chiyi/image2api
0
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix(leonardo): 参考图走 uploadInitImage 永久桶；音频参考需搭配图/视频参考

Browse Source
This commit is contained in:
chiyi
2026-08-10 11:14:15 +08:00
parent 52f9ec967c
commit b488142495
2 changed files with 149 additions and 68 deletions
Download Patch File Download Diff File Expand all files Collapse all files
backend/internal/provider/leonardo/video.go
+24 -19
View File
@@ -53,7 +53,7 @@ func (c *Client) GenerateVideo(ctx context.Context, cookie, model, prompt string
if len(img) == 0 {
continue
}
uploadID, upErr := c.uploadAsset(ctx, cookie, assetExtension(img, "png"), img)
uploadID, upErr := c.uploadInitImage(ctx, cookie, assetExtension(img, "png"), img)
if upErr != nil {
return nil, nil, upErr
}
@@ -65,24 +65,6 @@ func (c *Client) GenerateVideo(ctx context.Context, cookie, model, prompt string
if len(imageRefs) > 0 {
guidances["image_reference"] = imageRefs
}
var audioRefs []map[string]any
for _, aud := range refs.Audios {
if len(aud) == 0 {
continue
}
uploadID, upErr := c.uploadAsset(ctx, cookie, assetExtension(aud, "mp3"), aud)
if upErr != nil {
return nil, nil, upErr
}
audio := map[string]any{"id": uploadID, "type": "UPLOADED"}
if secs := MediaDurationSeconds(aud); secs > 0 {
audio["duration"] = secs
}
audioRefs = append(audioRefs, map[string]any{"audio": audio})
}
if len(audioRefs) > 0 {
guidances["audio_reference"] = audioRefs
}
var videoRefs []map[string]any
for _, vid := range refs.Videos {
if len(vid) == 0 {
@@ -101,6 +83,29 @@ func (c *Client) GenerateVideo(ctx context.Context, cookie, model, prompt string
if len(videoRefs) > 0 {
guidances["video_reference_base"] = videoRefs
}
// Leonardo rejects an audio reference that isn't paired with an image or
// video reference (audio_reference_only_compatible_with_image_or_video_reference).
var audioRefs []map[string]any
for _, aud := range refs.Audios {
if len(aud) == 0 {
continue
}
if len(imageRefs) == 0 && len(videoRefs) == 0 {
return nil, nil, errors.New("leonardo: audio reference requires an image or video reference")
}
uploadID, upErr := c.uploadAsset(ctx, cookie, assetExtension(aud, "mp3"), aud)
if upErr != nil {
return nil, nil, upErr
}
audio := map[string]any{"id": uploadID, "type": "UPLOADED"}
if secs := MediaDurationSeconds(aud); secs > 0 {
audio["duration"] = secs
}
audioRefs = append(audioRefs, map[string]any{"audio": audio})
}
if len(audioRefs) > 0 {
guidances["audio_reference"] = audioRefs
}
parameters := map[string]any{
"height": height,