feat: move keyframes beside transcript timeline

This commit is contained in:
2026-05-18 15:09:49 +08:00
parent 4991526bcc
commit c22bee4878
2 changed files with 222 additions and 162 deletions

File diff suppressed because one or more lines are too long

View File

@@ -1697,6 +1697,8 @@ function AudioIntakePanel({
const [audioFeatures, setAudioFeatures] = useState<AudioFeature[]>([])
const [audioFeatureStatus, setAudioFeatureStatus] = useState<AudioFeatureStatus>("idle")
const [manualBusy, setManualBusy] = useState(false)
const [extracting, setExtracting] = useState(false)
const [deletingFrame, setDeletingFrame] = useState<number | null>(null)
const [waveHoverTime, setWaveHoverTime] = useState<number | null>(null)
const videoRef = useRef<HTMLVideoElement | null>(null)
const rowRefs = useRef<Record<number, HTMLDivElement | null>>({})
@@ -1720,6 +1722,11 @@ function AudioIntakePanel({
)
}, [job, mediaDuration])
const activeSegment = job?.transcript.find((segment) => currentTime >= segment.start && currentTime <= Math.max(segment.end, segment.start + 0.2))
const frames = useMemo(() => job ? [...job.frames].sort((a, b) => a.timestamp - b.timestamp) : [], [job])
const selectedReferenceFrames = useMemo(
() => frames.filter((frame) => selectedFrames.has(frame.index)),
[frames, selectedFrames],
)
const waveTimeHint = waveHoverTime !== null
? `指针停点 ${waveHoverTime.toFixed(1)}s`
: activeSegment
@@ -1802,6 +1809,33 @@ function AudioIntakePanel({
}
}
const extractKeyframes = async () => {
if (!job) return
setExtracting(true)
try {
for (const frame of job.frames) {
if (selectedFrames.has(frame.index)) onToggleFrame(frame.index)
}
const updated = await analyzeJob(job.id, 12, "motion", "replace", "accurate")
onJobUpdate(updated)
toast.info("已按动作峰值逻辑重新抽取 12 张参考帧,完成后在时间轴左侧选择主角参考。")
} catch (e) {
toast.error("12 张关键帧抽取失败:" + (e instanceof Error ? e.message : String(e)))
} finally {
setExtracting(false)
}
}
const deleteReferenceFrame = async (idx: number) => {
if (!job || !onDeleteFrame) return
setDeletingFrame(idx)
try {
await onDeleteFrame(job.id, idx)
} finally {
setDeletingFrame(null)
}
}
if (!job) {
return <EmptyState text="先在左侧粘贴 TK 链接或上传本地视频。点击开始后,会先下载视频,再自动解析原音频文案、讲话人节奏和背景音。" />
}
@@ -1900,44 +1934,58 @@ function AudioIntakePanel({
/>
</div>
<div className="min-w-0">
<div className="mb-2 flex items-center justify-between gap-3">
<SectionTitle icon={<FileText className="h-4 w-4" />} title="逐句时间轴" />
<span className="rounded-md border border-white/10 bg-black/35 px-2 py-1 text-[11px] text-white/45">{job.transcript.length} </span>
</div>
{job.transcript.length ? (
<div className="overflow-hidden rounded-md border border-white/10">
<div className="grid grid-cols-[76px_minmax(0,1fr)] border-b border-white/10 bg-white/[0.04] px-3 py-1.5 text-[11px] font-semibold text-white/50">
<div></div>
<div> / </div>
</div>
<div className="max-h-[288px] overflow-y-auto 2xl:max-h-[346px]">
{job.transcript.map((segment) => {
const active = activeSegment?.index === segment.index
return (
<div
key={segment.index}
ref={(node) => { rowRefs.current[segment.index] = node }}
onClick={() => seekTo(segment.start)}
className={`grid cursor-pointer grid-cols-[76px_minmax(0,1fr)] gap-2 border-b px-3 py-1.5 text-[11.5px] leading-snug transition last:border-b-0 ${
active
? "border-emerald-300/18 bg-emerald-300/[0.12] text-white"
: "border-white/8 text-white/64 hover:bg-white/[0.045]"
}`}
>
<div className={`font-mono text-[10.5px] ${active ? "text-emerald-100" : "text-white/38"}`}>{segment.start.toFixed(1)}-{segment.end.toFixed(1)}s</div>
