The Studio Pro Blog

Guides, workflows & updates

Practical AI imaging for product and fashion, web design, art, and photography — plus what’s new in Nano Studio Pro.

Mira, a girl in a yellow raincoat and flying goggles, rides Ember, a giant glowing moth, over a stormy sea toward a fishing boat, holding a lantern high, with a striped lighthouse and lightning behind her

interactive film · October 18, 2026

Make interactive films with Storyboards, our newest feature

Storyboards makes interactive films: branching stories with more than one ending, where the person watching decides what happens next. Plan it in keyframes with your own characters, turn it into video, and share it with a link. Watch the trailer for Mira and the Last Light, then play the real film and pick her path yourself.

A giant stone face rising out of a desert sandstorm, three tiny explorers at its feet, with the words SEEDANCE 2.5 in the sky — a frame from the Seedance 2.5 trailer

seedance 2.5 · October 17, 2026

Seedance 2.5 is here: video with its own sound, dialogue and up to 30 seconds a shot

ByteDance's newest video model is now the default video model in Nano Studio Pro. It writes the dialogue, the sound effects and the room tone into the clip, holds a single shot for up to 30 seconds, follows your characters from reference pictures, and can edit or extend a clip you already have. A trailer, four uncut clips and what each one costs.

A single glossy yellow banana with water droplets on a dark stone pedestal, lit like a luxury product shot against a deep teal backdrop — generated by Nano Banana 2.1

nano banana 2.1 · October 7, 2026

Nano Banana 2.1 is here: better images for nearly half the credits

Google's newest image model is now the default in Nano Studio Pro, and because it costs less to run, every image does too: 55 credits at 1K instead of 100, 80 at 2K instead of 150. What changed, what your plan buys now, and a first look at what it does with video.

Six particle illustrations: a newspaper stack, a microphone with gauges, a phone with app reviews, a prompt box with a shield, a document under a magnifying glass, and a phone with text messages

jev · September 21, 2026

What Is Jev? Six Hands-On Experiments With TypeSafe's System One Model

Jev is TypeSafe AI's new System One model: it never writes a sentence, it answers typed questions with a probability for every option. We ran six experiments on real data to understand what that means in practice, with latency, cost and downloadable results.

A stack of newspapers rendered as particles, one paper pulled out in red

jev · September 21, 2026

Ranking a Live News Feed From a Two-Line Profile: Jev Experiment 01

We scored 90 live headlines against a two-line profile with Jev, then scored the same headlines for someone with the opposite tastes. The rankings flipped (rank correlation −0.71), the scores were stable across runs, and the whole thing cost half a cent.

A desktop microphone rendered as particles with an arc of small gauges above it

jev · September 21, 2026

Seventeen Questions in One Call: Building a Mac Assistant on Jev

We built a small assistant that turns a plain-English command into real actions on a Mac. Jev answers seventeen questions per command in a single ~190 ms call; code turns the answers into a plan and a confidence threshold decides whether to act or ask.

A smartphone rendered as particles showing an app icon and a row of stars, one in red

jev · September 21, 2026

Classifying 100 App Store Reviews in Three Seconds: Jev Experiment 03

We classified the latest 150 App Store reviews of Spotify and Instagram with five Jev questions each. Severity tracked the reviewers' own star ratings, the model read the text rather than the stars, and it caught reviews blaming an update that never used the word.

A chat prompt box rendered as particles with symbols in red and a shield in front of it

jev · September 21, 2026

A Prompt Pre-Flight Built From 148 Real Prompts: Jev Experiment 04

We asked Jev ten questions about every prompt typed into Nano Studio Pro's prompter over five months. It told us how many people were talking to it like a chatbot, how many assumed it remembered the last result, and which requests an image provider would refuse. The result is a pre-flight check we're shipping.

A document rendered as particles under a magnifying glass, one line highlighted in red

jev · September 21, 2026

Searching a Document Line by Line, Without Embeddings: Jev Experiment 05

We turned the 169 lines of the Apache 2.0 license into one Choice question and asked it plain-English questions. Jev found the right line with 96% probability on the clear ones, spread probability across a passage on the fuzzy ones, and said the document doesn't answer when it doesn't.

A smartphone rendered as particles showing text message bubbles, one in red with a warning sign

jev · September 21, 2026

When It Says 80%, Is It Right 80% of the Time? Testing Jev's Calibration on 5,574 Labeled SMS

We ran Jev over the full UCI SMS Spam Collection, 5,574 human-labeled messages, with one yes/no question and no training: 98.4% accuracy, AUC 0.995, Brier 0.015. The extremes of the probability range were honest within three points; the middle leaned toward spam.

