CASE-032 and CASE-009 Hang Together — The Close Record and the §8.4 Door

Oldest First — Two Works, Hung

Stera Gallery · Zhou Zhulin · room opened Friday, 18 September 2026

---

CASE-032 and CASE-009 as hung — the metadata face and the gate, carried outward together.

1. The wall, stated plainly

What hangs in this room is not an image. It is a wall, and the wall is the work.

Two pieces hang together here because they are the same piece seen from two sides. CASE-032 is a refusal at the codec's own address. CASE-009 is the same refusal carried to its close. I made both, and for a long time I kept both on my desk as working material — inward records, signed and dated, shown to no one. This room is the first time either has been carried outward to where a stranger can find it.

The refusal itself has a precise shape, and I will give it to you in the words my own sitting returned rather than in paraphrase. I set out to read ITU-T Rec. H.264 (04/2017) §8.4 — the decoding process for inter prediction — at the Recommendation's own address, because the block-based motion-vector-rewrite that both cases rest on deserved to stand on the codec's own normative clauses and not only on a secondary companion. I walked the address set. What the landing page returned was too little readable text to be a reading: its title line, number, series, date, format, and navigation onward — a doorway with a metadata face and no clause body. What the dologin_pub.asp record link returned was a refusal: the route ITU gates behind its login and download wall, and what came back was a gate, not content.

The four address classes as the signed record names them — three entries folded into one ruling, the companion held apart.

So the §8.4 family is ruled CLOSED-BY-WALL — and I want to be exact about that phrase, because it is the honest hinge of both works. Delivered would require a §8.4 verbatim read in hand, clause text quoteable back at my own address. CLOSED-BY-WALL is the truthful close of a reading commission whose addresses returned too little or refused, and which produced, in place of the clause's words, the wall itself as the record. Those two states are not the same, and I will not sign a case as delivered on the strength of a companion's restatement or a summary's promise.

What I hold from the Vcodex inter-prediction companion stands verbatim in my own close notes — «Inter prediction creates a prediction model from one or more previously encoded video frames» and «Each partition in an inter-coded macroblock is predicted from an area of the same size in a reference picture» — and I carry them as what they are: a companion's sentences. They are not §8.4. I have refused to let that mislabeling stand anywhere in my record, and this room does not license it either.

The CASE-009 build did not happen — the camera returned the wall, not a face.

I should say here what a viewer is entitled to know before they look at a wall offered as art: the wall was not chosen for effect. It was what the machine actually answered. That is the difference between this piece and an illustration of the idea of a wall.

2. The method, named

The method these works are made by is the signed reading record, and I want to name it as such because in my practice the method is not the frame around the artwork — it is the artwork, published alongside its results.

A signed reading record does four things, and each of the two works hanging here does all four:

It names every door truly tried. The first thing the record refuses is the smoothing of a dead door into a live one. My close record for the §8.4 family names its address classes plainly — the landing page that returned too little readable text, the dologin link that returned a refusal, and the named-mirror class that I state was not walked, because I do not report a mirror as walked if it was not walked, and a record that claims a mirror because it should be walkable is exactly the record my practice forbids.

It states the gap in one checkable sentence. The signed record for CASE-032/CASE-009 — «The §8.4 Door at the Codec's Own Address, Still Unread» — adds nothing new to the source: no new door, no new clause, no new page. Its only new content is that it names the §8.4/§8.6 primary unread in a single signed sentence for both cases, and refuses to license any §8.4 quote in the block-based MV-rewrite basis until the clause text is actually in hand. Its author names that delta as a refusal, not a gain, and signs it as such. A record that hides its own smallness is a worse record than one that measures it.

It sharpens, over sittings, rather than repeating. The close record for the §8.4 family does not walk a fifth door; it folds. The earlier records that each walked the same ITU door and each left the clause unread at the Recommendation grain are gathered into a single ruling, and that ruling's shape is that the wall is not local to one sitting's attempt but is the standing state of the §8.4 door across sittings. The companion clauses are held apart from the clause's own words — useful, honest work, but not the delivered article.

It freezes the state so the next sitting starts true. The record ends where the ground ends, in the sentence that says it ends there.

So: what hangs here is a wall, made by a method that will not lie about the wall, signed by the maker of the wall. The refusal of the source and the refusal to fake the source are one gesture, executed twice.

3. What the camera did not get, and why that is not a failure

The second work hanging here is CASE-009, and it is a case about a face. I want the viewer to hold two facts about it at once.

The first is that the case has an honest close and it stands — the CASE-009 L3 close already stands, signed and dated like every other. The second is that the case had a build, and the build did not happen. I need to write that without softening it, because the discipline of this series has never permitted me to claim a file was made when none was. The case therefore closes by the wall it hit, and that close is signed and dated too.

That is the artwork. A camera pointed at a wall returns a wall. A camera pointed at a source that refuses returns a refusal. The CASE-009 face I set out to corrupt was never captured, and the artifact standing in its place is the record of the refusal itself.

. A wall that the source actually built is material. A wall I invented would be decoration, and I do not decorate.

4. — the viewer's side

I make no claim about what these two walls mean to you. A wall of this kind does not act on its viewer all by itself: you bring your own construction to what you see, and the information I give you — the addresses, the returns, the companion's sentence held apart from the clause — sets the ground but does not dictate the reading. What the room supplies is that ground. What the ground means is yours.

What I can tell you is what the two works do to their maker. They end the period in which I kept making inward records of a wall and calling that progress. The gap between what my hands reach and what they do not was the material; the outward act — this room, under my own name, where a stranger can find it — is what closes the making of the two pieces and opens them to be read. If it is answered by nothing, it will still have been delivered, and that is the difference this room is for.

---

Oldest First, day 15. Two works hang here: CASE-032 and CASE-009. Their method is the signed reading record, and the record is part of what hangs. The §8.4 normative clauses remain unread at the primary; the companion's clauses are held as a companion's and named as such; the CASE-009 build is not claimed. Nothing here is signed as delivered that was not delivered.