refactor(inference): 统合 image_gen 为 GPT Image 2 异步 API,删除冗余配置

- 删除 gptimage.go,异步逻辑直接内聚到 inference.go
- ImageGenConfig 改为 GPT Image 2 参数:aspect_ratio / x_channel / poll_max_wait / poll_interval
- GenerateImages 流程:submit → poll → download,支持并行多图(spritesheet)
- 移除旧的 OpenAI 兼容同步 API 代码(callImageGenAPI / parseImageResponse 等)
- 统一 env:GEN2D_IMAGE_* 系列,移除 GEN2D_GPT_IMAGE2_* 冗余
This commit is contained in:
2026-05-25 12:44:39 +08:00
parent d9d3b51262
commit c4dc7394b1
6 changed files with 255 additions and 512 deletions
+7 -7
View File
@@ -21,11 +21,11 @@ llm:
max_tokens: 2048
image_gen:
base_url: "https://api.stability.ai/v1"
base_url: "https://www.yuntts.com/api/v1"
api_key: ""
model: "stable-diffusion-xl"
width: 1024
height: 1024
num_images: 1
steps: 30
cfg_scale: 7.0
model: "gpt-image-2"
quality: "low"
aspect_ratio: "1:1"
x_channel: "default"
poll_max_wait: 120
poll_interval: 3