This is a tarot video. A reader looks at picture cards and tells a story about someone's life. In this version we let up to ten cards sit on the board at the same time, and we made the video bigger so the cards are easier to see. The voice you hear is the real reader's own voice. The cards and the words on the chalkboard were added on top so you can follow along. Fix on June 29: the reader said "Page of Wands" then quickly fixed it to "Page of Cups." The first version of the video drew the wrong card. Now the right card — Ten of Pentacles — sits in spot 5. Tap Grown-up mode for the normal notes.
| The bug | At 1:38 the reader says "Okay, let me get one more with this Page of Wands. I mean Page of Cups." The loose card detector caught "Page of Wands" before she corrected herself, so the prior render placed a phantom Page of Wands as slot 5, and the real next pull — Ten of Pentacles — got bumped to slot 6. |
| The fix | The three rolling-caption cues between 00:01:38 and 00:01:43 are patched from "Page of Wands" to "Page of Cups" so the detector reads what the reader actually meant. Audio is bit-identical (re-mux of the prior render via --reuse-audio), so sync is preserved. The real Page of Wands pull at 02:26 is untouched. |
| Net effect | Slot 5 is now Ten of Pentacles (correct), Page of Wands now appears at its real pull moment later in the spread, and one phantom card is dropped from the on-screen count. |
| Cards on screen at once | Up to 10 (was capped at 5). Denser spreads stay on the board longer before any card is evicted. |
| Video window | Bigger — the player breaks out to a 1400px hero frame instead of sitting inside the 880px text column. |
| Voice / audio | Unchanged — the exact creator source audio, re-muxed bit-for-bit from the v1 render (no re-download, no TTS). |
| Card art / chips / chalk | Same RWS cards, same K1 keyword chips, same 144 chalk annotations — only the on-screen slot count and the page player grew. |
Source KghYp2skGtI (uploaded 2026-06-23). The audio was re-muxed straight from the v1 source-audio render, so the voice you hear is the reader herself. Same source_audio_render.py pipeline, run with --max-slots 10.
yuv420p / tv / bt709 / +faststart Main-profile encode (gate E2).| First five cards | Page of Cups, The High Priestess, Eight of Wands, Wheel of Fortune, Ten of Pentacles (fix) |
| Slot-5 correction | The reader misspoke ("let me get one more with this Page of Wands. I mean Page of Cups"), so the prior render placed a phantom Page of Wands at slot 5 and pushed Ten of Pentacles to slot 6. This re-render drops the misspeech and seats Ten of Pentacles in its real slot. Page of Wands still appears later in the spread at the cue where she actually pulls it. |
| Total placements | 30 distinct cards across 786 transcript cues |
| Voice owner | Creator source-audio (re-muxed from the v1 render, no synthetic voice) |
| Render mode | SOURCE-AUDIO · --max-slots 10 |