|
•••
|
|
Some weeks the visual tools all launch at once. This is not one of them. No major image or video model shipped this week, and the headlines went instead to the business of AI: an enormous acquisition, a set of leaked financials, a model still switched off by government order. So this issue does two things. First, a clear look at which image, video, and design tools are worth reaching for right now. Then a quick pass through the news that actually moved.
A lull is a good moment to stop chasing launches and get deliberate about your stack. Below is the honest state of each category, the tool to reach for and what it is genuinely good at, followed by the week's wider news in brief.
|
|
In Images
Match the tool to the job.
|
|
For pure aesthetic quality, Midjourney is still the one most creators reach for, especially for editorial, fashion, and atmospheric work. For photorealism and developer workflows, FLUX has the edge. When the image needs readable text, a logo, a poster, an ad with a headline, Ideogram and the newer text-focused models handle it where Midjourney still stumbles. For anything client-facing where the source of the training data matters, Adobe Firefly stays the safe choice, trained on licensed and public-domain material. And for logos and brand systems that have to stay editable, Recraft is the rare model that outputs real vector files rather than flat images. There is no single winner; the skill is the match.
|
|
In Video
The field settled, for now.
|
|
The video models that mattered earlier this year still set the bar. Veo remains the most complete all-around pick, with strong realism and synchronized audio in a single pass. Kling is the value choice when you need many iterations without a premium cost. Seedance is the standout for image-to-video, and accepts the widest mix of inputs, images, clips, and audio in one generation, which is why production creators favor it. Luma's newer work pushes into true high-dynamic-range output for color-critical pipelines, and on the open side, Wan and LTX are the serious self-hostable options. Nothing new shipped this week, but the lineup is deep enough that the question is fit, not novelty.
|
|
In Design
AI moved into the canvas.
|
|
On the design side, the generation models are now baked into the tools you already use. Canva's Magic Studio turns a prompt or a rough upload into a full layout in minutes, with a brand kit that keeps colors and fonts consistent. Figma's AI focuses on the practical work of interface design, suggesting components and layouts rather than generating art. And Adobe's Firefly lives directly inside Photoshop, Illustrator, and Express, so the generation happens where you already work. The pattern across all of them is the same: AI is less a separate tool you visit and more a layer inside the canvas you already have.
|
|
All AI News
The week, in brief.
|
|
Away from the visual tools, the week was loud. SpaceX, fresh off the largest IPO in Nasdaq history, agreed to acquire the maker of the coding tool Cursor for sixty billion dollars, the most expensive AI software acquisition yet, and briefly passed Amazon in market value. OpenAI's audited 2025 numbers surfaced, showing roughly thirty-four billion dollars spent against thirteen billion in revenue. The strain on computing power went public, with Microsoft routing some GitHub traffic through a rival's servers to keep it online. AI leaders joined a G7 session on governance in France. And Claude Fable 5, the model pulled offline by a government order last week, stayed dark, with its maker in Washington for talks. A quiet week for launches, a noisy one for everything else.
|
|
The Takeaway
Use the lull.
|
|
A quiet week is not a wasted one. When nothing new is demanding to be tried, it is the right time to get deliberate: settle on the tool for each job, tighten the workflow around it, and stop reaching for whatever launched most recently. The launches will come again soon enough. Until they do, the advantage goes to whoever used the calm to get their stack in order.
|
|
•••
I am putting together a stack guide: a one-page map of which image, video, and design tool to reach for in each situation, with the strength and the catch of each, kept current as the models change. Built so you can stop guessing and start matching the tool to the job.
Want it when it ships? Reply with send me the stack guide and I will get it to you.
|
|
A QUESTION FOR YOU
Which tool are you using out of habit rather than fit?
Reply and tell me what you reach for by default. The tool you use without thinking is the one most worth checking against the alternatives.
If this was useful, forward it to a creator who reaches for the same tool for everything.
|
|
Until next time,
Luxe Prompting
|
|
Luxe Prompting
AI Image Generation for Creators
|
|
