fix(inference): 修正文生图 API 调用参数,支持动态分辨率

- 去掉 response_format 参数(API 默认返回 URL,自动下载)
- Resolution 参数覆盖 API 调用尺寸(不再仅 mock 路径生效)
- parseImageResponse 改为动态 width/height 入参
- 新增 tools/gentest.go 生成测试脚本
- .gitignore 忽略 test_output/
This commit is contained in:
2026-05-25 00:21:16 +08:00
parent 0a9a923d55
commit 6bda5a1f35
3 changed files with 63 additions and 11 deletions
+1
View File
@@ -0,0 +1 @@
test_output/
+13 -11
View File
@@ -36,7 +36,6 @@ type imageGenRequest struct {
Prompt string `json:"prompt"` Prompt string `json:"prompt"`
N int `json:"n,omitempty"` N int `json:"n,omitempty"`
Size string `json:"size,omitempty"` Size string `json:"size,omitempty"`
ResponseFormat string `json:"response_format,omitempty"`
Steps int `json:"steps,omitempty"` Steps int `json:"steps,omitempty"`
CFGScale float64 `json:"cfg_scale,omitempty"` CFGScale float64 `json:"cfg_scale,omitempty"`
} }
@@ -57,7 +56,12 @@ func GenerateImages(ctx context.Context, prompt string, params AssetParams) ([]G
} }
if imgCfg.APIKey != "" { if imgCfg.APIKey != "" {
return callImageGenAPI(ctx, prompt, count) width, height := imgCfg.Width, imgCfg.Height
if params.Resolution > 0 {
width = params.Resolution
height = params.Resolution
}
return callImageGenAPI(ctx, prompt, count, width, height)
} }
size := params.Resolution size := params.Resolution
@@ -69,13 +73,12 @@ func GenerateImages(ctx context.Context, prompt string, params AssetParams) ([]G
} }
// callImageGenAPI 调用 OpenAI 兼容的 Images API,返回生成的图片。 // callImageGenAPI 调用 OpenAI 兼容的 Images API,返回生成的图片。
func callImageGenAPI(ctx context.Context, prompt string, count int) ([]GeneratedImage, error) { func callImageGenAPI(ctx context.Context, prompt string, count, width, height int) ([]GeneratedImage, error) {
reqBody := imageGenRequest{ reqBody := imageGenRequest{
Model: imgCfg.Model, Model: imgCfg.Model,
Prompt: prompt, Prompt: prompt,
N: count, N: count,
Size: fmt.Sprintf("%dx%d", imgCfg.Width, imgCfg.Height), Size: fmt.Sprintf("%dx%d", width, height),
ResponseFormat: "b64_json",
} }
if imgCfg.Steps > 0 { if imgCfg.Steps > 0 {
reqBody.Steps = imgCfg.Steps reqBody.Steps = imgCfg.Steps
@@ -108,7 +111,7 @@ func callImageGenAPI(ctx context.Context, prompt string, count int) ([]Generated
return nil, fmt.Errorf("image gen api error %d: %s", resp.StatusCode, string(b)) return nil, fmt.Errorf("image gen api error %d: %s", resp.StatusCode, string(b))
} }
return parseImageResponse(ctx, resp.Body) return parseImageResponse(ctx, resp.Body, width, height)
} }
// ======================== 图片编辑 API ======================== // ======================== 图片编辑 API ========================
@@ -138,7 +141,6 @@ func callImageEditAPI(ctx context.Context, imageData []byte, prompt string, coun
writer.WriteField("model", imgCfg.Model) writer.WriteField("model", imgCfg.Model)
writer.WriteField("n", strconv.Itoa(count)) writer.WriteField("n", strconv.Itoa(count))
writer.WriteField("size", fmt.Sprintf("%dx%d", imgCfg.Width, imgCfg.Height)) writer.WriteField("size", fmt.Sprintf("%dx%d", imgCfg.Width, imgCfg.Height))
writer.WriteField("response_format", "b64_json")
if err := writer.Close(); err != nil { if err := writer.Close(); err != nil {
return nil, fmt.Errorf("close multipart writer: %w", err) return nil, fmt.Errorf("close multipart writer: %w", err)
@@ -164,11 +166,11 @@ func callImageEditAPI(ctx context.Context, imageData []byte, prompt string, coun
return nil, fmt.Errorf("image edit api error %d: %s", resp.StatusCode, string(b)) return nil, fmt.Errorf("image edit api error %d: %s", resp.StatusCode, string(b))
} }
return parseImageResponse(ctx, resp.Body) return parseImageResponse(ctx, resp.Body, imgCfg.Width, imgCfg.Height)
} }
// parseImageResponse 解析 OpenAI 兼容的图片生成/编辑响应体。 // parseImageResponse 解析 OpenAI 兼容的图片生成/编辑响应体。
func parseImageResponse(ctx context.Context, r io.Reader) ([]GeneratedImage, error) { func parseImageResponse(ctx context.Context, r io.Reader, width, height int) ([]GeneratedImage, error) {
var genResp imageGenResponse var genResp imageGenResponse
if err := json.NewDecoder(r).Decode(&genResp); err != nil { if err := json.NewDecoder(r).Decode(&genResp); err != nil {
return nil, fmt.Errorf("decode response: %w", err) return nil, fmt.Errorf("decode response: %w", err)
@@ -195,8 +197,8 @@ func parseImageResponse(ctx context.Context, r io.Reader) ([]GeneratedImage, err
} }
images = append(images, GeneratedImage{ images = append(images, GeneratedImage{
Data: data, Data: data,
Width: imgCfg.Width, Width: width,
Height: imgCfg.Height, Height: height,
Format: "png", Format: "png",
}) })
} }
+49
View File
@@ -0,0 +1,49 @@
//go:build ignore
package main
import (
"context"
"fmt"
"os"
"gen2d/internal/config"
"gen2d/internal/service"
)
func main() {
cfg := config.Load()
// 清空 API key 强制走 mock
cfg.ImageGen.APIKey = ""
service.InitImageGenConfig(cfg.ImageGen)
service.InitLLMConfig(cfg.LLM)
in := service.PipelineInput{
AssetType: "sprite",
Prompt: "a cute cat warrior with golden armor",
Tags: []string{"pixel art", "fantasy", "16-bit"},
Params: service.AssetParams{
Resolution: 816,
Frames: service.FrameParams{Directions: 4, FramesPerDirection: 1},
Format: "individual",
},
}
output, err := service.RunPipeline(context.Background(), in)
if err != nil {
fmt.Fprintf(os.Stderr, "error: %v\n", err)
os.Exit(1)
}
os.MkdirAll("test_output", 0755)
for i, a := range output.Assets {
fname := fmt.Sprintf("test_output/gen_%d.%s", i, a.Format)
if err := os.WriteFile(fname, a.Data, 0644); err != nil {
fmt.Fprintf(os.Stderr, "write %s: %v\n", fname, err)
os.Exit(1)
}
fmt.Printf("saved %s (%d bytes)\n", fname, len(a.Data))
}
fmt.Printf("metadata: %+v\n", output.Metadata)
}