Adobe brings music, speech and sound effect generation to Firefly
Sponsored by Klovera
A castle under fire got me to my desk today. The three clips are all a single painted frame: archers on a wall as fire falls on a city, a witch under a full moon, and a food-truck cook mid-bite on a poster for an ad tool. Adobe leads the news, adding music, speech and sound effects to Firefly. The Secret is a two-camera rule for cutting a street interview, which matters because a buyer feels a wrong cut before they can name it. Trending takes it from here.
In today's paid edition: 3 clip recipes · 5 stories · the full Secret · member pricing on Klovera — Upgrade $3.99/mo.
📈 Trending
None of today's three clips needs motion to work. Each one is sold by a single frame you could print: archers waiting for fire to land, a cloaked figure under a moon the size of the sky, a cook mid-bite on what is really a poster for an ad tool. The thumbnail does the selling, and the clip just has to not ruin it.
1. Archers on a castle wall as fire rains down
▶ Watch on Instagram · @mrbashr_ai
A line of helmeted archers stands at a sandstone battlement, bows raised, looking across a river at a walled city of pale spires under an orange evening sky. Three burning projectiles trail black smoke as they fall toward it. The caption is five words, left because he was hurt, pasted over a siege. @mrbashr_ai made it with Flowwall and points viewers to a Discord for credits.
📈 119,211 views — 7.8× the account's median
Why it works:
- Open on the widest shot. A city, a river and an army read in one glance, so the scale lands before the thumb moves.
- Put a small line over spectacle. Five words about someone being hurt turn a siege into a feeling, and feelings get comments.
- Grade it like golden hour. Warm low sun and haze are what film looks like, so the eye accepts it as footage.
Steal this — paste into Klovera
A prompt that makes a clip like this
2. A witch on a rock, a moon, a castle
▶ Watch on Instagram · @minuit.art.ia
A figure in a pointed hat and a long rust-red cloak stands on a spike of black rock in a dark sea, back to us, a broom in hand. Behind, a full moon fills the frame with a castle in silhouette inside it, fog drifting past bare trees and small fires on the rock. The caption says only sweet dreams. @minuit.art.ia built it with Midjourney and Seedance.
📈 53,938 views — 2.4× the account's median
Why it works:
- Light with one source. The moon does all the lighting, so the figure becomes a silhouette and the eye has nowhere else to go.
- Keep the face hidden. A back turned to us invites the viewer to supply the story, which keeps them watching for a turn.
- Let costume become landscape. The cloak runs into the rock as one shape, which makes the frame feel designed rather than generated.
Steal this — paste into Klovera
A prompt that makes a clip like this
3. Higgsfield says you can run ads from inside ChatGPT
▶ Watch on Instagram · @higgsfield.ai
A cook in a red-striped shirt and apron sits on the bumper of a red and white food truck, biting into a foil-wrapped burrito, a Ferris wheel behind him, under the words Paid Ads. @higgsfield.ai's caption claims the pairing with GPT Sol 6.1 researches competitors, turns one product photo into a batch of ads and launches campaigns from inside ChatGPT. Comment ADS for the setup.
📈 387,416 views — 1.8× the account's median
Why it works:
- Lead with one finished image. A food-truck still that looks shot proves the tool before any feature list gets read.
- Promise the whole job. Strategy to finished variations in one prompt is the thing a small seller actually wants to hear.
- Gate the setup behind a comment. Asking for ADS turns every curious viewer into a comment, and comments push the post further.
Steal this — paste into Klovera
A prompt that makes a clip like this
Try this on your next idea: before you generate anything, pick the one frame that would stop a scroll on its own and build the rest of the clip outward from it.
🤖 News & Updates
Adobe adds music, speech and sound effects to Firefly
Adobe is bringing music, speech and sound effect generation into Firefly. Firefly launched in 2023 and has grown into an all-in-one creative studio for creating, editing and enhancing images, video and audio, with models from Adobe alongside Google, OpenAI, Runway, Kling AI and Luma AI. Adobe's framing is blunt: audio is the last thing creators add and the first thing viewers notice. If you cut product videos in Adobe tools, this matters because voice and soundtrack now sit next to the picture.
Suno adds spoken voice generation in public beta
Suno, until now a music generator, has launched Speech, a feature that generates spoken voices from a script or from a prompted description of the voice you want. It is in public beta on Suno's web and mobile apps, and it can generate the voiceover and the background music in the same pass. If you narrate product videos, this matters because a voice and a music bed made together tend to sit in the mix without a fight.
Instagram's Edits app gets an AI assistant that tells you how to post
Instagram is adding an AI creative assistant to Edits, its standalone editing app. Announced on Wednesday, it pulls data from your Instagram account and suggests what to change in a post, which makes Instagram the latest platform to build this kind of feedback in. Read the wording: it tells you how to post, and my read is that it grades you against what Instagram rewards. If you publish product Reels, this matters because the platform is showing its hand about what it wants from you.
4. @hedra_labs makes a 15-second matcha ad from one prompt inside Claude
5. Twitch is using AI to decide which streamers get ad dollars
Stories 4 and 5 in full, and what each means for what you make
✨ Sponsored
The street interview is the highest-performing ad format we have. An interviewer, a stranger, a mic and a real street. That is the whole recipe, and it has not needed changing.
Klovera makes them from a product link. Paste the link and you get the interviewer, the stranger, the mic and the street. See it at klovera.ai.
🤫 The Daily Secret
Wide, singles, wide: the two-camera cut
This week a skincare seller sent me a street interview ad to look at. Thirty seconds, two people, a mic, and not one frame with both of them in it. Interviewer tight. Stranger tight. Interviewer tight again. The performances were fine and the product line was clear, but by the third cut I had stopped believing they were on the same pavement. I kept waiting for the shot that would put them together, and it never came.
What she wanted was closeness. Every reaction big on the screen, every word readable off the lips, no dead space around the faces. She had read somewhere that phones reward tight shots, which is true, and she had taken it all the way: nothing but singles, cut on every line. The wish underneath was that the viewer would feel inside the conversation. That is the right wish. It is also exactly what the all-singles cut takes away.
The rest of the Secret: why it fails, three fixes and the one thing to do today
Lauren’s P.S.
See you inside.
— Lauren
Published by Internet Mastery
Go paid for $3.99/mo — every clip's recipe, all five stories, the full Secret, and member pricing on Klovera.