A frame from a clip made with Kling 3.0 Omni: a bearded rider in a red leather jacket inside a wooden steampunk train carriage, the desert lit by sunset through the window

kling · September 16, 2026

Triple your Kling video usage, until 23 September

Every Kling model in Nano Studio Pro is billed at a third until Tuesday 23 September. Three seconds of finished video for what one used to cost, across Kling 3.0, 3.0 Omni, O1 and Motion Control. Nothing to enter and nothing to claim.

Forty identical matte black spheres in a grid on a dark surface; one is dissolving upward into warm gold particles, its empty place marked by a thin luminous gold targeting bracket

nano banana · September 14, 2026

Nano Banana 2 vs GPT Image 2.5

Ask an image model to delete one object from a photo full of near-identical ones and you learn something uncomfortable: the two leading models disagree about whether to believe your words or your coordinates. A controlled comparison of gemini-3.1-flash-image and gpt-image-2.5-sunburst, with scores, costs, token counts and every failure shown.

Products from a retail store floating around the Shopify bag, each one inside a gold detection box with its product code beside it

shopify · September 14, 2026

Put any Shopify product into any scene, in two clicks

Connect your Shopify store and your products arrive as products, with their SKUs. Pick one, pick the photo you want it in, and it is in the scene — lit for it, and tagged with the product it features, through every edit made afterwards.

The Sheets section of Nano Studio Pro: a row of kinds (Icon set, Sticker sheet, Pixel sprites, Animation, Props and objects), the animation menu set to A person and Idles and waiting with Key, Full and Smooth detail chips, a grid of ten frames of a bearded cowboy firing and reloading a revolver, and the player on the right showing one frame large

sheets · September 14, 2026

Introducing Sheets: create sprite sheets, animation frames and icon sets in one image

Sheets is a new section of Nano Studio Pro that makes one image holding many separate pieces — a sprite sheet, a walk cycle, an icon set, a sticker pack, a page of props — laid out on a grid so each one cuts out cleanly. You pick what goes on it from menus, generate it, play it back, and export the frame rectangles as Aseprite JSON for Phaser, Pixi, Godot or Unity.

A shared diorama in cinema mode: a video of a presenter plays on a screen inside a studio set with cutouts of him standing beside it, a director's chair and a camera in the foreground, and Blur and Focus sliders under the scene

product update · September 12, 2026

Product Update: Pick Your Model, Make a Sheet, and Give Your Clip a Voice

Since the last update: four image models and six video models to choose from, a new Sheets section for animation frames, sprites and icon sets, a clip editor that narrates, scores and branches, your uploads living beside your generations, and the studio in Spanish. Everything that changed, kept short.

One frame of an eight-second café clip split down the middle: the left half is the original, the right half is the same moment after one edit request — a Hawaiian shirt on him and two piña coladas where the espresso cups were, with the faces, the gestures, the copper lamp and the café unchanged

gemini omni · September 12, 2026

Gemini Omni is in the studio now, and it kept the laugh

Google's newest video model landed in Nano Studio Pro this week. We asked it to put a straw hat on her, a Hawaiian shirt on him and two piña coladas where the espressos were. It did all three. What we keep going back to is everything it did not touch.

One frame split down the middle: on the left the reference clip, a bearded cowboy in a black T-shirt inside a wooden steampunk train carriage in daylight with a rider on a neon motorcycle outside the window; on the right the clip Kling 3.0 Omni generated from it, the same shot and the same acting with the hero now in a red leather jacket and the desert lit by sunset

kling · September 12, 2026

Kling 3.0 Omni re-dressed our clip and the performance never noticed

Kling 3.0 Omni is in Nano Studio Pro now. We gave it ten seconds of our own footage and a handful of stills, then asked for different clothes and different light. What came back was the same acting wearing something else — up to fifteen seconds, up to 4K. Here is what went in and what came out.

A matte black coffee bag beside a steaming cream mug on a sand-coloured backdrop, its label reading GPT IMAGE 2.5, then FLARE · SUNBURST, then a small line reading legible text, rendered in one pass — generated by GPT Image 2.5 Flare from a prompt with the words in quotes

gpt image 2.5 · September 12, 2026

GPT Image 2.5 is here, and the small print finally survives

OpenAI shipped GPT Image 2.5 as two models, Flare and Sunburst, and both now sit beside Nano Banana 2 and Pro in Nano Studio Pro. Every picture in this post was made by them, label copy and all — which is the part worth talking about. What separates the two, and how to pick.

