Guillermo

creative director @islaintelpr. recovering motion designer. testing every creative ai tool i can get my hands on

[Guillermo @gmartidesign]

thinking about every keyframe i ever set by hand in After Effects

[Higgsfield AI 🧩 @higgsfield_ai]

We vibe-coded these motion graphics with Claude Motion + Higgsfield Katana. All code, no After Effects.

This is insane.

[Guillermo @gmartidesign]

@AIWarper S.I. rebrand is a bold move. keeping the same feed or wiping the slate

[Guillermo @gmartidesign]

@nizamdesign clean cards. are these for a real product or exploration

[Guillermo @gmartidesign]

@EHuanglu crazy ads with a free prompt is generous. how much of the look came from the prompt vs the model defaults

[Guillermo @gmartidesign]

@nizamdesign design sense deciding what not to ship is still the job. vibe coding just made the bad ideas faster lol

[Guillermo @gmartidesign]

@techhalla "wait is this ai" is the only compliment that matters rn. was the first frame the hard part or the motion

[Guillermo @gmartidesign]

@CharaspowerAI time-stop with one character still moving is a nasty stress test. how many takes before the mall didnt melt

[Guillermo @gmartidesign]

the #1 image editor on the leaderboard leaves only about a fifth of the picture untouched on every edit

leaderboards score one edit. client feedback rounds get you to edit 14 lol

[Artificial Analysis @ArtificialAnlys]

We gave four frontier image editing models the same photo and 30 edits in a row: Ideogram 4.5 keeps most of the room intact, while the others drift, GPT Image 2.5 Sunburst most visibly.

We've seen some interesting demos of multi-turn editing consistency from the latest image editing models, so we ran our own test: GPT Image 2.5 Sunburst, #1 on our Image Editing leaderboard, against Ideogram 4.5, FLUX 3 and Nano Banana 2.1. Our leaderboard scores single edits; this tests what happens when 30 consecutive changes stack up, with each model editing its own previous output through a 30-step real estate staging sequence: light the fire, add a sofa, repaint the walls, swap day for twilight, and more.

Why are the final results so different?

@ideogram_ai's Ideogram 4.5 and @bfl_ml's FLUX 3 edit locally. On small edits like adding a vase of tulips, we measured that they left 95% or more of the image essentially untouched. GPT Image 2.5 (Sunburst) re-renders most of the entire image on every edit, leaving only about a fifth of the image unchanged, so small shifts in colour and detail compound over turns. Nano Banana 2.1 sits in between: its edits stay local, but the rest of the image shifts slightly and gradually darkens.

[Guillermo @gmartidesign]

meta finally ships a frontier agent and im mostly thinking about how long until this is just sitting in after effects babysitting renders

[Mark Zuckerberg @finkd]

Introducing Muse Wonder, our new family of frontier models with state of the art performance for personal and professional agentic work.

[Guillermo @gmartidesign]

@rileybrown

[Guillermo @gmartidesign]

@nizamdesign kitchen ops is such a good thing to design for. did the layout start from an ai screen or from scratch

[Guillermo @gmartidesign]

@henrydaubrez the lighting in this is doing heavy lifting. was it one model or a composite

[Guillermo @gmartidesign]

@nizamdesign ai consulting uis always drown people in options. what did you hide on purpose

[Guillermo @gmartidesign]

@adriankuleszo @designme testimonials that dont look like a wall of quotes are rare. what made you land on this layout

[Guillermo @gmartidesign]

@techhalla magnific one into seedance draft then hd is a smart cost split. how often does the draft version actually hold up

[Guillermo @gmartidesign]

also Synthesia Syren (upload footage, get video in your brands style) and Voyager, a desktop agent that drives Blender, DaVinci Resolve and After Effects for you

gonna be testing a bunch of these

[Guillermo @gmartidesign]

everything that dropped for creatives last week that actually matters 🧵

[Guillermo @gmartidesign]

images: Nano Banana 2.1 vs GPT Image 2.5 is the matchup everyones running. Magnific One is out. Midjourney is testing a thinking mode and has signups open for an MCP test