<div className="min-w-0">
<div className="truncate" title={segment.en}>{segment.en || <span className="text-white/30">-</span>}</div>
<div className={`mt-0.5 truncate text-[11px] ${active ? "text-emerald-50/80" : "text-white/42"}`} title={segment.zh}>{segment.zh || <span className="text-white/30"></span>}</div>
</div>
</div>
)
})}
</div>
<div className="grid gap-2 xl:grid-cols-[360px_minmax(0,1fr)] 2xl:grid-cols-[400px_minmax(0,1fr)]">
<SourceKeyframePicker
job={job}
frames={frames}
selectedFrames={selectedFrames}
selectedReferenceFrames={selectedReferenceFrames}
extracting={extracting}
deletingFrame={deletingFrame}
onToggleFrame={onToggleFrame}
onExtract={() => void extractKeyframes()}
onDeleteFrame={onDeleteFrame ? (idx) => void deleteReferenceFrame(idx) : undefined}
/>
<div className="min-w-0">
<div className="mb-2 flex items-center justify-between gap-3">
<SectionTitle icon={<FileText className="h-4 w-4" />} title="逐句时间轴" />
<span className="rounded-md border border-white/10 bg-black/35 px-2 py-1 text-[11px] text-white/45">{job.transcript.length} </span>
</div>
) : (
<EmptyState text={processing ? "音频解析中,完成后这里会按时间列出原文案和中文翻译。" : "下载完成后会自动解析音频;也可以点击右上角“解析音频”手动重试。"} />
)}
{job.transcript.length ? (
<div className="overflow-hidden rounded-md border border-white/10">
<div className="grid grid-cols-[76px_minmax(0,1fr)] border-b border-white/10 bg-white/[0.04] px-3 py-1.5 text-[11px] font-semibold text-white/50">
<div></div>
<div> / </div>
</div>
<div className="max-h-[288px] overflow-y-auto 2xl:max-h-[346px]">
{job.transcript.map((segment) => {
const active = activeSegment?.index === segment.index
return (
<div
key={segment.index}
ref={(node) => { rowRefs.current[segment.index] = node }}
onClick={() => seekTo(segment.start)}
className={`grid cursor-pointer grid-cols-[76px_minmax(0,1fr)] gap-2 border-b px-3 py-1.5 text-[11.5px] leading-snug transition last:border-b-0 ${
active
? "border-emerald-300/18 bg-emerald-300/[0.12] text-white"
: "border-white/8 text-white/64 hover:bg-white/[0.045]"
}`}
>
<div className={`font-mono text-[10.5px] ${active ? "text-emerald-100" : "text-white/38"}`}>{segment.start.toFixed(1)}-{segment.end.toFixed(1)}s</div>
<div className="min-w-0">
<div className="truncate" title={segment.en}>{segment.en || <span className="text-white/30">-</span>}</div>
<div className={`mt-0.5 truncate text-[11px] ${active ? "text-emerald-50/80" : "text-white/42"}`} title={segment.zh}>{segment.zh || <span className="text-white/30"></span>}</div>
</div>
</div>
)
})}
</div>
</div>
) : (
<EmptyState text={processing ? "音频解析中,完成后这里会按时间列出原文案和中文翻译。" : "下载完成后会自动解析音频;也可以点击右上角“解析音频”手动重试。"} />
)}
</div>
</div>
</div>
</div>
@@ -1945,9 +1993,7 @@ function AudioIntakePanel({
<SourceReferenceBuildPanel
job={job}
selectedFrames={selectedFrames}
onToggleFrame={onToggleFrame}
onJobUpdate={onJobUpdate}
onDeleteFrame={onDeleteFrame}
runtimeModels={runtimeModels}
/>
</div>
@@ -1956,25 +2002,137 @@ function AudioIntakePanel({
)
}
function SourceKeyframePicker({
job,
frames,
selectedFrames,
selectedReferenceFrames,
extracting,
deletingFrame,
onToggleFrame,
onExtract,
onDeleteFrame,
}: {
job: Job
frames: KeyFrame[]
selectedFrames: Set<number>
selectedReferenceFrames: KeyFrame[]
extracting: boolean
deletingFrame: number | null
onToggleFrame: (idx: number) => void
onExtract: () => void
onDeleteFrame?: (idx: number) => void
}) {
return (
<div className="min-w-0">
