Gallery

Every image in gallery/, what it is, and which scene made it. One full-quality render per scene, committed as the diffable record of what that scene looks like. The same scenes as Looking Glass quilts are 25-40 MB release assets rather than repo content -- see renders/README.md.

Regenerate any one with make still-<name> (make help lists them all), or the whole set with make gallery.

Render each at its own declared aspect. POV-Ray maps right to image width and up to image height whatever pixel dimensions you ask for, so a mismatched frame stretches the picture with no warning:

File	Scene	Aspect	Size
museum.png	museum/museum.pov	1.778	1920x1080
museum_pg.png	museum/museum_pg.pov	1.25	1500x1200
museum_2026.png	museum/museum_2026.pov	1.778	1920x1080
bell_jar_bj.png	bell_jar/bj.pov	0.75	900x1200
bell_jar_bj_holo.png	bell_jar/bj_holo.pov	1.778	1920x1080
bell_jar_bj_holo_2026.png	bell_jar/bj_holo_2026.pov	1.778	1920x1080
bell_jar_bj_portrait.png	bell_jar/bj_portrait.pov	0.5625	1080x1920
bell_jar_bdna.png	bell_jar/bdna.pov	0.75	900x1200
bell_jar_yinyang.png	bell_jar/yinyang.pov	1.25	1500x1200
porin_3porin.png	porin/3porin.pov	1.778	1920x1080
lambda_main.png	lambda/lambda_main.pov	1.778	1920x1080
vitrine_*.png	vitrine/exhibit_*.pov	1.778	1920x1080

The PyVista renders have no right vector to match and are rendered at the shape their scene is composed for.

---

Museum

pov-scenes/museum/ -- "Eric's Science Museum," 1995-99. The room the README's hologram is rendered from; what's on the walls is cataloged in about-the-image.md.

museum.pov -- the canonical cut, and the one render_museum_hologram.py drives.

museum_pg.pov -- the 1999 cut, the last one. Adds the DNA cartoon mobile, the Risedronate exhibit and a tree outside the window.

museum_2026.pov -- the modern variant, composed for holographic output. The 1999 cut at 16:9 with three changes and nothing else: the oval mirror hangs where the Risedronate picture did (the scene's own object, on a branch that was already there), and both pedestals move inward to flank the alcove, the left one centered under Lambda Repressor. museum_pg.pov is untouched.

museum_970211.pov (the 1997 cut), museum_dark.pov and worldmap.pov are in the tree and render; none is carried here.

---

Bell jar -- DNA still lifes

pov-scenes/bell_jar/ -- the still lifes the museum's own pedestals were built from: B-DNA and Z-DNA under glass, on a marble stand, over sea and sky.

bj.pov -- "DNA Under Glass": B-DNA under the jar, sea and sky behind.

bj_holo.pov -- the same still life recomposed for a Looking Glass panel: 16:9, title lifted into the sky above the dome, signature moved out over open water at the lower right. Both sit on (or within a fraction of a pixel of) the focal plane, where a light-field display holds them sharp; in bj.pov they are camera-pinned overlays 70-74 units from the eye, in front of the scene itself. The lens opens 53.13 -> 55.32 degrees to make room for the title, which bj.pov's framing has none of.

bj_holo_2026.pov -- the same frame with the glass made real. The 1996 jar has no interior, so it refracts nothing: it is a surface that filters what passes through it, and once the jar was given a wall that filter applied twice wherever the wall was seen edge-on, drawing two grey outlines down each side of the dome. This one is ior 1.52 with a Fresnel reflection and a heavier wall, so the wall refracts rather than tints -- one bright band at the silhouette, a visible rim where the glass meets the pedestal, the horizon bending as it crosses the wall, and the duplex magnified the way a real jar magnifies what is under it. The signature reads '26. Everything else -- eye, lens, framing, lettering -- is bj_holo.pov untouched. See BJ_CRYSTAL in bell_jar.inc; it is off by default, so the other scenes sharing that include are unchanged.

bj_portrait.pov -- the 9:16 companion, for the tall panels. Same eye and the same treatment of the lettering; the lens is set by the pedestal's width, which is what overruns a narrow frame.

bdna.pov -- B-DNA alone on a checkerboard, no jar.

yinyang.pov -- B-DNA and Z-DNA side by side under two jars.

