← All studio notes

Engineering & performance / DIJION

How to set a visual asset budget for mobile products

An inventory separating file size, resolution, memory and visible use so that visual quality can be assessed against device conditions.

A small image file does not necessarily use little memory at runtime. Package size, transfer time, decoded image memory and rendering cost are different concerns. A budget defined only as Compress all images leaves the problem being solved unclear.

Start with a usage inventory

Record where an asset appears, its displayed size and how many instances are visible together. A list thumbnail and a full-screen game background have different needs. If duplicated files exist, measure how reuse affects both package size and memory.

Scroll the table horizontally to see all columns.

An asset inventory to complete
FieldInformation to record
Identity and purposeScreen or scene using the asset
Source sizePixel dimensions and file size
DisplayRendered dimensions and density needs
RuntimeMeasured memory and loading behavior
ConcurrencyCopies visible at the same time
Quality boundaryUnacceptable blur or distortion

Separate the file from the decoded image

Android’s bitmap guidance discusses loading large images at a suitable target size. A compressed file and its in-memory representation differ. In a game engine, texture format, additional levels and platform support can change the cost. Avoid applying one file-size limit to every engine.

Choose a format for the actual use

Flat interface artwork, photographs and transparent characters need different decisions. Inspect edges, embedded text and transparent transitions against the real background. A smaller file does not always mean better loading or decoding on every device; evaluate support and visual quality together.

Connect the limit to an observed problem

Instead of inventing a universal kilobyte target, identify the assets required by the first screen and their behavior on the test device. If a list opens many full-resolution images at once, evaluate appropriate smaller versions. Investigate the lifetime of off-screen resources too.

Compare within the same scene

  1. Open the same screen with the same data and record starting conditions.
  2. Change the size or loading policy of one asset group.
  3. Inspect memory, loading and visible quality together.
  4. Leave and reopen the scene; resources that should be released must not accumulate.

This is a preparation plan, not a reported performance result. Record which device and quality boundary justified the budget. That makes visual quality and technical cost part of the same decision.

Sources