Stop resampling the extend start frame every hop

upload() ran fitStill (ffmpeg scale+pad) on every image except hero,
including the start/handoff still. Four chained extends resampled the
anchor frame four times and destroyed identity even with the hero
reference and lock text intact.

- upload(): only fit when the still's aspect truly differs from the
  job canvas (needsFit); never fit hero or start regardless.
- Track handoffSource (anchor|extract) and whether the start frame was
  fitted; surface both in job metadata and Details.
- extendSelected(): re-binding an already-selected clip on task switch
  must not stomp the user's overridden locks/settings back to the
  clip's saved values (regression introduced by the last hotfix push).
- Add tests/studio2-extend-fit.test.mjs: two extends at identical
  width/height reach Comfy with the anchor PNG byte-identical, no
  ffmpeg fit.

Co-authored-by: Cursor <cursoragent@cursor.com>
This commit is contained in:
Towsty
2026-09-11 21:12:24 -05:00
co-authored by Cursor
parent c6451b10ce
commit 65c69b779a
5 changed files with 144 additions and 4 deletions
+1
View File
@@ -17,6 +17,7 @@ const rows=computed(()=>{const m=props.metadata||{},s=m.settings||{};const time=
{label:'Identity still',value:m.identityStillId||'None'},{label:'Face lock',value:m.lockFace!==false?'On':'Off'},{label:'Outfit lock',value:m.lockOutfit!==false?'On':'Off'}, {label:'Identity still',value:m.identityStillId||'None'},{label:'Face lock',value:m.lockFace!==false?'On':'Off'},{label:'Outfit lock',value:m.lockOutfit!==false?'On':'Off'},
{label:'Aspect',value:s.aspect||'auto'},{label:'FPS',value:s.fps||24},{label:'Hide input',value:m.hideInput?'On':'Off'}, {label:'Aspect',value:s.aspect||'auto'},{label:'FPS',value:s.fps||24},{label:'Hide input',value:m.hideInput?'On':'Off'},
{label:'Start frame',value:m.startFrameSource?JSON.stringify(m.startFrameSource):'None'}, {label:'Start frame',value:m.startFrameSource?JSON.stringify(m.startFrameSource):'None'},
{label:'Handoff source',value:m.handoffSource || 'None'},{label:'Handoff fitted (resampled)',value:m.fitted==null?'Not recorded':m.fitted?'Yes':'No'},
{label:'End still',value:m.endStillId||'None'},{label:'Guide stills',value:m.guides?.length?JSON.stringify(m.guides):'None'}, {label:'End still',value:m.endStillId||'None'},{label:'Guide stills',value:m.guides?.length?JSON.stringify(m.guides):'None'},
{label:'Compiled prompt',value:m.compiledPrompt},{label:'Queued',value:time(m.queuedAt)},{label:'Started',value:time(m.startedAt)},{label:'Saved',value:time(m.savedAt)}, {label:'Compiled prompt',value:m.compiledPrompt},{label:'Queued',value:time(m.queuedAt)},{label:'Started',value:time(m.startedAt)},{label:'Saved',value:time(m.savedAt)},
{label:'GPU seconds',value:m.gpuSeconds},{label:'Wall seconds',value:m.wallTime},{label:'Comfy prompt ID',value:m.promptId},{label:'Host cleanup',value:m.purgeResult} {label:'GPU seconds',value:m.gpuSeconds},{label:'Wall seconds',value:m.wallTime},{label:'Comfy prompt ID',value:m.promptId},{label:'Host cleanup',value:m.purgeResult}
+10
View File
@@ -2,6 +2,7 @@ import { mkdtempSync, writeFileSync, readFileSync, rmSync } from 'node:fs'
import { join } from 'node:path' import { join } from 'node:path'
import { tmpdir } from 'node:os' import { tmpdir } from 'node:os'
import { runFfmpeg } from '../ffmpeg' import { runFfmpeg } from '../ffmpeg'
import { imageDimensions } from '../resolution'
export async function fitStill(data: Buffer, width: number, height: number) { export async function fitStill(data: Buffer, width: number, height: number) {
const dir=mkdtempSync(join(tmpdir(),'studio2-fit-')) const dir=mkdtempSync(join(tmpdir(),'studio2-fit-'))
try { try {
@@ -11,3 +12,12 @@ export async function fitStill(data: Buffer, width: number, height: number) {
return readFileSync(output) return readFileSync(output)
} finally { rmSync(dir,{recursive:true,force:true}) } } finally { rmSync(dir,{recursive:true,force:true}) }
} }
/** Only scale/pad when the still's aspect truly differs from the job canvas.
