What is the problem you are trying to solve?

In QField, multi-select Value Relation dropdowns successfully display and group checked items when reviewing a previously saved/plotted feature. However, this behavior breaks when creating a brand-new feature while utilizing the "Remember Last Value" pin.

When a field worker creates a new feature, the data carried over from the previous feature is active in the background cache, but QField treats the new record as completely blank visually. The carried-over items remain scattered throughout long lookup lists (e.g., a list of mixed species in forestry or biodiversity monitoring). Users are forced to scroll extensively through the entire list just to see what data has carried over or is already checked. In mixed environments (like woodlands containing a combination of conifer, broadleaf, native, and non-native trees simultaneously), splitting the data into cascading parent/child dropdowns isn't a viable workaround.

What is your proposed solution?

Update the user interface behavior within multi-select lookups so that carried-over data from the "Remember Last Value" cache is treated identically to pre-existing row data. When a new feature form opens, these active selections should automatically pin to the absolute top of the list view.

This enhancement should:

• Create a visual partition on new features: Group all carried-over/active selections at the very top of the widget immediately upon form creation, with the remaining unchecked options listed cleanly below them.

• Preserve project sorting: Maintain the project's underlying sort order (such as alphabetical order or custom priority columns configured in QGIS Desktop) within those two separate blocks.

• Update dynamically: Refresh the list order cleanly every time the dropdown widget is tapped open.

What is the value of this feature?

This fixes a major UX consistency gap between editing old features and creating new ones. It aligns QField with modern mobile UX standards, dramatically reduces form scrolling during rapid data collection, and eliminates critical data validation errors caused by carried-over selections being hidden deep within large lookup datasets.