Adding images in raw PDF

I am trying to make pdf's manually. I got the basics but the one thing I cannot figure out is images.

The thing I am trying right now, as a start, is adding a simple image in the form of binary code. The binary code has a length of 9 and should be able to represent a black and white image of 3x3. The code being: 111000111 (this should just be a black horizontal line through the middle). Ofcourse this is over simplified, not compressed and not usable for more complex images but I am desperate and just want to display SOMETHING :).

Hope someone can help and teach me more about this topic.

my new pdf (after johnwhitington comment except for point b)

%PDF-1.7

1 0 obj
  << 

     /Pages 2 0 R
     /Type /Catalog  >>
endobj

2 0 obj
  << 

     /Type /Pages
     /Count 1
     /Kids [3 0 R]  >>
endobj

3 0 obj
  << 

    /Type /Page
    /Parent 2 0 R  
    /Contents [4 0 R]
    /MediaBox [500 500]

    /Resources 
    <<
      /XObject 
      <<
        /Im1 5 0 R
      >>
    >>
  >>
endobj


4 0 obj 
  <<

  >>
    stream
      q
        1 0 0 1 100 100 cm
        /Im1 Do
      Q
    endstream
endobj


5 0 obj 
  <<
    /Type /XObject
    /Subtype /Image
    /Height 3
    /Width 3
    /BitsPerComponent 1
    /Length 9
    /ColorSpace /DeviceGray
  >>
    stream111000111endstream
endobj


trailer
  << /Root 1 0 R
   /Size 7
  >>
%%EOF

my old PDF:

%PDF-1.7

1 0 obj
  << 

     /Pages 2 0 R
     /Type /Catalog  >>
endobj

2 0 obj
  << 

     /Type /Pages
     /Count 1
     /Kids [3 0 R]  >>
endobj

3 0 obj
  << 

    /Type /Page
    /Parent 2 0 R  
    /Contents [4 0 R]

    /Resources 
    <<
      /ProcSet [/PDF /ImageB]    
      /XObject 
      <<
        /Im1 5 0 R
      >>
    >>
  >>
endobj


4 0 obj 
  <<

  >>
    stream
      q
        1 0 0 1 0 0 cm
        /Im1 DO
      Q
    endstream
endobj


5 0 obj 
  <<
    /Type /XObject
    /Subtype /Image
    /Height 3
    /Width 3
    /BitsPerComponent 1
    /Length 9
    /ColorSpace /DeviceGray
  >>
    stream111000111endstream
endobj


trailer
  << /Root 1 0 R
   /Size 7
  >>
%%EOF

@johnwhitington has covered the basic image programming in a PDF

The image is placed and scaled usually using four lines of code in a q block in the pages contents such as this working single line:-

q 192 0 0 192 100 100 cm /Img0 Do Q

192 is the dx and dy, 0 is the "skew" 100 is the x y, Img0 is the image number and Do is the stacked write code. None of this describes what real size Img0 is

To see a jpeg insertion using a hybrid text approach (after the text is prepared) see https://stackoverflow.com/a/75710613/10802527

So you need to inject the image pixels using any method that works in the editor, for RGB images JPG import is simplest HOWEVER JPG is not an ideal format for PDF as its pure binary and most text files cannot use pure binary inputs. So for MP4 video and JPEG images they need to be converted into a text safe format such as textual HeX 00 01 02 03 etc thus all 256 ANSI codes can be written into an ANSI editor, such as NotePad.

In other byte words 1 bit black is 00 and white is FF for RGB that's FFFFFF and 000000

Right, so how to write that pixel is like this set a pointer to an object in the page resources perhaps an entry like <</XObject <</Img0 6 0 R>>>> the

6 0 obj needs declarations such as number of pixels and colors and encoding type.

6 0 obj <</Type/XObject/Subtype/Image/ColorSpace/Device...

in your example we can see

16 0 obj 
  <<
    /Type /XObject
    /Subtype /Image
    /Height 3
    /Width 3
    /BitsPerComponent 1
    /Length 9
    /ColorSpace /DeviceGray
  >>
stream
 ­ 
endstream

note that will without any visible object between stream start and endstream produce this odd image WHY ! that looks like it should be

001101101

the answer is because its what's really there as binary stream which is not visible in an ANSI editor like notepad.

the characters are binary 20 AD A0 where 20 is a blank whitespace and AD is
binary 101 01101 and \n = A0

so those bytes look like

00100000
10101101
10100000

lets test that by widen to 6 bits and as expected we now get this

So the core issue is that in a text format the bits are taken as literal NOT visible 01's which cannot be convenient for handling imagery. At this level what we need is to start use an encoding such as ASCII HeX (/ASCIIHexDecode)

Answer

So a PDF is a Bitstream as a ByteStream and you want 111000111
that will be

111
000
111

thus its

11100000
00000000
11100000

easy use this or similar

stream
àà
endstream

where there is an invisible black character between the two à 's
result will be pixel perfect, but shows that black 00000000 characters are not easy to write in ANSI text they ideally need HEX coding one trick for RGB is use a white space for black as [space] = \x20 = thus as dark as a "safe" ANSI text character can be in its bytes format. so blackish and white as 8 bits would be ÿÿÿ ÿÿÿ

Hence for a similar, BUT blackish result as above we could use

    /Height 3
    /Width 3
    /BitsPerComponent 8
    /Length 27
    /ColorSpace /DeviceRGB
  >>
stream
ÿÿÿÿÿÿÿÿÿ         ÿÿÿÿÿÿÿÿÿ
endstream
endobj

A few hints:

a) You need Do not DO.

b) If it's one bit per component, and nine pixels, there will be two bytes (nine bits), not nine bytes in the image. '1' and '0' are 8-bit characters, not bits.

c) You don't need ProcSet these days.

d) To see your image on the screen, you'll want something like 1 0 0 1 100 100 cm to scale it up so it's visible.

e) Your page needs a /MediaBox