* A matching frame (typical extend handoff) must reach Comfy byte-identical. */
export function needsFit(data: Buffer, width: number, height: number) {
const size = imageDimensions(data)
if (!size?.width || !size?.height) return true
if (size.width === width && size.height === height) return false
const target = width / height, actual = size.width / size.height
return Math.abs(target - actual) > 0.01
}
+14 -4
View File
@@ -1,7 +1,7 @@
import { stylePrompt } from '~/shared/studio2/styles.mjs'; import { stylePrompt } from '~/shared/studio2/styles.mjs';
import { purge } from './cleanup'; import { purge } from './cleanup';
import { watchProgress } from './progress'; import { watchProgress } from './progress';
import { fitStill } from './media'; import { fitStill, needsFit } from './media';
import { resolveRequestSize } from './size'; import { resolveRequestSize } from './size';
import { queueSeeds } from '~/shared/studio2/seed.mjs'; import { queueSeeds } from '~/shared/studio2/seed.mjs';
import { readFileSync, mkdirSync, existsSync, unlinkSync } from 'node:fs'; import { readFileSync, mkdirSync, existsSync, unlinkSync } from 'node:fs';
@@ -29,9 +29,18 @@ type DiskFile = {
type: string; type: string;
}; };
function update(r: any, state: string) { r.state = state; r.updatedAt = Date.now(); saveRecord(r); } function update(r: any, state: string) { r.state = state; r.updatedAt = Date.now(); saveRecord(r); }
// Labels whose pixels must never be resampled: the hero identity still, and the
// extend/video start frame (which is either the exact anchor PNG or an already-canvas-sized
// extracted last frame). Fitting these every hop is what destroys identity across chained extends.
const NEVER_FIT = new Set(['hero', 'start']);
async function upload(r: any, name: string, data: Buffer): Promise<string> { async function upload(r: any, name: string, data: Buffer): Promise<string> {
const prefix = String(useRuntimeConfig().comfyFilenamePrefix).replace(/\/$/, '') + `/studio2/${r.id}/${r.index}`; const prefix = String(useRuntimeConfig().comfyFilenamePrefix).replace(/\/$/, '') + `/studio2/${r.id}/${r.index}`;
if (name !== 'hero') data = await fitStill(data, r.request.settings.width, r.request.settings.height); let fitted = false;
if (!NEVER_FIT.has(name) && needsFit(data, r.request.settings.width, r.request.settings.height)) {
data = await fitStill(data, r.request.settings.width, r.request.settings.height);
fitted = true;
}
if (name === 'start') { r.handoffFitted = fitted; saveRecord(r); }
const body = new FormData(); const body = new FormData();
body.append('image', new Blob([new Uint8Array(data)]), name + '.png'); body.append('image', new Blob([new Uint8Array(data)]), name + '.png');
body.append('subfolder', prefix); body.append('subfolder', prefix);
@@ -71,7 +80,8 @@ async function prepareGraph(r: any) {
const dir = join(studio2Root(), r.id); const dir = join(studio2Root(), r.id);
mkdirSync(dir, { recursive: true }); mkdirSync(dir, { recursive: true });
const dest = join(dir, 'handoff.png'); const dest = join(dir, 'handoff.png');
await resolveExtensionHandoffFrame({ ownerKey: r.owner, sourceClipId: q.startClipId, sourceVideoPath: clipVideoPath(r.owner, q.startClipId), destPath: dest }); const handoff = await resolveExtensionHandoffFrame({ ownerKey: r.owner, sourceClipId: q.startClipId, sourceVideoPath: clipVideoPath(r.owner, q.startClipId), destPath: dest });
r.handoffSource = handoff.source;
start = await upload(r, 'start', readFileSync(dest)); start = await upload(r, 'start', readFileSync(dest));
unlinkSync(dest); unlinkSync(dest);
} }
@@ -208,7 +218,7 @@ async function run(r: any, job: Job) {
saveRecord(r); saveRecord(r);
r.savedAt = Date.now(); r.savedAt = Date.now();
