git @ Cat's Eye Technologies NaNoGenLab / master naive-cut-up / README.md
master

Tree @master (Download .tar.gz)

README.md @master — view rendered · raw · history · blame 

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
naive-cut-up
============

Hypothesis
----------

We hypothesize that if we cut up a newspaper.  We also hypothesize that
we cut up a newspaper.

Apparatus
---------

*   Python 2.7.6 (probably works with older versions too)
*   [Pillow](http://python-pillow.github.io/) (it might work with PIL too)
*   Some scanned images of newspapers, books, etc., in PNG format, for example
    obtained by [fetch-chronam](../fetch-chronam/)

Method
------

*   Start with "blank" canvas.  For simplicity, we actually use one of the
    input images as the "canvas".
*   Pick a source image.  Pick a rectangle within that image.
*   Copy the image within the rectangle to a random location on the canvas.
*   Repeat from step 2 until we guess we've covered the canvas.

### Detailed procedure ###

First, we assume some PNGs of scanned newspaper pages involving some topic
(in this example, cheese) have been obtained.  (PNG format is probably not
a strict requirement; if I had built my install of Pillow with JPEG support
I could probably load in JPEGs directly.  Alas, I did not.)

Then we reorganize these files a little and find the largest one:

    $ mkdir pages
    $ mv *.png pages/
    $ ./naive-cut-up.py largest pages/*
    [...]
    LARGEST IS pages/ca_5.jp2.png

And we use that one as our base:

    $ mv pages/ca_5.jp2.png .

Then

    $ ./naive-cut-up.py cutup ca_5.jp2.png pages/*
    ca_5.jp2.png <PIL.PngImagePlugin.PngImageFile image mode=L size=6274x8906 at 0x7F427C776128>
    pages/ca_0.jp2.png <PIL.PngImagePlugin.PngImageFile image mode=L size=1788x2692 at 0x7F427C7765A8>
    pages/ca_10.jp2.png <PIL.PngImagePlugin.PngImageFile image mode=L size=1784x2646 at 0x7F427C776680>
    [...]
    pages/ca_8.jp2.png <PIL.PngImagePlugin.PngImageFile image mode=L size=1960x2728 at 0x7F427C77DFC8>
    pages/ca_9.jp2.png <PIL.PngImagePlugin.PngImageFile image mode=L size=1864x2680 at 0x7F427C781290>
    <PIL.PngImagePlugin.PngImageFile image mode=L size=1812x2626 at 0x7F427C77D9E0> (562, 174, 1166, 1049) <PIL.Image._ImageCrop image mode=L size=604x875 at 0x7F427C77D950>
    <PIL.PngImagePlugin.PngImageFile image mode=L size=1880x2767 at 0x7F427C776F80> (132, 508, 758, 1430) <PIL.Image._ImageCrop image mode=L size=626x922 at 0x7F427C776EF0>
    [...]
    <PIL.PngImagePlugin.PngImageFile image mode=L size=1892x2679 at 0x7F427C77DEF0> (612, 1340, 1242, 2233) <PIL.Image._ImageCrop image mode=L size=630x893 at 0x7F427C77DCB0>
    <PIL.PngImagePlugin.PngImageFile image mode=L size=1904x2745 at 0x7F427C776E18> (808, 1725, 1442, 2640) <PIL.Image._ImageCrop image mode=L size=634x915 at 0x7F427C77D290>
    Writing output.png...

And then we marvel at the result:

    $ ristretto output.png 

(You may wish to use a less clumsy image viewer than Ristretto, yourself.)

Observations
------------

It may be difficult to tell in this scaled-down sample, but the result was
surprisingly thematic in its reference to cheese:

![Newspaper cut-up on the theme of cheese](sample-cheese.jpg)