@koda-sl/baker-cli 0.305.0-dev.881741ff3 → 0.306.0-dev.2b6124eb2

package/dist/cli.js

@@ -67,7 +67,7 @@ import {
   validateCanvasDeep,
   withoutCastName,
   ytDlpBlockSignal
-} from "./chunk-ZILPAKE6.js";
+} from "./chunk-UBNLCVGN.js";
 import {
   csvOrJson,
   daysAgoIso,
@@ -29920,7 +29920,7 @@ import { defineCommand as defineCommand101 } from "citty";
 import { defineCommand as defineCommand96 } from "citty";
 
 // src/commands/avatars/casting.ts
-var CAST_FLAG_RULE = "Cast with `--avatar <handle>` on `baker studio generate`, `baker studio animate` and `baker canvas scaffold-ad`. Do NOT hand-roll it by pasting `subjectDescription` into the prompt or by passing the sheet through `--reference` \u2014 both render something that merely resembles them. `--avatar` is also the ONLY thing that carries their pinned VOICE and how they speak into a clip; without it the video model invents a different voice, and no later step can put theirs back.";
+var CAST_FLAG_RULE = "Cast with `--avatar <handle>` on `baker studio generate`, `baker studio animate` and `baker canvas scaffold-ad`. Do NOT hand-roll it by pasting `subjectDescription` into the prompt or by passing the sheet through `--reference` \u2014 both render something that merely resembles them. `--avatar` is also the ONLY thing that carries their ACCENT and delivery into a clip. An avatar has no pinned voice by design (ADR 0005): the video model renders the voice, and the region on their profile is the only thing steering it \u2014 without the flag the clip is performed flat, in whatever accent the model picks on its own.";
 var VERBATIM_RULE = "When you do write the description yourself \u2014 a hand-built canvas node, a landing image, anywhere `--avatar` does not exist \u2014 copy `subjectDescription` VERBATIM, word for word. Re-phrasing it per generation is the other reason a face drifts across a set.";
 var VIDEO_ROUTING = "In video, a photoreal presenter renders on `google/veo-3.1` (or `google/veo-3.1-fast`), never `bytedance/seedance-2.0` \u2014 it refuses photoreal human faces, AI-generated ones included. Scaffolding a video creative: `baker canvas scaffold-video \u2026 --real-face`.";
 function castingHints(avatar) {
@@ -30020,7 +30020,7 @@ function creationHints(created, profile) {
     ...casting,
     ...thin ? [thin] : [],
     ...castingHints({ handle: created.handle, status: created.status }),
-    `Once it is ready, cast it with \`--avatar ${created.handle}\` on \`baker studio generate\` / \`baker studio animate\` \u2014 that flag is what grounds the render on its identity sheet and, for a clip, what makes it speak in this avatar's own voice.`
+    `Once it is ready, cast it with \`--avatar ${created.handle}\` on \`baker studio generate\` / \`baker studio animate\` \u2014 that flag is what grounds the render on its identity sheet and, for a clip, what carries their accent and delivery into the read.`
   ];
   if (created.likeness === "licensed") {
     hints2.push(
@@ -30224,8 +30224,6 @@ registerSchema({
       description: "The company's statement that it may use this real person's image in advertising. Required with --source-image. Only the user can give this \u2014 ask them, never assume it.",
       required: false
     },
-    "voice-id": { type: "string", description: "Voice id to speak with in video", required: false },
-    "voice-description": { type: "string", description: "How the voice should sound", required: false },
     face: {
       type: "string",
       description: "Who picks the face, for an invented avatar: 'ask' (default) blocks on the user through `request_avatar_casting`, 'auto' keeps the one this build casts and asks nobody. Use 'auto' only when there is no one to answer \u2014 a Schedule, a batch, an explicit \"do it yourself\" \u2014 because a person who is there almost always wants this choice.",
@@ -30270,8 +30268,6 @@ var createCommand2 = defineCommand96({
       required: false,
       default: false
     },
-    "voice-id": { type: "string", description: "Voice id", required: false },
-    "voice-description": { type: "string", description: "How the voice should sound", required: false },
     face: {
       type: "string",
       description: "Who picks the face of an invented avatar: 'ask' (default, blocks on the user) or 'auto' (asks nobody)",
@@ -30339,8 +30335,6 @@ var createCommand2 = defineCommand96({
         ...handle ? { handle } : {},
         profile: { ...profile, subjectDescription },
         ...sourceImageIds ? { sourceImageIds, likenessConfirmed: true } : {},
-        ...args["voice-id"] ? { voiceId: args["voice-id"] } : {},
-        ...args["voice-description"] ? { voiceDescription: args["voice-description"] } : {},
         ...chatIdFromEnv() ? { chatId: chatIdFromEnv() } : {},
         // Sent so the backend can hold the sheet until the chosen portrait exists —
         // otherwise it fuses one from a face nobody picked and pays for it twice.
