split("\"images\":\\[\"")[1].split("\"")[0];// 解码 Base64 图片byte[] imageBytes = Base64.getDecoder().decode(imageBase64);ByteArrayInputStream bis = new ByteArrayInputStream(imageBytes);BufferedImage image = ImageIO.read(bis);// 将图片转换为二进制数组ByteArrayOutputStream baos = new ByteArrayOutputStream();ImageIO.write(image, "png", baos);return baos.toByteArray();} else {throw new IOException("请求失败: " + response.code());}}}/*** 切换模型** @param modelCheckpoint 模型名称* @throws IOException 如果请求失败*/public static void switchModel(String modelCheckpoint) throws IOException {// 更新模型配置String json = "{"+ "\"sd_model_checkpoint\": \"" + modelCheckpoint + "\""+ "}";RequestBody body = RequestBody.create(json, MediaType.parse("application/json; charset=utf-8"));Request request = new Request.Builder().url(BASE_URL + "/sdapi/v1/options").post(body).build();try (Response response = CLIENT.newCall(request).execute()) {if (response.isSuccessful()) {System.out.println("模型已切换到 " + modelCheckpoint);} else {throw new IOException("请求失败: " + response.code());}}}}
调用工具类实现
@Autowiredprivate SysOssService ossService;@Testvoid test8() throws Exception {String prompt = "A futuristic city with flying cars, highly detailed, 4k";String negativePrompt = "blurry, low quality, distorted, text";int steps = 30;int cfgScale = 7;int width = 512;int height = 512;String samplerIndex = "Euler a";String outputFilePath = "output.png";String modelInfo="anything-v5-PrtRE.safetensors [7f96a1a9ca]";// 调用方法生成图片byte[] imageBytes= StableDiffusionUtil.generateImage(prompt, negativePrompt, steps, cfgScale, width, height, samplerIndex, modelInfo);String fileName = "generated_image.png"; // 文件名String contentType = "image/png"; // 文件类型SysOssDTO ossDTO = ossService.upload(imageBytes, fileName, contentType);// 打印上传结果System.out.println("文件上传成功，访问路径: " + ossDTO.getUrl());}