<div className="mb-2 flex items-center justify-between gap-2">
<SectionTitle icon={<ImageIcon className="h-4 w-4" />} title="关键帧" />
<div className="flex items-center gap-1.5">
<span className="rounded-md border border-white/10 bg-black/35 px-2 py-1 text-[10px] text-white/42">
{selectedReferenceFrames.length || "全部"}
</span>
<button
type="button"
onClick={onExtract}
disabled={!job.video_url || extracting || job.status === "splitting"}
title="自动按动作峰值抽 12 张参考帧,更偏向手势、表情变化、节奏点和镜头变化"
className="inline-flex h-7 items-center justify-center gap-1 rounded-md bg-white px-2 text-[10.5px] font-semibold text-black transition hover:bg-white/90 disabled:cursor-not-allowed disabled:opacity-40"
>
{extracting || job.status === "splitting" ? <Loader2 className="h-3.5 w-3.5 animate-spin" /> : <Scissors className="h-3.5 w-3.5" />}
12
</button>
</div>
</div>
<div className="grid min-h-[205px] gap-1.5 rounded-md border border-white/10 bg-black/32 p-1.5 xl:grid-cols-[minmax(0,1fr)_86px] 2xl:min-h-[260px] 2xl:grid-cols-[minmax(0,1fr)_96px]">
<div className="min-w-0">
<div className="mb-1 flex items-center justify-between gap-2">
<span className="text-[10px] text-white/34"></span>
<span className="text-[9.5px] text-white/28"></span>
</div>
<div className="grid max-h-[178px] grid-cols-[repeat(auto-fill,minmax(38px,1fr))] gap-1 overflow-y-auto pr-0.5 2xl:max-h-[232px]">
{frames.map((frame, index) => {
const selected = selectedFrames.has(frame.index)
return (
<MediaAssetTile
key={frame.index}
src={effectiveFrameUrl(job.id, frame)}
alt={`关键帧 ${index + 1}`}
label={`参考帧 ${String(index + 1).padStart(2, "0")}`}
meta={`${frame.timestamp.toFixed(1)}s`}
className="aspect-[9/16]"
objectFit="contain"
selected={selected}
title={`关键帧 ${index + 1} · ${frame.timestamp.toFixed(1)}s`}
onClick={() => onToggleFrame(frame.index)}
topLeft={<span className="rounded bg-black/72 px-1 font-mono text-[9px] text-white/70">{String(index + 1).padStart(2, "0")}</span>}
topRight={<span className="rounded-full bg-black/72 p-0.5">{selected ? <Check className="h-3 w-3 text-emerald-200" /> : <Circle className="h-3 w-3 text-white/50" />}</span>}
onDelete={onDeleteFrame ? () => onDeleteFrame(frame.index) : undefined}
deleting={deletingFrame === frame.index}
deleteLabel={`删除关键帧 ${index + 1}`}
/>
)
})}
{!frames.length && (
<div className="col-span-full flex h-[106px] items-center justify-center rounded border border-dashed border-white/12 px-2 text-center text-[10.5px] leading-snug text-white/34">
</div>
)}
</div>
</div>
<aside className="min-w-0 rounded-md border border-emerald-300/12 bg-emerald-300/[0.035] p-1">
<div className="mb-1 flex items-center justify-between gap-1">
<span className="text-[9.5px] font-semibold text-emerald-50/70"></span>
<span className="font-mono text-[9px] text-emerald-50/55">{selectedReferenceFrames.length}</span>
</div>
{selectedReferenceFrames.length ? (
<div className="grid max-h-[168px] grid-cols-2 gap-1 overflow-y-auto pr-0.5 2xl:max-h-[222px]">
{selectedReferenceFrames.map((frame) => {
const order = frames.findIndex((item) => item.index === frame.index)
const label = order >= 0 ? String(order + 1).padStart(2, "0") : String(frame.index)
return (
<MediaAssetTile
key={frame.index}
src={effectiveFrameUrl(job.id, frame)}
alt={`已选关键帧 ${label}`}
label={`已选 ${label}`}
meta={`${frame.timestamp.toFixed(1)}s`}
className="aspect-[9/16]"
objectFit="contain"
selected
title={`点击取消选择 · ${frame.timestamp.toFixed(1)}s`}
onClick={() => onToggleFrame(frame.index)}