@@ -36963,6 +36957,22 @@ async function reviewRenderedVideo(opts) {
     await rm5(dir, { recursive: true, force: true });
   }
 }
+async function reviewContextBeside(canvasPath) {
+  const raw = await readFile11(path15.join(path15.dirname(canvasPath), "prompt.json"), "utf-8").catch(() => null);
+  if (!raw) return {};
+  try {
+    const blueprint = JSON.parse(raw);
+    const global = blueprint.global ?? {};
+    const text2 = (value) => typeof value === "string" && value.trim() ? value.trim() : void 0;
+    return {
+      ...text2(global.market) ? { market: text2(global.market) } : {},
+      ...text2(global.branding?.name) ? { subject: text2(global.branding?.name) } : {},
+      ...text2(global.transcript) ? { script: text2(global.transcript) } : {}
+    };
+  } catch {
+    return {};
+  }
+}
 async function sceneWindowsBeside(canvasPath) {
   const dir = path15.join(path15.dirname(canvasPath), "scenes");
   const files = await readdir6(dir).catch(() => null);
@@ -37321,6 +37331,7 @@ ${describeRewrites(healed.rewrites)}
     const review = reviewable ? await reviewRenderedVideo({
       video: reviewable,
       windows: await sceneWindowsBeside(filePath),
+      ...await reviewContextBeside(filePath),
       brief: opts.brief?.trim() || void 0,
       onSpend: (usage) => reviewUsage.push(usage)
     }) : null;
@@ -38644,7 +38655,11 @@ var AdSpec = z37.object({
   end_card: z37.union([
     z37.literal(false),
     z37.object({
-      /** The words on the button. Defaults to the last beat's line. */
+      /**
+       * The words on the button. No default — with none the card carries the mark
+       * alone, because falling back to the last beat's line printed that sentence
+       * twice on the closing frame. See `endCardCta`.
+       */
       cta: z37.string().min(1).optional()
     })
   ]).optional(),
@@ -38971,6 +38986,9 @@ function adSpecToBlueprint(spec, opts = {}) {
     estimated_duration_s: Math.round(clock * 100) / 100,
     global: {
       transcript: spec.beats.map((b) => b.say).join(" "),
+      // Resolved, not declared: `marketFor` falls back to the voice language and then to
+      // the script itself, so an ad that never named a place still says where it is set.
+      ...market ? { market } : {},
       voiceover: {
         mode: voiceoverMode(spec, nativeSpeech),
         ...spokenVariety(spec, avatarAccent) ? { accent: spokenVariety(spec, avatarAccent) } : {},
@@ -38996,8 +39014,9 @@ function adSpecToBlueprint(spec, opts = {}) {
     scenes
   };
 }
+var MIN_BEAT_WORDS = 4;
 function beatsTooShortToPlay(spec) {
-  return spec.beats.map((beat, i) => ({ i, spoken: beat.say.trim().split(/\s+/).filter(Boolean).length / WORDS_PER_SECOND2 })).filter(({ spoken }) => spoken < SPOKEN_CLIP_DURATION_S).map(({ i }) => i + 1);
+  return spec.beats.map((beat, i) => ({ i, words: beat.say.trim().split(/\s+/).filter(Boolean).length })).filter(({ words: words2 }) => words2 < MIN_BEAT_WORDS).map(({ i }) => i + 1);
 }
 
 // src/engine/scaffold/ad-brand-overlay.ts
@@ -39248,7 +39267,7 @@ registerSchema({
     spec: {
       type: "string",
       required: true,
-      description: 'Path to the ad spec JSON. Shape: { format?, market?, brand?, cast?, end_card?, voice?, music?, beats: [{ say, show, on_camera?, cast? }] }. Fill `brand` from src/brand/BRAND.md \u2014 `palette` (its hex tokens, most important first) and `logo` (the repo path to the mark) are what make the ad look like the client rather than like stock. `market` is where the ad is SET; omitted, it is inferred from the voice language, and getting it wrong is what fills a Spanish ad with British houses. `say` is ONE clause ending in its own punctuation \u2014 it becomes a caption card verbatim, so two sentences in one beat produce a card holding both. `show` is the shot brief for that line. Set `on_camera` ONLY when that beat\'s subject talks to camera. A beat\'s `sound` is one sound EFFECT for that shot \u2014 and MOST BEATS SHOULD NOT HAVE ONE. Add it only where the picture shows a specific physical event the viewer expects to hear (a drill tightening a bolt, a door closing, rain on glass, a graphic snapping in). On a talking head or a calm lifestyle shot it reads as a stock sting, and one per beat turns a quiet ad into a noisy one. A music bed plus two well-placed effects beats eight. `voiceover: false` makes it a MUSIC-LED ad: nobody speaks, the `say` lines become on-screen text captioned straight from the script, and it needs `music`. `cast` is WHO the ad is about, and an ad with people in it needs one: `{ "avatar": "marta" }` names a cast avatar (`baker avatars list`) and every beat is grounded on that avatar\'s identity sheet with its subject description copied verbatim \u2014 the same face here as in the rest of the company\'s work. Without it each beat invents its own stranger, which is how one 28-second ad came back with five different men playing one customer. `{ "description": "..." }` is the fallback when there is no avatar, and it scaffolds a canvas that asks you to drop a photo before it can run \u2014 so prefer the avatar. A beat sets `cast: false` for a shot they are not in. `end_card` is on by default whenever `brand` is set: the last beat becomes a flat brand plate with the mark and a call to action drawn over it. `{ "cta": "..." }` sets the button copy, `false` keeps the footage.'
