← All Framer updates
Framer Agents 3D Transforms — Rotate, Perspective, and Depth You Can Still Edit on Canvas cover
Framer update · 7 min read · 7 tags · By Abdus Salam

Framer Agents 3D Transforms — Rotate, Perspective, and Depth You Can Still Edit on Canvas

Agents can wire Rotate XYZ, Perspective, Origin, Depth, Backface, and the rest of Framer's 3D stack into real Frames and Effects — then you keep tweaking every layer on Canvas like a normal design file.

update agents ai 3d animation canvas design

TL;DR — On September 1, 2026, Framer added full 3D Transform support to AgentsRotate XYZ, Perspective, Origin, Depth, Backface, and the rest of the property stack Agents need for dimensional effects. Framer’s differentiator: 3D layers stay selectable and editable on Canvas after the Agent finishes, built from normal Frames and Effects rather than a locked render. Framer’s showcase logo animation used Opus 5 Fast with light reasoning.

If you have been hand-keyframing perspective on logo marks while wishing Agents could reach the same panel, this release closes that gap. The sections below translate Framer’s announcement into what to prompt, which model settings Framer demonstrated, and how to keep 3D work remix-friendly on client files.

Agents 3D transforms at a glance

Property clusterWhat Agents can setPractical use
Rotate XYZRotation on all axesLogo flips, card tilts, subtle hero parallax
PerspectiveScene depth and vanishing readProduct mockups that feel dimensional, not flat
OriginTransform pivot pointSpins that hinge from a corner or center mark
Depth / Z translationLayer separation in 3D spaceStacked UI cards, floating badges, layered type
BackfaceVisibility of reverse sidesFlip animations and reveal loops
Canvas editabilityLayers remain selectable post-AgentClient revisions without rerunning the whole prompt
Prior releaseHow 3D Agent support extends it
Reasoning and Fast ModeFramer’s demo pairs Opus 5 Fast + light reasoning for creative 3D iteration
Opus 5 in AgentsOpus-class planning quality when 3D must align across nav-heavy sites
Logo ShadersShader-driven logo polish plus native 3D transforms on Frames when you need both
July canvas capabilitiesAgents already touch Page Effects and overlays; 3D transforms add another motion lever

What Agents 3D transforms change

1. Full 3D Transform coverage in Agent prompts

Before September 1, dimensional work often meant manual property edits or shader shortcuts while Agents handled flatter layout tasks. Framer now documents all 3D Transform properties as in scope for Agents, including the axes and scene controls designers already know from the Canvas inspector.

That matters for creative agencies and product marketing teams who want Agent speed on logo loops, hero cards, and launch teasers without surrendering fine control after the first pass.

2. Editable Canvas layers, not a black box

Framer stresses a workflow detail competitors often skip: 3D layers remain selectable and customizable on Canvas. The September 1 logo animation example is built from fully editable Frames and Effects. You can nudge Origin after the Agent run, duplicate a layer for a second variant, or restyle fills without regenerating from scratch.

For retainers, that editability is the difference between shipping and opening a ticket to rerun the Agent.

3. Proof point: logo animation with Opus 5 Fast

Framer’s release page shows a logo treatment built in Framer using Agents. The documented stack: Opus 5 Fast with light reasoning. That is a signal for scoped creative prompts (one mark, one hero, one component variant), not a mandate for every 3D task.

When a 3D pass spans multiple routes or shared components, consider higher reasoning on a branch and compare staging quality before you merge.

How 3D transforms fit the Agents stack

Framer’s 2026 Agent arc moved from breadth (July canvas capabilities) to control (August reasoning and Fast Mode) to creative depth (September 3D transforms):

  1. July 29 canvas expansion — Page Effects, overlays, CMS video, richer summaries.
  2. August 12 reasoning + Fast Mode — planning depth and Opus 5 acceleration in the picker.
  3. September 1 3D transforms — Agents can wire native dimensional properties on Frames and Effects.

Read together, Framer is pushing Agents from layout assistant toward motion-capable co-designer, while keeping Canvas as the source of truth you can still hand-edit.

How to use Agents for 3D transforms

  1. Branch the project — never benchmark 3D Agent output on production main.
  2. Scope one focal treatment — logo lockup, hero product stack, or pricing card tilt; avoid “make everything 3D” prompts.
  3. Start with Opus 5 Fast + light reasoning for creative loops, matching Framer’s documented demo pairing.
  4. Name the properties you care about — e.g. “Rotate Y on hover, Perspective 1200, Origin bottom center” — so the Agent does not guess pivots.
  5. Inspect layers on Canvas after the run; tweak Origin and Depth manually before requesting a second Agent pass.
  6. Stage and performance-check — preview on mobile; reduce simultaneous animated 3D layers above the fold if scroll stutters.
  7. Merge only after human QA — Agent depth is fast to iterate, but client brands still deserve a final eye on motion timing.

Three workflow recipes to try this week

Recipe 1

Logo mark tilt loop at light reasoning

A client wants a subtle 3D flip on their SVG logo for a launch teaser — one component, no sitemap changes.

