TURN RECORD: THE ERROR CHAIN IS ENTERED AT 0x1BA30 BECAUSE rp+0x48 TESTS ZERO.
HYPOTHESIS (LOW BYTE OF THE ID) TESTED AND REJECTED.

Trace of the composition path:
   r9F2  a8=00000000        ; the call at 0x1B9EC returns 0
   ERRCHAIN_1BA30           ; <-- 0x1B884 JREQ 0x1BA30 taken (rp+0x48 == 0)
   ERR_1BAA6 / ERR_1BAC6 / ERR_1BB96
   DRIVER_RET a8=00000002   ; the chain reports 2 (0x1BA68 MOVK 0x2,A11)

The tests:
   0x1B878 A5 = @0xa6000
   0x1B87E MOVE *A5(0x48),A11,0x0   ; NOTE: SETF 8 -- an 8-BIT read of byte 9
   0x1B884 JREQ 0x1BA30             ; zero -> the BadValue error chain
and the caller had stored a 32-bit value there:
   0x63080 MOVE *A9(0x20),*A8(0x48) ; rp+0x48 <- request w0 = photo_id
So an 8-bit test against the low byte of the photomap id looked plausible.

TESTED: created the photomap with an id whose low byte is non-zero
   MAP = 00400B01  (previously 00400B00)
   full sequence: CreatePhoto(w6=1) -> SetTransport(mode=3, org=1) -> PutTile
     -> Export(21)
   RESULT: export = err2,err2 ; white vs black framebuffer AE = 0
So the low-byte theory is WRONG: the id's low byte is not what that flag reads,
or rp is a table entry (0xa6000 + index*32) whose 0x48 field is set elsewhere.

STATE (unchanged and reproducible)
   import: element+0x28 = 0xFF white / 0x00 black ; QueryResource comp_COUNT=1
   export: err2 in every configuration tried; window never changes
   viewer: xshow (core X) is the only thing that has displayed an image

STATUS.  NO client-chosen XIE image.