+      description: 'Path to the ad spec JSON. Shape: { format?, market?, brand?, cast?, end_card?, voice?, music?, beats: [{ say, show, on_camera?, cast? }] }. Fill `brand` from src/brand/BRAND.md \u2014 `palette` (its hex tokens, most important first) and `logo` (the repo path to the mark) are what make the ad look like the client rather than like stock. `market` is where the ad is SET; omitted, it is inferred from the voice language, and getting it wrong is what fills a Spanish ad with British houses. `say` is ONE clause ending in its own punctuation \u2014 it becomes a caption card verbatim, so two sentences in one beat produce a card holding both. `show` is the shot brief for that line. Set `on_camera` ONLY when that beat\'s subject talks to camera. A beat\'s `sound` is one sound EFFECT for that shot \u2014 and MOST BEATS SHOULD NOT HAVE ONE. Add it only where the picture shows a specific physical event the viewer expects to hear (a drill tightening a bolt, a door closing, rain on glass, a graphic snapping in). On a talking head or a calm lifestyle shot it reads as a stock sting, and one per beat turns a quiet ad into a noisy one. A music bed plus two well-placed effects beats eight. `voiceover: false` makes it a MUSIC-LED ad: nobody speaks, the `say` lines become on-screen text captioned straight from the script, and it needs `music`. `cast` is WHO the ad is about, and an ad with people in it needs one: `{ "avatar": "marta" }` names a cast avatar (`baker avatars list`) and every beat is grounded on that avatar\'s identity sheet with its subject description copied verbatim \u2014 the same face here as in the rest of the company\'s work. Without it each beat invents its own stranger, which is how one 28-second ad came back with five different men playing one customer. `{ "description": "..." }` is the fallback when there is no avatar, and it scaffolds a canvas that asks you to drop a photo before it can run \u2014 so prefer the avatar. A beat sets `cast: false` for a shot they are not in. `end_card` is on by default whenever `brand` is set: the last beat becomes a flat brand plate with the mark and a call to action drawn over it. `{ "cta": "..." }` sets the button copy \u2014 there is no default, so with no `cta` the card carries the mark alone, and a `cta` repeating the closing line is dropped rather than printed twice. `false` keeps the footage.'
