Mohammad Nasravi

Chat with Mohammad on WhatsAppگفت‌وگو با محمد در واتساپتواصل مع محمد عبر واتساب

AI Daily: Google Omni 1.2 Flash hits 15 seconds, 30% cheaper

Branded AI Daily card over a dark editing bay where a long row of fifteen glowing cyan glass panels recedes into the distance.

Written by Mohammad Nasravi inنوشته شده توسط محمد نصراوی در

Google’s video model now makes 15 seconds in one take, at $0.07 a second for 720p. That is the length of a social ad. The price that matters is still the price of a take you keep.

Omni 1.2 Flash: 15 seconds, edits by conversation, 30% off

On 5 November Google made Gemini Omni 1.2 Flash generally available in the Gemini API and Gemini for business, and put it in Flow, AI Studio and Google Ads. Clips run 3 to 15 seconds, up from 10. You draft at 360p or 720p, then upscale to 1080p or 4K. 720p costs $0.07 a second, 30% below Omni 1.1. You refine by conversation: swap a product or a background, change the light, change the ratio. Google says packaging type and brand colours hold across camera angles, and lists fast sports action, rapid script changes and complex physics as still in development (Google, 5 Nov). Paid-tier rate limits doubled, and Omni 1.1 Flash requests route to 1.2 from 3 December (API changelog, 5 Nov).

My read: A 15-second clip at 720p costs $1.05. I reject 60–70% of my takes, so budget about three tries: a little over $3 per keeper, before your own hours. Google’s post prices 720p only, so check the upscale rate before you quote. The packaging claim is the one to test. It is Google’s claim, not my result. On the Festival Toffee 3D ad my job included the product sheet, and every shot was checked against it. A longer take means more frames where small print can slip, so that check gets harder.

Vids gives away three 15-second clips a month

Vids moved to Omni 1.2 Flash the same day. Anyone with a Google account can make up to three 15-second videos a month at no cost, from a prompt or reference images (Google, 5 Nov). Google’s help page still says “Generated images and videos are for use only within Vids”, and gives a different number: “Most users can generate up to 50 videos per month” (Google Help, read 7 Nov). Custom image avatars in Vids are on hold: “Rollout for this feature is paused” (Workspace recap, 6 Nov).

My read: Three clips is 45 seconds a month. At my rejection rate that is one clip you would show anyone. Free is a sketchpad. I used Vids on a paid explainer in October, on the older model: 22 clips, six in the bin. Takes came back with domes, minarets and nonsense shop signs nobody asked for, and the look stayed cartoon until I wrote “photorealistic live-action” into the prompt. Test whether 1.2 fixes that before you trust it. And ask legal about that “only within Vids” line before a free clip goes into paid media.

Google voices can now be remixed, and two old models get a date

On 4 November the Gemini API added voice remixing for Gemini 3.8 TTS: change a prebuilt or designed voice with plain instructions and store it as a new voice, in the API or AI Studio. On 2 November Gemini 2.5 Pro and 2.5 Flash left the AI Studio picker. From 1 December API calls to them route to newer models, with breaking changes (API changelog). Replicating a real person is a separate route: a paid project, plus a consent recording from the same adult speaker (Google docs).

My read: Remixing solves a small, real problem: a house voice with no actor’s contract behind it. The limit is that nobody owns the base voice, so do not call it exclusive. I localise one hero film into several language versions, and the voice is what has to survive each market. Test yours in every language before you commit. If any tool in your pipeline calls Gemini 2.5 by name, find it this week.

What I’d do today

Spend $3. Run one 15-second test on your real pack with the small print in shot, three takes. Compare the last frame against your artwork. If the type holds, you have a cheap first draft. If it drifts, keep building from short shots. If you want that film directed, this is what I do.

Sources

خبرهای هوش مصنوعی را در تلگرام دنبال کنخلاصهٔ خبرها و برداشتِ من از کارِ واقعی، در کانالِ Nasravi.com در تلگرام.
عضویت در کانال تلگرام

Written by Mohammad Nasravi — AI creative director, in-house at a major FMCG group. Four years putting AI-made commercials past real brand teams, including Shirin Asal campaigns for Gulfood 2026 and Canton Fair 2025. More about the work · Start a project نوشتهٔ محمد نصراوی — کارگردان خلاق هوش مصنوعی، درون یک گروه بزرگ FMCG. چهار سال است کارِ ساخته‌شده با هوش مصنوعی را از فیلتر تیم‌های برند رد می‌کند؛ از جمله کمپین‌های شیرین عسل برای گلفود ۲۰۲۶ و کانتون فر ۲۰۲۵. دربارهٔ کارها · شروع پروژه بقلم محمد نصراوي — مدير إبداعي للذكاء الاصطناعي من داخل مجموعة FMCG كبرى. أربع سنوات من تمرير إعلانات مصنوعة بالذكاء الاصطناعي عبر فرق علامات حقيقية، منها حملات Shirin Asal في جلفود 2026 ومعرض كانتون 2025. المزيد عن العمل · ابدأ مشروعك