r.wallTime = (r.savedAt - r.startedAt) / 1000; r.wallTime = (r.savedAt - r.startedAt) / 1000;
const metadata = { ...structuredClone(q), id: r.id, sourceStillId:r.sourceStillId,heroReferenceAttached:r.heroReferenceAttached,sampleLatent:r.sampleLatent,sampleDenoise:r.sampleDenoise,kind: video ? 'video' : 'image', graphId: r.graphId, promptId: r.promptId, queuedAt: r.queuedAt, startedAt: r.startedAt, savedAt: r.savedAt, gpuSeconds: r.gpuSeconds, wallTime: r.wallTime, outputWidth: asset.width || s.width, outputHeight: asset.height || s.height, purgeResult: 'Left on host' }; const metadata = { ...structuredClone(q), id: r.id, sourceStillId:r.sourceStillId,heroReferenceAttached:r.heroReferenceAttached,sampleLatent:r.sampleLatent,sampleDenoise:r.sampleDenoise,kind: video ? 'video' : 'image', graphId: r.graphId, promptId: r.promptId, queuedAt: r.queuedAt, startedAt: r.startedAt, savedAt: r.savedAt, gpuSeconds: r.gpuSeconds, wallTime: r.wallTime, outputWidth: asset.width || s.width, outputHeight: asset.height || s.height, purgeResult: 'Left on host', handoffSource: r.handoffSource || null, fitted: video && q.startClipId ? !!r.handoffFitted : null };
await attachStudio2Metadata(r.owner, asset.id, metadata); await attachStudio2Metadata(r.owner, asset.id, metadata);
if (!r.outputs.some((o:any)=>o.id===asset.id)) r.outputs.push({id:asset.id,kind:video?'clip':'still',studio2:metadata}); if (!r.outputs.some((o:any)=>o.id===asset.id)) r.outputs.push({id:asset.id,kind:video?'clip':'still',studio2:metadata});
saveRecord(r); saveRecord(r);
+3
View File
@@ -44,6 +44,9 @@ export function extendSelected(form, clip, xaigen = false) {
const next = { ...form, mode: 'extend', engine: xaigen && form.engine === 'ltx' ? 'ltx' : 'minimax' } const next = { ...form, mode: 'extend', engine: xaigen && form.engine === 'ltx' ? 'ltx' : 'minimax' }
const target = clip?.id ? clip : null const target = clip?.id ? clip : null
if (!target?.id) return next if (!target?.id) return next
// Already bound to this clip: switching task must not stomp the user's overrides
// (locks, CFG, etc.) by re-copying the clip's saved settings over them.
if (form.startClipId === target.id) return next
try { return bindClip(next, target, xaigen) } try { return bindClip(next, target, xaigen) }
catch { return { ...next, startClipId: target.id } } catch { return { ...next, startClipId: target.id } }
} }
+116
View File
@@ -0,0 +1,116 @@
import test from 'node:test'
import assert from 'node:assert/strict'
import { createRequire } from 'node:module'
import { readFileSync, mkdtempSync, writeFileSync, existsSync } from 'node:fs'
import { tmpdir } from 'node:os'
import { join } from 'node:path'
import ts from 'typescript'
const require = createRequire(import.meta.url)
function load(path, resolve = require) {
const source = readFileSync(new URL(path, import.meta.url), 'utf8')
const code = ts.transpileModule(source, { compilerOptions: { target: ts.ScriptTarget.ES2022, module: ts.ModuleKind.CommonJS } }).outputText
const exports = {}
new Function('require', 'exports', code)(resolve, exports)
return exports
}
const { runFfmpeg } = load('../server/utils/ffmpeg.ts', id => id.startsWith('node:') ? require(id) : {})
const { imageDimensions } = load('../server/utils/resolution.ts')
const { fitStill, needsFit } = load('../server/utils/studio2/media.ts', id =>
id.startsWith('node:') ? require(id) : id === '../ffmpeg' ? { runFfmpeg } : id === '../resolution' ? { imageDimensions } : {}
)
const { isValidAnchorFile } = load('../server/utils/extensionFrame.ts', id =>
id.endsWith('.json') ? { default: {} } : id.startsWith('node:') ? require(id) : {}
)
// A fake PNG carrying a real signature + IHDR width/height for imageDimensions() to read.