     },
     avatar: {
       type: "string",
@@ -39468,14 +39487,15 @@ var scaffoldAdCommand = defineCommand111({
       process.exit(1);
       return;
     }
-    if (!noBrand && resolvedLogo?.toLowerCase().endsWith(".svg")) {
-      const svg = await readFile16(resolvedLogo, "utf-8").catch(() => null);
+    const markPath = spec.data.brand?.logo?.trim();
+    if (!noBrand && markPath?.toLowerCase().endsWith(".svg")) {
+      const svg = await readFile16(markPath, "utf-8").catch(() => null);
       if (svg === null) {
         writeJson({
           ok: false,
           error: {
             code: "VALIDATION_ERROR",
-            message: `The brand mark at \`${resolvedLogo}\` could not be read, so there is no way to tell whether it is a real logo or a typed stand-in. An ad is not shipped on a mark nobody could open.`,
+            message: `The brand mark at \`${markPath}\` could not be read, so there is no way to tell whether it is a real logo or a typed stand-in. An ad is not shipped on a mark nobody could open.`,
             fix: `Check the file is there and readable, or put the client's logo in \`${BRAND_DIR}/logos/\` \u2014 the file their designer delivered \u2014 an outlined SVG, or the PNG they gave you.`
           }
         });
@@ -39487,7 +39507,7 @@ var scaffoldAdCommand = defineCommand111({
           ok: false,
           error: {
             code: "VALIDATION_ERROR",
-            message: `The mark at \`${resolvedLogo}\` draws its wordmark with live \`<text>\`, not outlines. That is a typed stand-in, not a logo: it renders in whatever font the machine happens to have, and the one that shipped this way came out reading "greenle\xE1" in every frame.`,
+            message: `The mark at \`${markPath}\` draws its wordmark with live \`<text>\`, not outlines. That is a typed stand-in, not a logo: it renders in whatever font the machine happens to have, and the one that shipped this way came out reading "greenle\xE1" in every frame.`,
             fix: `Ask the client for their real logo file and put it in \`${BRAND_DIR}/logos/\` \u2014 the one their designer delivered, as an outlined SVG or a PNG they gave you. Do not redraw it, and do not export this file to PNG: rasterising it only bakes the invented letters into pixels, where nothing downstream can tell they are wrong.`
           }
         });
@@ -39495,7 +39515,7 @@ var scaffoldAdCommand = defineCommand111({
         return;
       }
     }
-    if (resolvedLogo && !await markIsReal(resolvedLogo)) {
+    if (!noBrand && markPath && !await markIsReal(markPath)) {
       writeJson({
         ok: false,
         error: {
@@ -39621,7 +39641,7 @@ var scaffoldAdCommand = defineCommand111({
         // stops, the frames are shown, and continuing pays for none of them twice.
         handle || staged ? `Wrote ${spec.data.beats.length} beats to ${outPath}. Run it with \`baker canvas run ${outPath} --until frames --brief "<what they asked for, their words>"\`, SHOW the frames it returns and ask whether to carry on \u2014 then continue with the \`--run-id\` it gives you. This ad carries ${handle ? "a person's face" : "the client's mark"}, and a wrong look costs a whole ad to discover after the clips.` : `Wrote ${spec.data.beats.length} beats to ${outPath}. Run it with \`baker canvas run ${outPath} --brief "<what they asked for, their words>"\`.`,
         ...autoCast ? [
-          `Cast \`${autoCast}\` \u2014 the spec described this person in words, and they are one of this company's avatars. Every shot now grounds on their identity sheet and they speak in their own pinned voice. Write \`"cast": { "avatar": "` + autoCast + '" }` yourself next time; a written description is for someone the company does not have.'
+          `Cast \`${autoCast}\` \u2014 the spec described this person in words, and they are one of this company's avatars. Every shot now grounds on their identity sheet and carries their accent and delivery. Write \`"cast": { "avatar": "` + autoCast + '" }` yourself next time; a written description is for someone the company does not have.'
         ] : [],
         `These beats run about ${blueprint.estimated_duration_s}s. If the ask was longer, add beats \u2014 the length is the sum of the lines, so a 25s ad needs roughly 25s of script and cannot be stretched by holding shots.`,
         ...logoPath && !staged ? [`Could not read the logo at \`${logoPath}\` \u2014 the ad has no mark on screen. Give the path as it appears in src/brand/BRAND.md, relative to the workspace root.`] : [],
@@ -39669,7 +39689,7 @@ var scaffoldAdCommand = defineCommand111({
 function shortBeatHint(short, total) {
   if (short.length === 0) return [];
   return [
-    `Beat${short.length === 1 ? "" : "s"} ${short.join(", ")} of ${total} ${short.length === 1 ? "is" : "are"} under ${SPOKEN_CLIP_DURATION_S}s \u2014 a beat lasts exactly as long as its line takes to read, so a two-word line is a three-second shot, and a run of those plays as a slideshow rather than an ad. Do NOT pad them: that buys seconds of somebody standing still. Merge them, so each beat carries a whole thought \u2014 an ad of this length usually wants two or three beats, not six. Edit the spec and re-run this command.`
+    `Beat${short.length === 1 ? "" : "s"} ${short.join(", ")} of ${total} ${short.length === 1 ? "is" : "are"} a fragment rather than a line \u2014 a few words with a whole shot to themselves. Every beat's picture holds for at least ${SPOKEN_CLIP_DURATION_S}s, so a two-word beat buys six seconds of somebody standing still, and a run of those plays as a slideshow rather than an ad. Do NOT pad them. Merge them into the beat before or after, so each one carries a whole thought. Edit the spec and re-run this command.`
   ];
 }
 async function fetchClientLogo() {