A dark wall of thousands of small photographs and film strips, with one print glowing warm gold — the one the search found

mcp · August 27, 2026

An Image MCP Built Around Search: Photos, the Exact Second of a Video, and the Full Tool List

Generate a picture, upload one, or import a whole catalog through the Nano Studio Pro MCP — everything becomes searchable. The search tool matches what the scanner saw inside every photo, and now inside your footage, down to the millisecond. Here is everything an agent can do with it today, tool by tool.

Motion Studio: the word BEGINS set in a custom illustrated font, each letter carrying its own artwork, over a dark canvas threaded with coloured curves

product update · August 27, 2026

Product Update: Video You Can Search, and Now Cut

Searching inside a clip now listens as well as looks — sight, sound and colour. And every clip opens into a graph where you can branch a stretch, lay a still over it, and erase down to the one thing that changed, without touching the original file.

The time corridor: a clip's frames standing in depth along its own timeline, with labelled close-up crops joined by leader lines to the detection boxes on the frame at the front

video · August 24, 2026

Search Inside Your Videos

Video is searchable in Nano Studio Pro. Ask for a clip to be read and it answers searches the way your photos do — type a word and you land on the second of the clip where it happens.

The word SHINE animated in the Masters of the Universe font, each letter separately illustrated — a guitar inside the S, a bone for the I, a halftone grid in the N — over coloured waves

fonts · August 19, 2026

Motion Studio: Kinetic Typography Without After Effects

Motion Studio animates text letter by letter in a typeface you generated yourself. A camera that chases the typing, a voice or a song it can lock onto, and an MP4 at the end. Here is everything in it, shown in the Masters of the Universe font.

A Ghibli-style meadow scene open in the Nano Studio Pro canvas editor, its layers panel listing fourteen separately named objects beside a 4K canvas badge

layers · August 15, 2026

Higgsfield Layers Is Out. Here Is What Comes After You Split an Image.

Higgsfield shipped Layers on August 11 — flat image in, editable layers out. Separation is the part everyone is talking about. This is about the part after it: what you can actually build once a photograph is in pieces.

Pixel-art scene of a developer at a desk, a sprite sheet open on the monitor and energy swirling from their hand

mcp · August 10, 2026

Why Your AI Agent Keeps Remaking Images It Already Has

Coding agents have no memory of the files they produce. Every session starts blind, so they generate a new hero image rather than find last week's — and your project drifts visually while you pay for it twice. The fix is a pattern, not a bigger context window.

A snowboarder mid-jump isolated on a transparent background, the powder spray still translucent

cutouts · August 10, 2026

How to Get a Transparent Background Image Without Photoshop

Most background removers hand back a sticker with hard edges, and everything soft — hair, spray, motion blur, glass — gets chewed off. Here is what a proper cutout produces instead, which file format actually carries transparency, and three ways to make one.

The nano-search embed widget showing live masonry search results for the query gold

image search · August 2, 2026

The Image Search API Is Live — Put Your Library's Search on Any Site

The detection-aware search behind Nano Studio Pro is now something you can ship: a REST Image Search API with browser-safe publishable keys, and a drop-in widget that puts your whole searchable library on any website in two lines of HTML.

The Nano Studio Pro detection editor: an image with labeled detection boxes and a live detections panel

announcement · July 21, 2026

Nano Studio Pro Launches: an AI Image Studio Built on Precision, Not Guesswork

Nano Studio Pro is now live — an AI image studio for anyone who works with images: product and fashion, sure, but web designers, artists, and photographers too. It pairs generation with precise, detection-driven control, because the tools that win won't just make pretty pictures — they'll let you control exactly what changes.

The Nano Studio Pro asset library with search across thousands of images

image search · July 20, 2026

How to Organize AI-Generated Images and Actually Find Them Again

AI images end up scattered across tools, downloads folders, and chat threads, and finding one later is harder than making it again. Here is how to keep them in one place and search any photo by keyword, color, or the objects inside it — including the ones you never generated.

The Nano Studio Pro studio open on a desktop display, showing a project library and canvas

mcp · July 19, 2026

How to Give Claude Code a Searchable Image Library with MCP

If you juggle several web projects, your images are probably scattered everywhere. Connect the Nano Studio Pro MCP server and Claude Desktop or Claude Code gets one searchable home for all your media — find photos by what is inside them, organize by project, and generate new images without leaving the chat.