Entry 100

The magic portal fades out as the monsters come out: its objects at their own transparency

2026-09-25rendertest

Filescrates/piney-effect/src/portal.rscrates/piney-effect/examples/draw_probe.rstools/test_effect_draw_rs.pycrates/piney-game/src/session.rsdocs/engine/effects.md

The user saw that on a field an enemy portal did not vanish right after its monsters came out. It stayed up, full, and went away only about when the battle mode ended.

What the portal does

ccMagicCircle::main was already ported and checked in eemu (test_battle_spawn_rs's circle_main and frame), and it runs right in the port.

Its acts:

• Act 0: it wakes when Kite comes within 3000.
• Act 1: after 32 frames it switches to the opening, ANM_xmagcir2.
• Act 2: after 22 frames the monsters come out (entryCircleObject).
• Act 3: it waits for the opening's 111 frames to end.
• Act 4: 66 frames later the entry control deletes it.

Logged on event 3's east portal: the monsters out at frame 181; the opening's end at 268; the portal deleted at 337; the battle mode on at 220 and still on after 364.

The opening's picture. Its Anime chunk fades the portal's objects through their alpha tracks (tools/anim.py pose):

• the ball flares to 0.98 at frame 30 and is gone by 60;
• the rings grow to twice their size, then shrink to nothing by 100;
• the sphere fades from 0.9 to 0 between 60 and 110.

ccAnm::Draw hands each object to ccObj::Draw at its own transparency times the anm's, and draws none at or below 1/128. So the portal fades out over the frames after the monsters appear. Its last frame drawn is the opening's 108th, about 86 frames after the monsters. Act 4 passes with only its last sparks.

The cause

portal::MagicCircle::render is piney-effect's draw of the portal, the one the runtime uses with the sparks. It drew every object at the portal's own transparency and never read the pose's alpha. So the sphere and rings stayed full until the entry was deleted, about when the battle mode ends.

piney-world's own draw of the circle (foe::Circle) did apply the objects' alphas; test_foe_rs checks it. But the runtime does not use it once the effects draw the portals.

The render now goes through nodes::anm. That gives each object its matrix and its own transparency, in ccAnmIndex order; the palette swap and morphs are kept.

Checked

• eemu: ModelsAgainstGame. tools/test_effect_draw_rs.py now runs the game's ccAnm::Draw / ccClump::Draw on XMAGCIR.CCS as well. The file is decoded by the game's own DecodeSetup, and draw_probe adds the same file after the effect files, as the field does. The cases:

  • the clump;
  • ANM_xmagcir1 and ANM_xmagcir2, stepped 1 to 179 frames at 256, into the opening's held end.

  In all: 1,140 clump draws (896 models) and 750 animation draws (2,872 models). 70 of the draws are the portal's (107 models). 0 mismatches in models, order, transparency and lwMatrix (worst 2.3e-6 relative).

• A unit test. the_opening_fades_the_portal_out: the ball is drawn at the opening's frame 29 and gone by 70; the last frame with anything drawn is between 100 and 110 (it is 108).

• Shots. portal_shots is an ignored test. Kite walks up to event 3's east portal, the camera reset behind him. A shot every 10 frames, from the wake to 40 frames after the deletion.

  • Before the fix (scratch/combat/portal/sheet_before.png): the full cage sphere through act 3 and act 4, gone at the deletion, with "BATTLE MODE OFF" 20 frames later.
  • After (portal_after/sheet_after.png): the goblin comes out and the sphere fades to a sliver and a glint. Nothing is left by act 4, the last sparks aside.

• The other checks. test_effect_rs and test_foe_rs pass. So do the workspace's tests, clippy, fmt and the docs check.