Thanks! On my cellphone not even enough of the UI was working for me to discover those URLs. I suspect a certain amount of error recovery is in order for wgetting all 2238 images. 2000 seems to be the maximum resolution available, which is under 100dpi. A few of the images seem to have been uploaded to https://commons.wikimedia.org/wiki/Category:Codex_Atlanticus.

There are a couple of scans of a 43-page Italian edition published by Ulrico Hoepli on the Archive: https://archive.org/details/codex-atlanticus-leonardo-da-vin... https://archive.org/details/codex-atlanticus-leonardo-da-vin... but they seem to be of very poor quality.

I'm done downloading now (with a sleep of 1 second between pages), and I have 1064125470 bytes of JPEG files, a very reasonably torrentable size. I'll see if I can put together a torrent and upload to the Archive and Commons...

Or in PowerShell on Windows:

  1..1119 | % { iwr "https://codex-atlanticus.ambrosiana.it/assets/2000/000R-$_.jpg" -OutFile "000R-$_.jpg" }
  1..1119 | % { iwr "https://codex-atlanticus.ambrosiana.it/assets/2000/000V-$_.jpg" -OutFile "000V-$_.jpg" }

Some people around me swear PowerShell has better user experience than unix shells, but then I keep seeing examples like these. How on earth could people prefer this compared to `wget https://codex-atlanticus.ambrosiana.it/assets/2000/000V-{1.....`?

In this case presumably the main difference is not PowerShell vs. bash but iwr vs. wget? Because I think this is roughly equally bad (untested):

    for page in {1..1119}; do
        iwr "https://codex-atlanticus.ambrosiana.it/assets/2000/000R-$page.jpg" -OutFile "000R-$page.jpg"
        iwr "https://codex-atlanticus.ambrosiana.it/assets/2000/000V-$page.jpg" -OutFile "000V-$page.jpg"
    done

Also until recently bash didn't have {42..53} syntax. You had to use `seq`. There was an alternative name for `seq` in Unix Power Tools, `jot`, because it wasn't standard: https://docstore.mik.ua/orelly/unix/upt/ch45_11.htm. This section was by ORA author and sysadmin Linda Mui (https://www.oreilly.com/pub/au/268), but I don't know if she wrote `jot` or just popularized it.

Any idea on how to best compile it to an ebook? Just stuffing the jpgs into a pdf rarely works well...

I usually do what rarely doesn't work well for you, but it works decently for me. You get 1 page per image and the image isn't compressed or touched at all.

  apt install img2pdf
  img2pdf *.jpg -o leonardo-da-book.pdf

wouldnt this mess up the order? I think you are supposed to view it like R1, V2, R2, V2, etc

Yes, this was just an example. Using wildcard expansion will give you whatever order the your current shell seems fit. Bash does alphabetical order.

More like

    echo $(for page in {1..1119}; do for side in R V; do
      echo "000$side-$page.jpg"; done; done)

I haven't that done this in some time, but templating some markdown code for pandoc and creating an ebup might be a viable avenue.

Maybe what rarely works well for NoMoreNicksLeft is having a gigabyte of JPEGs in a single HTML chapter inside the epub? In that case you could do something like divide the files into 373 "chapters" of 6 pages each?

One of the fragmentary editions I linked on the Archive uses the .cbr Comic Book Reader format; perhaps that is a better format than .epub for high-resolution scans of every page?

Oooh... I have even less luck with epub, when the pages are an image-per-page.

Easy way would be to just drop them in a zip and label it .cbz. Most readers handle CBR/CBZ just fine.

Oh, is .cbz that simple? Does it use the file order of the zipfile members or some other order? (https://acbf.fandom.com/wiki/ACBF_Editor_-_Creating_Metadata says it uses alphabetical order, which is the wrong order in this case.)

It may be useful to use zip -Z store. JPEG data isn't going to get much benefit from another layer of LZ77.

Calibre comes with a ebook-convert command, that one might work

ocrmypdf (rocks!)