// This is not a decodable image; tests never run ffmpeg against it (fit is either skipped
// or mocked), they only need imageDimensions() to report the right size.
function fakePng(width, height) {
const buf = Buffer.alloc(64) // isValidAnchorFile requires >= 64 bytes
buf.set([0x89, 0x50, 0x4e, 0x47], 0) // PNG signature start
buf.writeUInt32BE(width, 16)
buf.writeUInt32BE(height, 20)
return buf
}
const PNG_32x32 = fakePng(32, 32)
const text = readFileSync(new URL('../server/utils/studio2/runner.ts', import.meta.url), 'utf8')
const tree = ts.createSourceFile('runner.ts', text, ts.ScriptTarget.Latest, true)
function extractUpload(scope) {
const nodes = tree.statements.filter(n =>
(ts.isFunctionDeclaration(n) && n.name?.text === 'upload') ||
(ts.isVariableStatement(n) && n.getText(tree).startsWith('const NEVER_FIT'))
)
const js = nodes.map(n => ts.transpileModule(n.getText(tree), { compilerOptions: { target: ts.ScriptTarget.ES2022 } }).outputText).join('\n')
return new Function(...Object.keys(scope), `${js};return upload`)(...Object.values(scope))
}
function fixture() {
const uploaded = []
const record = { id: 'r', index: 0, files: [], request: { settings: { width: 32, height: 32 } } }
const scope = {
fitStill: async (data) => { throw new Error('fitStill must not be called for an already-canvas-sized handoff frame') },
needsFit,
saveRecord() {},
useRuntimeConfig: () => ({ comfyFilenamePrefix: 'preview/MiniMax_H3' }),
comfyFetch: async (_url, opts) => {
uploaded.push(opts.body)
return { ok: true, json: async () => ({ name: 'start.png', subfolder: 'preview/MiniMax_H3/studio2/r/0' }) }
},
scopedFile: () => true
}
return { scope, record, uploaded }
}
test('upload() never fits the start/hero labels when the still already matches the job canvas', async () => {
const f = fixture()
const upload = extractUpload(f.scope)
await upload(f.record, 'start', PNG_32x32)
assert.equal(f.record.handoffFitted, false)
assert.equal(f.record.files.length, 1)
})
test('upload() never fits the hero label even when it mismatches the canvas aspect', async () => {
const f = fixture()
f.record.request.settings = { width: 1344, height: 768 }
const upload = extractUpload(f.scope)
// PNG_32x32 is square, canvas is 16:9 -- would trigger a fit for any other label.
await upload(f.record, 'hero', PNG_32x32)
assert.equal(f.record.handoffFitted, undefined) // only tracked for 'start'
})
test('upload() still fits a source still whose aspect truly differs from the job canvas', async () => {
const f = fixture()
f.record.request.settings = { width: 1344, height: 768 }
let fitCalled = false
f.scope.fitStill = async (data) => { fitCalled = true; return data }
const upload = extractUpload(f.scope)
await upload(f.record, 'compose', PNG_32x32)
assert.equal(fitCalled, true)
})
test('needsFit is false for an exact canvas match and true for a real aspect mismatch', () => {
assert.equal(needsFit(PNG_32x32, 32, 32), false)
assert.equal(needsFit(PNG_32x32, 1344, 768), true)
})
test('two extends at identical width/height: the anchor PNG reaches Comfy byte-identical, no ffmpeg fit', async () => {
const dir = mkdtempSync(join(tmpdir(), 'studio2-anchor-'))
const anchorPath = join(dir, 'anchor-last.png')
writeFileSync(anchorPath, PNG_32x32)
assert.ok(isValidAnchorFile(anchorPath))
for (const hop of [0, 1]) {
const f = fixture()
const upload = extractUpload(f.scope)
const anchorBytes = readFileSync(anchorPath)
const key = await upload(f.record, 'start', anchorBytes)
assert.equal(f.record.handoffFitted, false, `hop ${hop} must not resample the anchor frame`)
const sentBody = f.uploaded[0]
const sentBlob = sentBody.get('image')
const sentBytes = Buffer.from(await sentBlob.arrayBuffer())
assert.deepEqual(sentBytes, anchorBytes, `hop ${hop} must upload the anchor PNG byte-identical`)
assert.ok(key.length > 0)
}
})