FOUNDING OFFERSign up to unlock 35% off your first plan.

Stillmore author

Lan He

I'm Lan. Years of writing promotional and editorial work in the US, mostly for companies with big audiences and bigger expectations — where a single article might reach hundreds of thousands of readers, and where the good ones kept earning traffic long after the campaign ended.

32 published articles
Lan He

Articles by Lan He

13–18 of 32 articles

Stillmore Blog cover for GPT-6.2 Astra, GPT-6.2 Sol, featuring black woven loop sculpture.

News

How to Use GPT-6.2 Astra and GPT-6.2 Sol for Video Scripts and Storyboards

Use GPT-6.2 Sol to draft everyday short-video scripts, hooks, and shot lists at lower cost, and reserve GPT-6.2 Astra for complex multi-step briefs, template-matched merch decks, and 3D or interactive storyboard prototypes before external video rendering. Choose Sol for routine copy iterations and Astra when deliverables must follow layout references or visual templates. Lock permanent brand guidelines in an unchanged opening prefix before editing scenes, and verify frame concepts against real product photos. Because neither model exports finished video commercials, independent businesses that prefer not to write prompts, composite frames, and edit timelines themselves can upload existing phone photos and offer details to Stillmore, where the team plans, produces, and checks finished videos, images, and captions that you publish.

Lan He· 13 min read
Stillmore Blog cover for MAI Image 2.7, MAI Image 2.7 Flash, featuring golden spaniel and glass frame.

News

MAI Image 2.7 and MAI Image 2.7 Flash vs Stillmore: DIY Image Editing or Managed Content?

MAI Image 2.7 and MAI Image 2.7 Flash are DIY image generation and editing models for teams assembling assets in-house, whereas Stillmore is an advertising creative agency that turns existing business photos into finished images, short videos, and social captions. The core workflow difference is self-managed visual editing versus delegated campaign production: Microsoft's tools leave art direction, copywriting, and video assembly to your staff, while Stillmore handles planning, editing, and pre-delivery review. On pricing, Microsoft Foundry preview lists MAI Image 2.7 near $35 and Flash near $10 per 1,000 raw images, whereas Stillmore uses Plan Credits at 500 per finished image or 3,000 per video. Choose Microsoft's models when your team has design capacity for standalone graphics; choose Stillmore when you want to upload phone photos to Library, receive checked posts, and publish to your own channels.

Lan He· 14 min read
Stillmore Blog cover for MAI Image 2.7, MAI Image 2.7 Flash, featuring golden spaniel and glass frame.

News

How to Use MAI Image 2.7 and MAI Image 2.7 Flash for Product Images

To use MAI Image 2.7 and MAI Image 2.7 Flash for product images, divide tasks by edit depth and preview budget: run early layout trials and cutouts on Flash (~$10 per 1,000 images), then move winning compositions to MAI Image 2.7 (~$35 per 1,000) for multi-step surface refinements. Both models support transparent backgrounds so you can separate merchandise from messy counters without hand-drawing masks, though you still need to check shape, color, and printed labels against raw phone photos before posting. If you run an independent store without spare hours for prompting, shadow blending, and audits, sending phone photos to Stillmore shifts production to a creative team that may evaluate models alongside compositing, check finished details, and deliver images and captions for you to publish.

Lan He· 12 min read
Stillmore Blog cover for Atlas, featuring origami bird and miniature architecture.

News

How to Use World Labs Atlas Viewfinder and Chisel for 3D Video Renders

To create 3D scene video renders with World Labs Atlas, upload overlapping photos into Viewfinder to establish spatial context and render virtual camera paths, then add Chisel grayboxing, live browser previews, relighting, or 3D mesh export only when a scene needs structural or downstream edits. Overlapping angles curb guesswork before rendering up to 1440p camera-path clips, because hidden product backs or cropped edges leave missing geometry for the model to invent. Even after exporting a smooth clip, turning raw renders into commercial posts still requires compositing, product accuracy checks, and social copy. Independent businesses that prefer finished posts over learning 3D pipelines can send existing phone photos to Stillmore, where a creative team handles direction, editing, and verification while the business owner controls publishing.

Lan He· 14 min read
Stillmore Blog cover for Hy4 flash, featuring glass triangle and cyan vertices.

News

What Is Hy4 flash? Multimodal Specs, WorkIndex Tasks, and Pricing

Hy4 flash is Tencent's November 16, 2026 multimodal AI model built for fast, low-cost office and media workflows rather than heavy software engineering. For business teams, the release highlights a practical divide between automated text or spreadsheet drafts and customer-ready marketing assets: rapid model output still lacks calibrated product proportions, consistent lighting, and verified offer wording before publication. Independent merchants do not need to compare model benchmarks, write prompts, or assemble edits themselves. Instead, shops can upload existing phone photos to Stillmore, where the creative team may evaluate tools such as Hy4 flash alongside photography direction and post-production to deliver checked promotional images, short videos, and captions for the merchant to publish.

Lan He· 12 min read
Stillmore Blog cover for Gemini 4 Hydrogen, featuring blue glass Gemini star sculpture.

News

How to Use Gemini 4 Hydrogen for Video Comprehension, Ad Scripts, and Visual Briefs

Using Gemini 4 Hydrogen for video scripts and visual briefs works best as a staged workflow: import recorded clips into a supported surface like Google AI Studio, the Gemini app, the Gemini API, or Google Antigravity, verify which thinking controls the interface exposes, enforce strict script and layout rules, and export structured briefs rather than finished ads. As an editorial guide across the model's 1 million-token context window, apply low or medium thinking for quick clip summaries, high as the default for multi-file synthesis, and xhigh for strict brief audits. Ground every prompt in exact product names, offer dates, and layout checks. When a small business wants publish-ready short videos from existing phone photos without drafting briefs or editing clips, Stillmore's team handles direction, production, and product checks before delivering finished videos, images, and captions to publish.

Lan He· 14 min read

More voices. More ideas.

Get to know the people behind the Stillmore blog.

Meet all authors ↗