topLeft={<span className="rounded bg-black/72 px-1 font-mono text-[8.5px] text-emerald-100/80">{label}</span>}
topRight={<span className="rounded-full bg-black/72 p-0.5"><Check className="h-2.5 w-2.5 text-emerald-200" /></span>}
/>
)
})}
</div>
) : (
<div className="flex h-[90px] items-center justify-center rounded border border-dashed border-emerald-300/12 px-1 text-center text-[9.5px] leading-snug text-white/30">
使
</div>
)}
</aside>
</div>
</div>
)
}
function SourceReferenceBuildPanel({
job,
selectedFrames,
onToggleFrame,
onJobUpdate,
onDeleteFrame,
runtimeModels,
}: {
job: Job
selectedFrames: Set<number>
onToggleFrame: (idx: number) => void
onJobUpdate: (job: Job) => void
onDeleteFrame?: (jobId: string, idx: number) => Promise<void> | void
runtimeModels?: RuntimeModels
}) {
const [extracting, setExtracting] = useState(false)
const [subjectBusy, setSubjectBusy] = useState(false)
const [subjectAssetBusy, setSubjectAssetBusy] = useState<string | null>(null)
const [deletingFrame, setDeletingFrame] = useState<number | null>(null)
const [subjectStyle, setSubjectStyle] = useState<SubjectStyleMode>("transparent_human")
const [subjectDirection, setSubjectDirection] = useState("")
const [characterLibrary, setCharacterLibrary] = useState<CharacterLibraryItem[]>([])
@@ -2011,9 +2169,9 @@ function SourceReferenceBuildPanel({
})
}, [actorAssets])
const referenceCountLabel = selectedReferenceFrames.length
? `使用已选 ${selectedReferenceFrames.length}`
? `已选 ${selectedReferenceFrames.length}参考帧`
: frames.length
? `默认使用全部 ${frames.length}`
? `默认使用全部 ${frames.length}参考帧`
: "待抽帧"
useEffect(() => {
@@ -2028,22 +2186,6 @@ function SourceReferenceBuildPanel({
return () => { cancelled = true }
}, [])
const extractKeyframes = async () => {
setExtracting(true)
try {
for (const frame of job.frames) {
if (selectedFrames.has(frame.index)) onToggleFrame(frame.index)
}
const updated = await analyzeJob(job.id, 12, "motion", "replace", "accurate")
onJobUpdate(updated)
toast.info("已按动作峰值逻辑重新抽取 12 张参考帧,完成后在这里人工选择主角参考。")
} catch (e) {
toast.error("12 张关键帧抽取失败:" + (e instanceof Error ? e.message : String(e)))
} finally {
setExtracting(false)
}
}
const generateSimilarActor = async () => {
if (!frames.length) {
toast.warning("请先自动抽帧 12 张,或在原版视频上手动补帧。")
@@ -2095,16 +2237,6 @@ function SourceReferenceBuildPanel({
}
}
const deleteReferenceFrame = async (idx: number) => {
if (!onDeleteFrame) return
setDeletingFrame(idx)
try {
await onDeleteFrame(job.id, idx)
} finally {
setDeletingFrame(null)
}
}
const regenerateSubjectAsset = async (asset: SubjectAsset) => {
if (!actorSource) return
setSubjectAssetBusy(`regen:${asset.id}`)
@@ -2150,99 +2282,15 @@ function SourceReferenceBuildPanel({
return (
<div className="min-w-0">
<div className="mb-2 flex items-center justify-between gap-3">
<SectionTitle icon={<ImageIcon className="h-4 w-4" />} title="关键帧 / 相似主体" />
<SectionTitle icon={<Sparkles className="h-4 w-4" />} title="相似主体 / 主体模板" />
<div className="flex items-center gap-2">
<span className="rounded-md border border-white/10 bg-black/35 px-2 py-1 text-[11px] text-white/45">
{referenceCountLabel}
</span>
<button
type="button"
onClick={() => void extractKeyframes()}
disabled={!job.video_url || extracting || job.status === "splitting"}
title="自动按动作峰值抽 12 张参考帧,更偏向手势、表情变化、节奏点和镜头变化"
className="inline-flex h-8 items-center justify-center gap-1 rounded-md bg-white px-3 text-[11px] font-semibold text-black transition hover:bg-white/90 disabled:cursor-not-allowed disabled:opacity-40"
>
