I’ll begin by setting out my conditions and disclaimers. I am RebiltG, first generation Tombsayer. The work documented here was done long before the discovery of the Tomb Index. At that time I operated in various headless squads, but I was — and remain — unaffiliated with any House or sect. What follows is an account of the process of decrypting the contents of a Tomb. I make no claims for the interoperability of the methods described. Technical detours will inevitably be required by the narrative, but understand that I cannot disclose the basic equipment and procedures necessary for file extraction. I believe that such barriers to entry are essential to maintain order in Tomb lore and prevent the mobbing of ’Sayer Discords by illiterates. Tomb lore is inscrutable by design, and initiation into its practices is self-selecting.

ORION was not the first Tomb whose opening I contributed to, but its contents are the best induction in wicked Tomb poetics that I know of. ORION is GENESIS VI. Early. It weighs 23.205.361, to give you some idea of what we had to contend with. Pearl violet disc body, grape green specular disc surface. Bracket with canonical brushed aluminium finish and patination. Glimpses of its eponymous constellation reflected on the glass overlay. It had been inactive on chain for over a year when it was opened.

After extraction, ORION appeared to contain only two .png files. These crude, monochrome images are reproduced here. An immediate squad consensus held that their orb-like forms must depict stars in the Orion constellation, and that they were likely part of a larger set of images, partially lost or damaged by the extraction process. Such uncertainty is mentally fraying for any ’Sayer, and a sense of futility pervaded the work from the outset. We clung to the belief that the markings on the surfaces of the orbs would be enough to identify the stars and their relative positions. That vector might then point us ... well, anywhere. The celestial map on ORION’s lower cartridge includes two prominent two-star clusters. The obvious alignment of one of these clusters — comprised of the binary system Pi5 Orionis and its neighbour Pi6 Orionis — with ORION’s centre plate led us to deem this the natural entry point. The two .pngs were subsequently referred to as Pi5 and Pi6. The squad split into two and work began.

Breakthroughs had been made in decrypting other Tomb content using multivariate image processing. The figures on SESSION, for instance, had been uncovered by running image tiles, apparently as small as 24 × 24 pixels, through genuine fractal enlargement algorithms and mosaicing the outputs. Our squads cycled Pi5 and Pi6 through every image-clarifying script repository to which we could get access. Tens of millions of derivative images were generated, from which subsets of hundreds of thousands were selected by machine-learning for eyeballing against astronomical reference images. After months of fruitless analysis, the whole effort was finally abandoned. It was lonely, punishing work, and at its unsatisfactory conclusion, those ’Sayers with any notoriety to bolster departed for other, potentially higher value, extraction sites. That option was not open to me. I stayed on ORION, connecting with a squad labouring aimlessly on deciphering metadata.

Most text strings recovered from Tombs, either from within discrete files, or their surrounding metadata, use 1337 and other ciphers to resist brute-force anagram and address resolution. Not only does this exponentially increase ’Sayer compute, it makes definitive resolution arguably impossible, since any given digit might properly belong to an alphanumeric address on chain, or be a substitute for a plain text glyph. Pi5 and Pi6’s original filenames were, respectively,

70_16h7h3_3r0ff_47wh_70_533_50_3476r_4_h4d.fp

70_my_1mbcl_4_741nm0un_45_1_17_533_1n_ndm1.rm

If these do contain complete, comprehensible messages, they remain hidden.

At this point I must humbly admit to the significance of my own contribution on ORION. In fact it was my journey from multivariate imaging to metadata that presaged the fateful discovery. During countless hours eyeballing Pi5 derivatives, I often manually post-processed ML-selected images with my own custom curves and profiles, in pursuit of greater definition. Days were spent hunched over the trackpad, forlornly plotting Bézier control points onto the unyielding face of Pi5’s histogram. Histograms are the software graphs we used to represent the pixel values in an image. They are human-legible, and experienced ’Sayers can use them to enhance certain features of an image without even referring to a live preview. It was only much later, as the metadata squad resorted to scrutinising ever more absurdly obscure fragments of text, that something prompted those histograms to resurface from my subconscious. It dawned on me that we were overlooking a whole class of visual metadata.

Having withstood an arsenal of brute-force computational incursions, and a relentless barrage of human attention, ORION gave up its secrets to a simple keyboard shortcut. In that extraordinary moment we saw that Pi5 and Pi6 share an identical histogram. Their whole celestial pretence was a decoy. We had taken it literally! ORION’s containing only two images was an economical device to establish the significance of their common characteristic. The second image was what ’Sayers call a confirmer, and ORION’s true cargo was the histogram itself. More than mere metadata, it was a meta-image. It directed us, unambiguously, to one of the most highly prized Tombs in GENESIS, a connection which would eventually enable the speculative deduction of the logic of the entire GENESIS lineage. I suppose that discovery, on ORION, is the source of whatever notoriety I might now claim as a ’Sayer.

This text is part of an occasional, unauthorised series set in an extended universe based on David Rudnick’s Tomb Series. Technical note: view the histograms of Pi5 and Pi6 in Adobe Photoshop to relive the most profound moment in the narrator’s career. In Photoshop, the Tombsayer’s preferred histogram view is found here: Image/Adjustments/Curves. The histogram presented there is generated from cached data. To refresh it, click the Options … button in the Curves window to trigger the display to update and reveal its truth. Other applications may present histogram data differently.