Auto strategy (compose.auto / code.auto)

April 4, 2026 · View on GitHub

Same API surface as scaled with prefixes asdp / ahdp / awdp / assp / … — see COMPOSE-API-CONVENTIONS.md §3.

What it is

Auto combines linear scaling with 300 dp reference up to 480 dp on the axis; beyond that it adds a logarithmic term so tablet growth is more restrained than a pure line, without being as “flat” as fluid’s plateaus.

Calculation used

With inv = 1/300, transition = 480, sensitivity = 0.4:

  • If dim ≤ 480: scale = dim × inv (same as linear scaled in that range).
  • If dim > 480: scale=(480×inv)+sensitivity×ln(1+(dim−480)×inv)\text{scale} = (480 \times \text{inv}) + \text{sensitivity} \times \text{ln}(1 + (\text{dim} − 480) \times \text{inv})
  • Result: base × scale; with a, multiply by the pre-computed aspect-ratio factor (DimenCache.currentAspectRatioMul); custom sensitivity uses 1+k×logNormalizedAr1 + \text{k} \times \text{logNormalizedAr}.

Implementation: calculateAutoDpCompose in DimenAutoDp.kt.

How to use

import com.appdimens.dynamic.compose.auto.asdp
import com.appdimens.dynamic.compose.auto.ahdp

Modifier.padding(16.asdp)

Prefixes: asdp, ahdp, awdp (+ Sp, px, inverters, a/i).

Code: com.appdimens.dynamic.code.auto.

Why use it

One token set that breathes on phones but does not double everything on tablets, without hand-tuned breakpoints.

When to use it

  • Apps that want less redesign between phone and tablet.
  • When scaled grows too much on tablet but power or fluid do not match the desired feel.

Advantages and trade-offs

  • Pros: continuous transition; familiar linear behavior on phones.
  • Cons: past 480 dp the curve changes character — validate long copy and grids there.

Prototype with scaled; if tablet is the only pain point, try auto on the same tokens before mixing many strategies on one screen.

Back to index