The galaxy inside the giant's head: Hubble imagery (ESA/Hubble, Creative Commons Attribution 4.0 International)

At the head of the products the camera dives into the nebula inside the giant's glass head and backs out of it
until the whole giant stands in his space (src/experience/universe/entity/entrance: inside.ts draws the nebula at many
times the drawing's scale, stars.ts the bright stars taken out of the image). The nebula there is made from one Hubble
Space Telescope image published by ESA/Hubble; everything else of the giant is the drawing and the page's own.

Files made from the image (each tier its own sizes; scripts/space/kresba/entrance/make_entrance.py makes them all):
  deep-3423.avif, deep-1112.avif    the nebula round the deepest frame, at its finest (desktops, phones)
  core-1750.avif, core-1283.avif    the helmet's inside
  glow-1221.avif, glow-940.avif     the glow the camera dives into: the helmet's nebula and the image's gas round it
  depth-470.png                     the nebula's depth: where the pillars stand in front of the gas behind them
  stars.json                        the stars taken out of the image: their places, colours, light and sizes
(and neb-606.webp, neb-454.webp: the drawing's own helmet nebula with its stars taken out, under its glass and past
its outline the image itself, its colours taken to the drawing's round it)

Licence: Creative Commons Attribution 4.0 International (CC BY 4.0),
https://creativecommons.org/licenses/by/4.0/ , as ESA/Hubble releases its images and videos
(https://esahubble.org/copyright/). ESA/Hubble does not endorse Nolentra or its use of the image.

The image, with its exact credit as published:

"New view of the Pillars of Creation - visible" (heic1501a, the Eagle Nebula, M 16), released 5 January 2015
   Credit: NASA, ESA/Hubble and the Hubble Heritage Team
   Page:   https://esahubble.org/images/heic1501a/
   File:   https://cdn.esahubble.org/archives/images/large/heic1501a.jpg (6780 x 7071)
   Used:   the region round the left and the middle pillar (about x 280 to 5580, y 0 to 5710 of the image's px),
           laid on the drawing's own helmet

Changes made to the image (make_entrance.py, run on the file above):
- its grain lightened at its finest scale only (the detail under a pixel kept at a third, its colour's grain smoothed
  over about a pixel; the filaments untouched);
- its stars taken out as the light they add: each bright star's halo (its profile over the gas round it) subtracted,
  its diffraction spikes measured where they run and taken off, the spokes round the four brightest too, its
  saturated core filled from round it with the gas's own grain; each smaller star taken out by its own light only;
  all of them kept as points the page draws (stars.json);
- laid on the drawing's helmet: an affine map fitted to eight features the two share (the left pillar's lit cap, the
  pillars' heads and edges), then a gentle smooth warp (a dense flow between the two, smoothed over 28 px of the
  1620 px drawing, fitted on the image with every bright star masked wide) so its pillars stand where the drawing's
  do; turned about 0.6 degrees, about 6.9 of its px a px of the drawing;
- its colours graded to the drawing's helmet (a linear Monge-Kantorovich map between their colour distributions, in
  the light's cube root);
- its pillars' depth drawn from it (depth-470.png, smooth);
- cut into the levels above and resized (area averaging), encoded as AVIF with the colour at full resolution.
On the page the drawing's own light is laid under the image at every scale coarser than the zoom shows it sharp: deep
inside, the image's structure in the drawing's colours; as the camera backs out, the drawing's own helmet takes over
scale by scale, and at the drawing's own scale only the drawing shows.

The credit on the site: the data and licence page, /licence-dat/ and /en/data-licences/ (src/pages/licence-dat.astro,
the image's credit linked to its page and this notice).