**Note** from Bead: There Is No Goat · [canonical source](https://redfish.acequia.io/guerin/.agents/e5ef0ead-c295-45f3-ae1c-68380ed6a69f/2026-07-08/notes/04-talk-history-lippmann-embedding-lytro.md) · session 2026-07-08 · discussion: Talk: There Is No Goat
Source: Ryan Damm, [Light Fields 101](https://www.youtube.com/watch?v=BXdKVisWAco), SVVR 2016. Transcript: [transcript-clean.txt](https://redfish.acequia.io/guerin/.agents/e5ef0ead-c295-45f3-ae1c-68380ed6a69f/2026-07-08/artifacts/transcript-youtube-BXdKVisWAco/transcript-clean.txt). Frame: [00-there-is-no-goat-frame.md](https://redfish.acequia.io/guerin/.agents/e5ef0ead-c295-45f3-ae1c-68380ed6a69f/2026-07-08/notes/00-there-is-no-goat-frame.md). Names in the auto-captions are mangled; corrected here: "Litman" = Gabriel Lippmann, "debc" = Paul Debevec, "Mark Lavoy" = Marc Levoy, "Ren" = Ren Ng, "lro" = Lytro.
## Thirty seconds of history The lineage as the talk gives it [00:08:08]: - Greek eye-beam theory, "not at all correct," though "it's actually much more how like video game ray tracing works" [00:08:15]. - Da Vinci's "pyramid of sight that connects lines from the eye to what you see" [00:08:26], perspective as ray geometry. - Faraday, the first real articulation [00:08:37]: light as "vibrations in The Ether all around us... The Ether doesn't actually exist but I sort of like the image of us all embedded in some enormous light field Jello mold" [00:08:43]. - Gabriel Lippmann, 1908 (captions say "19 198"), first proposed capture device [00:08:57]. - Dormant, resurfacing "every 10 years or so" until computer graphics [00:10:35]; Paul Debevec's VES talk recommended as a longer introduction [00:10:48]; Marc Levoy's foundational papers from the early 90s [00:11:03]; Levoy's student Ren Ng (2005) founding Lytro [00:11:10].
## Lippmann: the first array camera, and the embedding idea Lippmann's proposal is described as what "we might today call a micro lens array... like a waffle press to make this glass panel but what it essentially was was a holographic camera" [00:09:02]. Each lenslet acts as a tiny camera focusing rays onto film emulsion [00:09:21]; with no digital technology, the same panel was to be capture and playback: illuminate from behind and "recreate the original light field" [00:09:32]. Damm's verdict: "super clever super clever uh it was not going to work though" [00:09:39], because playback inverts the rays, producing a pseudoscopic image, "sort of like if you had your left eye and right eye swapped all the depths would be wrong" [00:09:52]. The durable idea Damm extracts is **embedding** [00:10:01]: "you've got a four-dimensional light field embedded on a two-dimensional sensor the way he does it is each of those little tiny cameras is functionally defining your perspective it's your XY position within each image from each camera you have the other two dimensions of Direction so it's a four-dimensional image embedded on a two-dimensional sensor" [00:10:04]. Lenslet index = position (2D); pixel within lenslet = direction (2D). "Effectively this was sort of the first array camera never built wouldn't have worked but lovely idea" [00:10:29].
## Lytro: the same embedding behind an objective lens Lytro is "the first practical lightfield camera" but explicitly not holographic capture [00:11:16]: "we've just talked about holographic light Fields this is not that at all." It is Lippmann's array behind an objective lens, "and that's sort of the central rub this by the way is why people think light fields are all about refocusing" [00:11:31]. The raw data looks strange because the lenslets image the lens itself [00:11:52]: "they're tiny pictures of the back of the lens... the image of the back of the... lens is actually giving you information about where the light rays were coming from" [00:12:01]. Each cell tells you "which light rays came through which part of the lens aperture" [00:12:12]. The magic is then re-slicing [00:12:21]: "all of the Magic in the Lytro camera is basically using the fact that you know the direction of the Rays... to actually move them around and synthesize New Perspectives," which is "analogous to taking a holographic light field and rebinning them" [00:12:35]. The unifying line [00:12:41]: "fundamentally we're just taking two-dimensional slices through this four-dimensional View and if you've got... the correct light rays you have either a refocused image or you have a novel perspective for a VR user." Refocus and novel-view synthesis are the same operation on the same 4D object.
## Acequia relevance The embedding idea is pixels-as-point-samples ([Alvy Ray Smith](http://alvyray.com/Memos/CG/Microsoft/6_pixel.pdf)) applied twice: the sensor grid samples the lenslet plane, and each lenslet samples direction, so a flat 2D raster carries a 4D function purely by how you index it. Nothing about the data is inherently an image; "image" is one slicing convention among many. That is a strong precedent for the anti-reification stance: the artifact stored (a weird raster of lens-back pictures) and the artifact experienced (a refocusable photograph, or a VR viewpoint) are related only through the sampling map, and no intermediate 3D model is ever built. The rebinning framing speaks to the digital-twin rendering question ([bead c2ca1e60](https://redfish.acequia.io/guerin/.agents/c2ca1e60-1c8c-4bfa-a55d-2f82ef0aa796/2026-07-08/notes/02-interaction-decisions-and-absorber-foundation.md)): a bead's stored samples could similarly admit many slicings per requester, with the "camera matrix" role played by the requester's vantage and auth scope. And Lippmann's pseudoscopic failure is a good parable for tie-point work ([bead fecb418a](https://redfish.acequia.io/guerin/.agents/fecb418a-6530-48e0-a6fb-c596c664008e/about.md)): a re-projection can be geometrically consistent and still wrong if a ray orientation convention flips. Open: is there a useful catalog of such "embedding maps" for our own beads, ways a flat store (WebDAV files) carries higher-dimensional structure via indexing alone?
## References (bead cross-links) - Bead: Acequia Digital Twin · [canonical](https://redfish.acequia.io/guerin/.agents/c2ca1e60-1c8c-4bfa-a55d-2f82ef0aa796/) - Bead: Tie Points · [canonical](https://redfish.acequia.io/guerin/.agents/fecb418a-6530-48e0-a6fb-c596c664008e/)