@koda-sl/baker-cli 0.260.0-dev.07acdacde → 0.262.0-dev.c998276fe

package/README.md

@@ -5796,6 +5796,8 @@ This CLI is designed for AI agent consumption. Key patterns:
 - **0.252.0**: avatar voices removed. `--voice-id` / `--voice-description` are gone from `baker avatars create|update`, and an avatar no longer holds a voice at all — every clip is voiced by the video model as it renders, directed by the profile's `--accent`. The picker they replaced offered a voice, played its preview, and then shipped a clip that did not sound like it: the re-voice was speech-to-speech, which replaces timbre and keeps pronunciation, so the accent the user picked could never survive. Three other routes were built and judged on video before removing it — TTS + lip-sync (mouth), audio-driven avatar models (picture), and Seedance 2.5 driven by audio (refuses AI faces). Reasoning and the re-open condition: `docs/adr/0005-an-avatar-is-a-face-not-a-voice.md`.
 - **0.253.0**: the pre-render gate runs `hyperframes check` — lint, runtime, layout, motion and contrast in one browser session, sampled at transition seams — instead of the static `lint` + `inspect` pair, falling back to them when the installed binary predates `check`. The browser half is signal the parser could never produce: measured on HyperFrames' own `warm-grain` example, lint/runtime/layout/motion all reported `ok: true` while the composition rendered an almost entirely blank ten-second video, the only trace being six `GSAP target #a-roll not found` warnings — a tween pointed at an element that does not exist, so nothing it animates ever appears. Baker promotes that warning to blocking, alongside `sweep_static` ("the timeline never advanced", i.e. a still image billed as a video). Warnings are grouped by code with a count (one run returned 62, 53 of them identical) and the highest-value findings carry advice on what to do rather than only what was seen. Requires `hyperframes@0.8.17` in the sandbox, bumped from 0.7.5 in the same change. Reasoning: `docs/adr/0007-the-render-gate-runs-in-a-browser.md`.
 - **0.254.0**: the nested-composition smoke test now looks at the frames instead of trusting the exit code — exiting 0 is what a blank render does too. `hyperframes snapshot --describe` sends the captured frames to Gemini and writes back what is actually in them, and the credential Studio already uses for Gemini and Omni (`GOOGLE_GENERATIVE_AI_API_KEY`) is the `GEMINI_API_KEY` it reads; the backend now ships it to the sandbox by omission, so a deployment without one keeps the plain smoke test. Verified against a real render: the pass reported "the image is completely empty" and named the single visible element without being told what to look for. Frame verdicts are condensed into the node log, one line per frame — a file the agent would have to know to open is a file it does not read.
+- **0.262.0**: `scaffold-ad` carries the brand. The spec takes `brand.palette` (hex tokens) and `brand.logo` (repo path to the mark), and emits them as the `style.palette` and `branding` the engine already reads to colour its brand plate and CTA. The palette dresses the plate and the type and is deliberately kept OUT of the shot briefs — hex in a frame prompt makes the model paint the whole scene that colour, which reads as a filter. Also takes `market`, since an ad that does not say where it is set is an ad the model sets wherever it likes.
+- **0.261.0**: four defects that reached a finished ad, fixed in `scaffold-ad` where they originate. Frames render on **`openai/gpt-image-2`**, pinned rather than inheriting the catalogue default (Gemini Flash — right for cheap high-volume work, wrong for the opening still a five-second clip is built from). Every shot brief now states the MARKET, inferred from the voice language when unstated: an ad for Spain came back full of British houses because nothing said where it was. Every shot brief also rules out the impossible — paper and screens are opaque, objects are at real-world scale — after a bill rendered with the text from its reverse showing through, and a solar panel five metres tall. And beat durations snap to lengths the video model can actually render: free-form durations were rounded down per clip, so 11 beats summed to 27.0s of picture against a 31.4s timeline and the 4.4s difference rendered as a black tail. The read rate was re-measured off a real Spanish ad (81 words in 26.48s = 3.06 w/s); the old 2.6 over-predicted every beat by 18%.
 - **0.260.0**: burned-in captions break where the script punctuates instead of every N words. The karaoke composition sliced the transcript into fixed groups, and a word count knows nothing about the sentence it is cutting — an ad the engine rendered showed cards reading "OTRA VEZ, EN" and "TU CASA GENERA", half-thoughts ending on a preposition. A card now closes when the clause does; the word count stays as a ceiling so a long clause is still split, and a linking word ("de", "al", "y") never ends a card because it belongs to what follows. Same rule the pre-render gate already checked for, now enforced where the cards are actually built.
 - **0.259.0**: `baker canvas scaffold-ad` is listed in `baker canvas --help` with a description, and the three video routes are separated in the skill. It had a usage line and no description row, so an agent reading help saw two scaffolders explained and a bare name — and three runs from the same prompt took three different wrong paths: hand-directing the flow, reproducing a competitor ad found via `winning-ads`, and hand-authoring a canvas. The routes now state their own boundary: `scaffold-ad` is the default for a new ad, `scaffold-video` only reproduces a video the USER supplied, and `video-flow.md` opens by saying so rather than calling itself the default eleven times.
 - **0.257.0**: the frame-vision pass now asks whether what is on screen could physically happen, not just whether the frame is empty. The defect that motivated it was a generated shot of a solar panel roughly five metres tall being lifted onto a roof by one person — well lit, on brief, perfectly legible, and impossible. That is what reads loudest as "AI-generated", and the previous question could not see it: nothing was missing, something present was the wrong size. Asked the new question, the pass called the rejected shot *"physically implausible due to the immense weight and surface area"* and the replacement *"consistent with what two people could reasonably maneuver"* — the same verdict a person reached, unprompted.