{extracting || job.status === "splitting" ? <Loader2 className="h-3.5 w-3.5 animate-spin" /> : <Scissors className="h-3.5 w-3.5" />}
12
</button>
</div>
</div>
<div className="h-[250px] overflow-y-auto rounded-md border border-white/10 bg-black/32 p-2 2xl:h-[290px]">
<div className="grid gap-2 xl:grid-cols-[minmax(0,1fr)_138px] 2xl:grid-cols-[minmax(0,1fr)_158px]">
<div className="min-w-0">
<div className="flex items-center justify-between gap-2">
<span className="text-[10.5px] text-white/34"></span>
<span className="text-[10.5px] text-white/30"></span>
</div>
<div className="mt-2 grid grid-cols-[repeat(auto-fill,minmax(38px,1fr))] gap-1">
{frames.map((frame, index) => {
const selected = selectedFrames.has(frame.index)
return (
<MediaAssetTile
key={frame.index}
src={effectiveFrameUrl(job.id, frame)}
alt={`关键帧 ${index + 1}`}
label={`参考帧 ${String(index + 1).padStart(2, "0")}`}
meta={`${frame.timestamp.toFixed(1)}s`}
className="aspect-[9/16]"
objectFit="contain"
selected={selected}
title={`关键帧 ${index + 1} · ${frame.timestamp.toFixed(1)}s`}
onClick={() => onToggleFrame(frame.index)}
topLeft={<span className="rounded bg-black/72 px-1 font-mono text-[9px] text-white/70">{String(index + 1).padStart(2, "0")}</span>}
topRight={<span className="rounded-full bg-black/72 p-0.5">{selected ? <Check className="h-3 w-3 text-emerald-200" /> : <Circle className="h-3 w-3 text-white/50" />}</span>}
onDelete={onDeleteFrame ? () => void deleteReferenceFrame(frame.index) : undefined}
deleting={deletingFrame === frame.index}
deleteLabel={`删除关键帧 ${index + 1}`}
/>
)
})}
{!frames.length && (
<div className="col-span-full flex h-[106px] items-center justify-center rounded border border-dashed border-white/12 text-[11px] text-white/34">
12
</div>
)}
</div>
</div>
<aside className="min-w-0 rounded-md border border-emerald-300/12 bg-emerald-300/[0.035] p-1.5">
<div className="mb-1 flex items-center justify-between gap-1">
<span className="text-[10px] font-semibold text-emerald-50/70"></span>
<span className="rounded border border-emerald-300/15 bg-black/24 px-1.5 py-0.5 font-mono text-[9px] text-emerald-50/55">{selectedReferenceFrames.length}</span>
</div>
{selectedReferenceFrames.length ? (
<div className="grid max-h-[176px] grid-cols-3 gap-1 overflow-y-auto pr-0.5 2xl:max-h-[210px]">
{selectedReferenceFrames.map((frame) => {
const order = frames.findIndex((item) => item.index === frame.index)
const label = order >= 0 ? String(order + 1).padStart(2, "0") : String(frame.index)
return (
<MediaAssetTile
key={frame.index}
src={effectiveFrameUrl(job.id, frame)}
alt={`已选关键帧 ${label}`}
label={`已选 ${label}`}
meta={`${frame.timestamp.toFixed(1)}s`}
className="aspect-[9/16]"
objectFit="contain"
selected
title={`点击取消选择 · ${frame.timestamp.toFixed(1)}s`}
onClick={() => onToggleFrame(frame.index)}
topLeft={<span className="rounded bg-black/72 px-1 font-mono text-[8.5px] text-emerald-100/80">{label}</span>}
topRight={<span className="rounded-full bg-black/72 p-0.5"><Check className="h-2.5 w-2.5 text-emerald-200" /></span>}
/>
)
})}
</div>
) : (
<div className="flex h-[106px] items-center justify-center rounded border border-dashed border-emerald-300/12 px-2 text-center text-[10px] leading-snug text-white/30">
</div>
)}
</aside>
</div>
<div className="mt-2 border-t border-white/8 pt-2">
<div className="rounded-md border border-white/10 bg-black/32 p-2">
<div>
<div className="mb-1.5 flex flex-wrap items-center justify-between gap-2 text-[10px] text-white/36">
<div className="flex items-center gap-2">
<span></span>