  1. Branch, select Opus 5 Fast with light reasoning (Framer's documented demo pairing).
  2. Prompt: add a slow Rotate Y loop on the logo Frame only — set Perspective and Origin bottom-center; do not change nav or footer.
  3. Scrub Origin and Depth on Canvas after the run; merge when the loop reads clean at mobile width.
Recipe 2

Hero product card stack with Depth

A SaaS homepage needs three feature cards separated in Z space with a shared perspective scene.

  1. Branch and scope the hero section only — reference existing card Frames by name if possible.
  2. Prompt: apply shared Perspective to the hero container; translate each card on Z with slight Rotate X; preserve copy and CMS bindings.
  3. Review backface visibility and shadow overlap in staging before merging to main.
Recipe 3

Compare Agent 3D vs manual before client review

You already hand-built a 3D hero; the client asks for a faster variant for A/B staging.

  1. Duplicate the hero Frame group on a branch; run a light-reasoning Agent pass with explicit Rotate and Perspective targets.
  2. Toggle between manual and Agent variants in staging; note which layers stayed editable without rerunning the prompt.
  3. Ship the variant that needs fewer manual Origin fixes — document the winning prompt for the retainer playbook.

Framer templates for 3D Agent experiments

These remix-friendly files give Agents strong hero and brand surfaces without inventing a sitemap from scratch. Open one on a branch when you want to stress-test Perspective and Rotate XYZ on real marketing layout.

Browse more in the template directory or start from the free Framer templates hub when you want a different niche baseline.

Official Framer resources worth bookmarking

More reading on yoframer

The bottom line

Framer Agents gained native 3D Transform support on September 1Rotate XYZ, Perspective, Origin, Depth, Backface, and the rest of the dimensional stack, with Canvas layers you can still edit after the run. Framer’s demo pairs Opus 5 Fast and light reasoning for scoped creative loops; scale up reasoning when 3D must align across routes. Branch first, scope one focal treatment, hand-tweak Origin on Canvas, then merge when staging performance holds.

Get started free on Framer →

Frequently Asked Questions

Quick answers about Framer's September 1, 2026 Agents update — full 3D Transform support in Agent prompts, which properties are in scope, why Canvas editability matters, recommended model settings, and where to read Framer's official release notes.

What 3D Transform properties can Framer Agents edit now?

Framer published the feature on September 1, 2026 at https://www.framer.com/updates/agent-3d-transforms. Agents now support the full 3D Transform stack Framer documents on Canvas, including Rotate XYZ, Perspective, Origin, Depth, Backface, and related properties. The release positions Agents for creative effects you iterate quickly, not just flat layout passes.

Can I still select and edit 3D layers on the Framer Canvas after an Agent run?

Yes. Framer highlights this as unique to its workflow: 3D layers remain selectable and customizable on Canvas after Agents build them. The September 1 example logo animation is built from fully editable Frames and Effects, not a baked export you cannot touch. That makes Agent-assisted 3D work practical for client revisions.

Which Agent model did Framer use for the 3D logo animation example?

Framer says the showcase logo animation on https://www.framer.com/updates/agent-3d-transforms was made with Opus 5 Fast and light reasoning. That pairing favors quick creative iteration over deep multi-route planning. See /framer-updates/framer-opus-5-agents-model-update/ and /framer-updates/framer-agents-reasoning-fast-mode-update/ for model and control context.

Should I use light or higher reasoning for 3D Transform Agent prompts?

Framer's own demo uses light reasoning with Opus 5 Fast for a focused logo animation pass. Start light when the prompt scopes one hero or logo treatment. Reach for higher reasoning only when 3D transforms must align across multiple routes or component variants. Always run experiments on a branch first — see /framer-updates/framer-agents-branching-community-update/.

How is Agent 3D support different from Framer Logo Shaders or Video Shaders?

Logo Shaders and Video Shaders (covered at /framer-updates/logo-shaders/ and /framer-updates/framer-video-shaders-update/) apply shader effects to media. Agent 3D Transforms operate on Framer's native 3D Transform properties for Frames and Effects on Canvas. You can combine both on a project, but they solve different problems: dimensional layer transforms versus shader-driven media treatment.

Will heavy 3D Transform stacks hurt published site performance?

Framer ships 3D transforms as web-ready Canvas output, but depth, perspective, and animated rotations still add paint and compositing work in the browser. Treat Agent-built 3D like any motion-heavy hero: test on staging, limit simultaneous animated 3D layers above the fold, and prefer one focal 3D treatment per viewport rather than stacking many independent transforms.

Where can I read Framer's official Agents 3D Transforms release notes?

The canonical post is at https://www.framer.com/updates/agent-3d-transforms. Browse adjacent Agent releases at https://www.framer.com/updates and our editorial hub at /framer-updates/.

Keep exploring

Browse every free Framer template in the free Framer templates hub and full directory, or open the full Framer updates index.

More updates