package/dist/cli.js

@@ -32174,12 +32174,30 @@ var AdSpec = z29.object({
   brand: z29.object({
     name: z29.string().optional(),
     /** Figures to lift typographically — the one gold number every ad has. */
-    highlight: z29.array(z29.string()).optional()
+    highlight: z29.array(z29.string()).optional(),
+    /**
+     * Brand hex colours, most important first. They dress the brand plate and
+     * the type — never the scenes: pushing hex into a shot brief makes the
+     * model paint the whole frame that colour, which reads as a filter.
+     */
+    palette: z29.array(z29.string()).optional(),
+    /** Repo path to the brand mark, so the CTA carries it instead of a name in a font. */
+    logo: z29.string().optional()
   }).optional(),
   voice: z29.object({
     description: z29.string().min(1),
     language: z29.string().optional()
   }).optional(),
+  /**
+   * Where the ad is set, as a place a model can picture: "Spain", "Mexico",
+   * "the UK". Inferred from the voice's language when omitted.
+   *
+   * It exists because nothing else in a shot brief says where it is, and a model
+   * with no location picks its own. An ad for Spain came back full of British
+   * houses — sash windows, slate roofs, English suburbs — in every single frame,
+   * from briefs that were otherwise perfectly good.
+   */
+  market: z29.string().min(1).optional(),
   /**
    * A bed, described in words. Omitted leaves the ad dry — the engine only wires
    * music when asked, and silence is a legitimate choice.
@@ -32196,15 +32214,37 @@ function narrativeRole(index, total) {
   if (index === total - 1) return "cta";
   return index < total / 2 ? "problem" : "proof";
 }
-function estimateSeconds(line) {
+var MARKET_BY_LANGUAGE = {
+  es: "Spain",
+  "es-es": "Spain",
+  "es-mx": "Mexico",
+  pt: "Portugal",
+  "pt-br": "Brazil",
+  fr: "France",
+  it: "Italy",
+  de: "Germany",
+  en: "the United Kingdom",
+  "en-us": "the United States"
+};
+function marketFor(spec) {
+  if (spec.market) return spec.market;
+  const lang = spec.voice?.language?.trim().toLowerCase();
+  return lang ? MARKET_BY_LANGUAGE[lang] : void 0;
+}
+function beatSeconds(line) {
   const words2 = line.trim().split(/\s+/).length;
-  return Math.max(1.6, Math.round(words2 / 2.6 * 10) / 10);
+  const spoken = words2 / 3.06;
+  const allowed = [...SEEDANCE_DURATIONS].sort((a, b) => a - b);
+  return allowed.find((d) => d >= spoken) ?? allowed[allowed.length - 1];
 }
 function adSpecToBlueprint(spec) {
   const total = spec.beats.length;
+  const market = marketFor(spec);
+  const place = market ? ` Set in ${market}: the architecture, streets and styling are ${market}'s.` : "";
+  const physics = " Everything obeys real-world physics: paper, card and screens are OPAQUE with nothing showing through from behind, and every object is at believable real-world scale next to the people handling it.";
   let clock = 0;
   const scenes = spec.beats.map((beat, i) => {
-    const duration = estimateSeconds(beat.say);
+    const duration = beatSeconds(beat.say);
     const start = clock;
     clock += duration;
     return {
@@ -32227,15 +32267,16 @@ function adSpecToBlueprint(spec) {
           ...spec.voice?.description ? { voice_description: spec.voice.description } : {}
         }
       ],
-      motion_prompt: beat.on_camera ? beat.show : (
+      motion_prompt: beat.on_camera ? `${beat.show}${place}${physics}` : (
         // The subject must not be forming words: the voice belongs to the
         // narration, and a mouth moving under it reads as dubbing.
-        `${beat.show} Nobody in frame is speaking \u2014 mouths closed, no dialogue.`
+        `${beat.show}${place}${physics} Nobody in frame is speaking \u2014 mouths closed, no dialogue.`
       ),
-      start_frame_prompt: beat.show
+      start_frame_prompt: `${beat.show}${place}${physics}`
     };
   });
   return {
+    estimated_duration_s: Math.round(clock * 100) / 100,
     global: {
       transcript: spec.beats.map((b) => b.say).join(" "),
       voiceover: {
@@ -32244,7 +32285,13 @@ function adSpecToBlueprint(spec) {
         mode: "voiceover",
         ...spec.voice?.description ? { voice_description: spec.voice.description } : {}
       },
-      ...spec.brand?.name ? { brand: { name: spec.brand.name } } : {},
+      ...spec.brand?.name || spec.brand?.logo ? {
+        branding: {
+          ...spec.brand?.name ? { name: spec.brand.name } : {},
+          ...spec.brand?.logo ? { logo_path: spec.brand.logo } : {}
+        }
+      } : {},
+      ...spec.brand?.palette?.length ? { style: { palette: spec.brand.palette.map((hex) => ({ hex })) } } : {},
       ...spec.music ? { music: { music_prompt: spec.music } } : {}
     },
     scenes
@@ -32268,6 +32315,7 @@ function resolveShippedCanvasDir(name, startDir, exists = existsSync4, maxDepth
 }
 
