An early request said the plan looked too small on desktop. After enlarging it, a later request said it was too large to see the selection controls and plan together. The statements sound opposite, but the purpose was consistent: visitors needed to choose a type and comfortably inspect the result.
The calculation therefore needed more than an image size. It needed the space remaining after the header, contact bar, title and selectors. The repair notes and final source describe measuring those areas to set the desktop plan height. I changed the target from filling the screen with a drawing to making selection and reading work together.
Not all viewport height belongs to the drawing
Two browser windows with the same height can offer different plan space when a title wraps or a fixed contact bar grows. The occupied header position is also absent from a simple image-width calculation. Giving the drawing the full viewport height can push its controls or important material below the visible area. I account for the necessary controls first.
The final source measures the header and fixed lower area, the title, both selector groups and detail links. It distinguishes hidden elements from visible occupied space and includes layout-specific gaps and a minimum height. Those values belong to the inspected page structure. They are not a universal formula to copy into every website.
Keep an overview and a route to read the original detail
Fitting a plan into one view can make small labels harder to read. Cropping the bottom would lose information instead of solve the reading problem. The historical repair record pairs an uncropped fit with an original-size viewing link. Comparing the overall arrangement and inspecting small details become two supported views rather than competing demands on one image size.
Several supplementary drawings require more than a successful first image. Each drawing’s relationship to controls and separate area-information material matters. I should not expand one image’s success into a claim about every sheet. The reading route needs to survive differences in material length and purpose.
Measure again when the layout changes
A window resize or a late-loading image can invalidate the first measurement. The final repair code schedules a fresh calculation on resize, load, image load and selection changes, and observes some control sizes with ResizeObserver. Repeated requests are gathered into a frame-level calculation schedule rather than independently repeated without coordination.
I have inspected those recalculation connections in the archived source. This does not freshly test every font and image-loading combination on the original services. Actual application should include wrapped titles, short browser heights and different selected aspect ratios. A resize handler’s presence and an observed satisfactory result are separate kinds of evidence.
A detail-page landing position should include the way to choose
Scrolling a detail page directly to the drawing can leave its selector above the viewport. The historical record changes the landing point to the start of the selection area. A visible image is insufficient if the way to compare it with another type has disappeared. Scroll position is part of using the material, not merely a decorative transition.
Desktop and mobile need not repeat the same control arrangement. The user kept one upper selector on mobile, making its connection to the drawing especially important. Rather than shrink the large desktop arrangement indiscriminately, I check the actual remaining operating position. Different screens require explicit choices about what stays available and what becomes smaller.
Explain the conditions behind a one-view target
The historical desktop calculation applies above a particular width boundary. A very short viewport or unusually tall title and controls may still require scrolling because the drawing retains a minimum height. That minimum is not a promise that arbitrary material fits every screen. Screen conditions and material conditions should accompany the design target.
This article is supported by original requests, repair source and historical local connection records. Those records separately identify a stage before server deployment and assign visual review to the user; they do not prove that every real screen review was completed. The method I want to leave is how to calculate the place for information and its controls, rather than a number that simply makes every plan larger.