Photoshop's Relight lets you drag lights around a finished photo, at 30 credits a go

Sponsored by Klovera

Wednesday, and the ghost costume in today's lead clip is still making me smile. All three clips are character pieces built for a phone: a toddler under a bedsheet, a woman outrunning a T-Rex to claim it as a ride, and part six of a glitter-soaked serial. The news leads with Photoshop's Relight, which lets you drag a light source around a finished photo. The Secret is about how far apart two people should stand. Spacing is a trust signal, and buyers read it instantly. Over to Trending.

📈 Trending

All three of today's clips rest on one character in a costume that does the talking: a sheet with eyeholes, a beaded desert outfit at a computer, an electric-blue suit on a horse hung with bells. None of them needs dialogue to explain itself. The look is the hook, and the story arrives second.

1. A toddler in a bedsheet ghost costume says hello

▶ Watch on Instagram · @rudenko.aivideo

A small child stands on a rug in a sunlit bedroom, a white sheet pulled over the head with two eyeholes cut for round blue eyes. Arms out, the pose sits halfway between spooky and asking for a hug. The caption is a drawn-out hello, so the picture does the talking. It comes from @rudenko.aivideo, in a soft animated-film style.

📈 49,683 views — 20× the account's median

Why it works:

• Hide the face, keep the eyes. Two eyes through a sheet carry the emotion, and the hidden face keeps people looking for more.
• Light it like a memory. Late-afternoon sun through a curtain reads as a real childhood bedroom, so the render feels found rather than made.
• Give the caption one word. A single hello leaves the viewer to supply the story, and supplying the story is what keeps them watching.

2. Seedance ad: outrun a T-Rex, claim it as a ride

▶ Watch on Instagram · @brain.glitch_ml

A spec energy-drink ad from @brain.glitch_ml, made with Seedance 2.5 through OpenArt. The title card shows a woman in beads and feathers at a desktop computer in a cave. Per the caption, she summons a saddled T-Rex from the machine, it turns on her, she grabs a can, leaps clear and takes the beast as her ride. The product is the turning point.

📈 222,204 views — 12× the account's median

Why it works:

• Make the product the fix. The can arrives at the moment of danger, so the viewer sees what it does instead of hearing a claim.
• Mix eras in one frame. A beaded outfit beside a beige computer in a cave is wrong enough to stop a thumb before anything moves.
• Reverse the threat into a win. The chase becomes a ride, so the story pays off the fear it set up and leaves people grinning.

3. Glitchata part six: a Seedance serial in glitter

▶ Watch on Instagram · @arcadecalifornia

Part six of Glitchata, a serial from @arcadecalifornia, made with Arcade Studio, Midjourney, ChatGPT and Seedance 2.5 in a chain. The frame shows a rider in an electric-blue embroidered suit and sombrero on a horse hung with bells and flowers, on a wildflower hillside under a swirling sky. The caption calls it a quick play from a much bigger script, about meeting a best friend somewhere strange.

📈 58,708 views — 2.6× the account's median

Why it works:

• Saturate past what's real. Colour turned up this far tells the viewer it's a dream, so no one bothers checking for seams.
• Number the episodes. A part six promises five earlier parts, and a new viewer who likes this one has somewhere to go.
• Lead with one loaded costume. Bells, embroidery and a blue sombrero on one rider give the eye a single place to land and stay.

Try this on your next idea: pick one costume or one prop that is wrong for its setting, and build the whole clip around the moment it moves.

🤖 News & Updates

Photoshop's Relight lets you drag lights around a photo, at 30 credits a press

Photoshop's beta build has a Relight layer that puts a movable light on an image. Add it from the Layer menu, and a white light lands in the centre; drag it anywhere on or off the canvas, add more lights or hide them. Original pixels stay untouched, and Done renders at full resolution. The catch: every press of Done costs 30 credits. If you shoot product stills, this matters because you can move the key light after the fact instead of reshooting.

Eleven v4 and v4 Turbo launch, the most emotive voice model yet

Eleven v4 is out: a text-to-speech model built to read tone, pacing, emotion, character and context, with a low-latency variant, v4 Turbo. Multi-speaker scenes are meant to sound like a conversation rather than lines glued together. The company says listeners preferred it roughly three quarters of the time in blind tests against rival models, and Creator+ plans get 3x credits until October 12. If you voice ads with synthetic speech, this matters because delivery is what makes a script sound sold rather than read.

Imagen adds full manual RAW editing to its AI photo workspace

Imagen has launched Intelligent Studio, a post-production workspace that lets photographers cull and edit in whatever order suits the job. For the first time the platform includes a full manual RAW editor: JPEG and RAW files, with control over tone, colour, detail, lens corrections and creative effects, alongside a customised AI profile. Imagen says it can sit beside Lightroom rather than replace it. If you shoot your own product photos, this matters because culling, AI edits and manual fixes now happen in one window.

✨ Sponsored

🤫 The Daily Secret

The forearm rule for two people in frame

A seller I know makes clips for a cleaning spray, and this week she sent me two versions of the same scene: two women at a kitchen counter, one holding the bottle, the other asking about it. In the first, the model had put them shoulder to shoulder, arms touching, heads almost tilted together. It looked like a couple sharing a morning. In the second, she'd asked for more room, and they ended up a full stride apart, both facing the lens, the bottle marooned on the counter between them. It looked like a photo from a trade show. Neither clip was bad. Neither one sold a thing.

What she wanted was simpler than either. She wanted the feeling of two friends talking about something one of them already uses. A recommendation you overhear, not a pitch you're handed. That kind of clip works because the viewer joins the second woman: curious, a bit sceptical, ready to be convinced by someone she trusts. And trust, in a silent first frame, is measured in inches.

See you inside.
— Lauren

Published by Internet Mastery