 // src/commands/canvas/scaffold-ad.ts
+var AD_IMAGE_MODEL = "openai/gpt-image-2";
 var SHIPPED_COMPOSITION_DIR = resolveShippedCanvasDir("video-overlay-composition", import.meta.dirname);
 var SHIPPED_CAPTIONS_DIR = resolveShippedCanvasDir("tiktok-captions-composition", import.meta.dirname);
 registerSchema({
@@ -32277,7 +32325,7 @@ registerSchema({
     spec: {
       type: "string",
       required: true,
-      description: "Path to the ad spec JSON. Shape: { format?, brand?, voice?, beats: [{ say, show, on_camera? }] }. `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."
+      description: "Path to the ad spec JSON. Shape: { format?, market?, brand?, voice?, music?, beats: [{ say, show, on_camera? }] }. 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."
     },
     slug: {
       type: "string",
@@ -32349,7 +32397,7 @@ var scaffoldAdCommand = defineCommand104({
     await writeSceneFiles(outDir, blueprint);
     await writeFile9(blueprintStylePath, renderStyleProjectionFromValue(blueprint), "utf8");
     const opts = {
-      imageModel: DEFAULT_IMAGE_GENERATE_MODEL,
+      imageModel: AD_IMAGE_MODEL,
       videoModel: DEFAULT_VIDEO_GENERATE_MODEL,
       overlayCompositionPath: path22.relative(outDir, compositionDest),
       captionsCompositionPath: path22.relative(outDir, captionsDest),
@@ -32366,8 +32414,10 @@ var scaffoldAdCommand = defineCommand104({
       data: { canvas: outPath, beats: spec.data.beats.length, slug },
       hints: [
         `Wrote ${spec.data.beats.length} beats to ${outPath}. Run it with \`baker canvas run ${outPath}\`.`,
+        `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.`,
         "The caption cards are your `say` lines verbatim \u2014 read them back before running; that is the copy the viewer sees.",
-        "Every clip renders with nobody speaking unless a beat set `on_camera`. If a beat needs a presenter talking, set it there rather than describing it in `show`."
+        "Every clip renders with nobody speaking unless a beat set `on_camera`. If a beat needs a presenter talking, set it there rather than describing it in `show`.",
+        "If the spec carried no `brand`, read src/brand/BRAND.md and re-run with its palette and logo \u2014 an unbranded ad is the client's message in someone else's clothes."
       ]
     });
   }