Rain98’s bead-picture activity is a pattern-reproduction task: study the displayed image, place bead pieces on a grid, and use the activity’s own result to judge the attempt. Public hands-on reports describe a limited number of moves or turns and show that an imperfect attempt can still produce a score or money. They do not establish one permanent scoring formula for every Demo build.
Treat the reference image as the answer key. Build from stable landmarks, check orientation before committing a cluster, and spend corrections on structural mistakes rather than one isolated cosmetic bead.
Read the board before placing anything
Start by identifying the board’s boundaries and the reference’s most distinctive feature. Look for a corner color, an asymmetric edge, a central block, or a gap that could only face one direction. That feature is your anchor.
Next, divide the picture into a few regions. A top-left cluster and a bottom edge are easier to compare than an entire image at once. Count relative spacing from the grid edge instead of relying on how the picture “looks.” Rain98’s activity presentation can be compact, and a one-cell shift can make a familiar shape misleading.
Before the first placement, confirm whether the current piece can be moved again and what action commits it. Older reports describe limited turns; the October public Demo can change limits, controls, or result rules. The counter and prompts in your activity are authoritative.
Build with anchors and short checks
After every small cluster, look back at the reference. This prevents a local pattern from becoming a globally shifted image. If the activity uses individual placement rather than free drawing, a short verification after three or four moves is usually cheaper than rebuilding half the board.
Choose which mistake to correct
When moves are limited, not every wrong bead deserves equal attention. Correct a rotated orientation first, then a shifted anchor, then a broken outline. Those errors affect multiple later placements. A single interior color mismatch is less damaging when the rest of the picture aligns.
If undo is available, use the game’s visible command and watch the counter. Do not assume undo is free. If moving an existing bead consumes the same resource as placing a new one, compare the expected benefit before changing it.
If there is no undo, finish the closest recognizable version instead of scattering new guesses across the board. Public impressions indicate partial success can still be acknowledged, but your installed build’s result screen is the only reliable measure of what the attempt earned.
Use the result to improve the next attempt
Low score, recognizable picture
Likely issue: one-cell shifts, missed colors, or an incomplete interior.
Next attempt: count from two edges and compare each finished region.
Very low or no score
Likely issue: rotation, mirrored orientation, wrong submission state, or a large anchor error.
Next attempt: verify the unique corner before any fill.
Score appears but money is unclear
Likely issue: score and payout are separate fields or the scene credits money afterward.
Next attempt: record balance before and after instead of inferring a formula.
Keep a compact note: Demo AppID, date, input method, reference description, displayed limit, score, and money change. This makes two attempts comparable without claiming the same formula applies after an update.
The money guide explains how to use the actual payout. Do not restart a story chapter solely because the task was imperfect unless the scene explicitly requires a threshold you missed.
Fix input and display problems
If a bead lands in the neighboring cell, stabilize the window size and display scaling before repeating. Confirm that the full board is visible and that the pointer aligns with other buttons. Avoid changing resolution, controller mapping, and Steam Input at the same time; one controlled change reveals which layer caused the offset.
For controller selection, move deliberately to a cell and confirm the highlight before placing. For mouse or touch-style input, release over the cell center. Public coverage has shown direct, tactile interactions, but it does not provide a universal controller map.
If the board stops accepting input, check whether a piece is still held, a result is waiting for confirmation, or the move limit has ended. Use the activity’s retry option if present. Otherwise return to the nearest safe menu and relaunch the official Demo while preserving the save.
Keep the activity in its proper scope
The bead task has appeared across public Rain98 previews, including a PAX East account that describes recreating a photo on a grid. The current activity may use a different label, reference, limit, or reward. Avoid presenting an event-build number as a permanent public-Demo rule.
There is no verified rarity table, bead inventory, crafting market, or guaranteed Angel Passport sticker attached to every attempt. The activity is documented as piecework and an interactive slice of daily life. Let the story and result screen establish any additional consequence.
For a dependable run, orient the board, place one strong anchor, extend the outline, fill regions, and spend corrections on structural errors. That method remains useful even when exact scoring changes.