BJ_WALL = 0.06 in bell_jar/bell_jar.inc, the value chosen and the default. It was picked against 0.09, which is visible but chunky and distorts the duplex behind it; that comparison frame is no longer kept, and re-rendering bj.pov with BJ_WALL overridden reproduces it. Setting it to 0 restores the original zero-thickness surface, which refracts once and reads as a soap film.

---

Porin

pov-scenes/porin/ -- the beta-barrel membrane protein, as a ribbon cartoon over water under a rainbow.

3porin.pov -- self-contained apart from rainbow.inc; the scene the release .ini is written for.

3porin2.pov is in the tree but has no still: it is the stock POV-Ray "Basic Scene Example" template with #include "3porin.inc" appended, and 3porin.inc only #declares porin -- nothing instantiates it, so the scene renders sky and ground and no barrel.

---

Lambda repressor

pov-scenes/lambda/ -- the 1998 "Lambda Repressor" poster scene: the 1LMB PDB file, converted to a mesh, clamped onto its operator DNA over open water.

lambda_main.pov -- the repressor over sea and sky, with the poster's chrome titling.

---

Vitrine -- the standard exhibit case

pov-scenes/vitrine/ -- a museum case built in exhibit units: the molecule is normalized to a unit sphere by the enclosing radius pdb2pov writes into every file, and the room is built around that. These four are one set, not four scenes. The same camera, lighting and depth budget hold a 2.5x range of molecular radius with no per-structure tuning, which is the entire claim -- any one of them alone demonstrates nothing.

exhibit_gfp.pov -- 1EMA, 1,866 atoms, enclosing radius 31.2 A. The 11-strand beta barrel, the smallest of the set.

exhibit_hemoglobin.pov -- 2HHB, 4,779 atoms, 40.3 A.

exhibit_ompf.pov -- 2OMF, 8,481 atoms, 51.0 A. The trimeric porin, modern heir to porin/3porin.pov above.

exhibit_f1atpase.pov -- 1BMF, 23,481 atoms, 79.0 A. The largest, on the same plinth at the same camera.

exhibit_ompf_cartoon.pov -- the same porin trimer as a Richardson cartoon rather than as atoms, via quiltwright.pymol. porin/3porin.png above is the same subject drawn the same way in 1994, by an exporter whose output survives in no repository; this one comes from a PDB ID and a command line. Its geometry is generated rather than committed -- 8.9 MB for one trimer -- so the wrapper carries the command that makes it.

The same subject, the same room, the same camera, thirty-two years apart. Left: 3porin.inc, 7,835 triangles from the lost 1994 exporter, brought onto the pdb2pov -o contract by pov-scenes/vitrine/porin_1994.inc, which measures and centers the original geometry without altering it. Right: 2OMF from the RCSB, 75,792 triangles through quiltwright.pymol.

They are not quite the same molecule, which is the honest reading of the picture: the 1994 mesh is a monomer -- one 16-strand barrel -- while the modern one is assembly 1, the biological trimer. The archive image was of a single subunit, and nothing recorded that until the two stood side by side.

---

PyVista datasets

Rendered by scripts/render_pyvista_hologram.py rather than POV-Ray -- candidates surveyed in pyvista-datasets.md.

st-helens -- Mt. St. Helens post-eruption DEM (examples.download_st_helens), terrain relief with no texture.

damavand -- Mt. Damavand volumetric data (examples.download_damavand_volcano), a single conical peak.

brain -- classic VTK brain.vtk volume, a human head MRI (examples.download_brain).

mouse-brain -- Allen Institute mouse brain CCFv3 average template, 50 µm (see pyvista-datasets.md for the other resolutions).