top of page

๐—ช๐—ฅ๐—ข๐—ก๐—š ๐—˜๐—ก๐—— ๐—ข๐—™ ๐—ง๐—›๐—˜ ๐—ฆ๐—ง๐—œ๐—–๐—ž ๐—˜๐˜ƒ๐—ฒ๐—ฟ๐˜†๐—ผ๐—ป๐—ฒ'๐˜€ ๐—ฒ๐˜…๐—ฐ๐—ถ๐˜๐—ฒ๐—ฑ ๐—ฎ๐—ฏ๐—ผ๐˜‚๐˜ ๐—ฐ๐—ผ๐—ป๐˜๐—ฟ๐—ผ๐—น๐—น๐—ถ๐—ป๐—ด ๐—ฃ๐—ต๐—ผ๐˜๐—ผ๐˜€๐—ต๐—ผ๐—ฝ ๐˜„๐—ถ๐˜๐—ต ๐—–๐—น๐—ฎ๐˜‚๐—ฑ๐—ฒ ๐—–๐—ผ๐—ฑ๐—ฒ ๐˜ƒ๐—ถ๐—ฎ ๐— ๐—–๐—ฃ.

  • Writer: candyandgrim
    candyandgrim
  • Apr 19
  • 2 min read

Watching AI rename layers and batch-resize 40 images from a terminal feels like magic.


Except I've been automating Photoshop with Actions and Droplets since 2008.



That's not a dismissalโ€”the layer management piece is genuinely new, and MCP as a protocol matters. But the excitement is pointed at a problem that's largely already solved.



Here's where I'd actually want it:



Adobeย ๐—–๐—ต๐—ฎ๐—ฟ๐—ฎ๐—ฐ๐˜๐—ฒ๐—ฟ ๐—”๐—ป๐—ถ๐—บ๐—ฎ๐˜๐—ผ๐—ฟ. The image-to-puppet pipeline isn't iterative...it's laborious and unforgiving. One mis-tagged layer and the rig breaks. An MCP that could scaffold a puppet setup correctly, first time? Hours back, every session.



๐—”๐—ณ๐˜๐—ฒ๐—ฟ ๐—˜๐—ณ๐—ณ๐—ฒ๐—ฐ๐˜๐˜€. Plugins already handle keyframe automation on many levels: Mister Horse, Motion Tools, Limber, Newton, Joysticks & Sliders. The unsolved problem is different. Bounce, roll, float, walk cycle, IK, head turn, morphโ€”these take genuine skill and patience because getting them right requires feel, not just execution. That's where AI-assisted iteration earns its place.



๐—œ๐—ป๐——๐—ฒ๐˜€๐—ถ๐—ด๐—ป. The layout engine is excellent. The interactive and animated layer for ePub, interactive PDF, and published content is one of the most painful creative experiences in the entire Adobe suite. A print tool with interactive features grafted on and largely left to rot.



Maxonย ๐—–๐—ถ๐—ป๐—ฒ๐—บ๐—ฎ ๐Ÿฐ๐—— / ๐—ซ-๐—ฃ๐—ฎ๐—ฟ๐˜๐—ถ๐—ฐ๐—น๐—ฒ๐˜€. Physics simulation is a different kind of pain entirely. You spend 80% of your time nudging parameters, re-simulating, watching it break, nudging again. An AI that could hold your intent and iterate toward it isn't a nice-to-have. It's transformative.



And then there's ๐—ฝ๐—ฎ๐—ถ๐—ฑ ๐—ฎ๐—ฑ ๐—ฝ๐—ฟ๐—ผ๐—ฑ๐˜‚๐—ฐ๐˜๐—ถ๐—ผ๐—ปโ€”30 variants, multiple ratios, multiple resolutions. Smart teams templatise inย Figmaย or After Effects. But underneath the template, the judgment calls still happen manually every time. When does the H2 disappear? When is the CTA too long for the container? Combine that with inconsistent GIF output, HTML5 as a discipline in its own right, and assets destined for a dozen different platformsโ€”and this might be the one place where even MCP feels like a stretch. Not yet.



The butt of the spear is aimed at a solved problem. The sharp end is somewhere nobody's looking yet.



ย 
ย 
ย 

Comments


bottom of page