AI cannot draw a QR code that works

What happened and why it matters

I asked an image generator for a poster with a square scannable code on it. It produced a handsome poster with a convincing code: the right shape, the right texture, the little target squares in three corners. It does not scan. None of them ever scan.

The reason tells you where the boundary of this whole class of tool is

An image generator is drawing a picture of a code. It has seen thousands of them and learned what they look like. Dense black and white blocks, those three distinctive corner markers, a quiet border. It reproduces the appearance faithfully.

But one of these codes is not a texture. It is an encoding. Every little square is the result of a specific calculation over your specific text, wrapped in mathematics that lets a scanner recover from smudges. The pattern is not decoration. The pattern is the data.

Drawing something that resembles it is like writing a paragraph in convincing handwriting using letter shapes that are not letters. It looks like writing from across the room, and it says nothing.

So the method is boring and completely reliable

Make the code with a real encoder, then place it into whatever artwork you want. Let the generator do the thing it is good at, which is the picture. Let a proper piece of software do the thing it is good at, which is being correct.

What surprised me more than the generator part

While testing, I collected the ways a technically valid code still fails in the real world. Codes printed light on dark are rejected by a lot of phone cameras. Too little clear space around the edge and the scanner cannot find where the code starts. A code sized for a page, printed on a business card, becomes unreadable.

Each of those produces a code that is mathematically perfect and practically useless. Which is why the last step is not optional, and it is the step people skip because the image looks right: scan it yourself, with a phone, from the distance a real person will stand.

How to do it yourself

Making a code with a real encoder, the settings that decide whether it scans in the real world, placing it into artwork, and the test that has to happen before you publish anything.

1. Make it with proper software, never with an image generator

# Python
pip install "qrcode[pil]"
python3 - <<'PY'
import qrcode
q = qrcode.QRCode(
    version=None,                                   # smallest that fits
    error_correction=qrcode.constants.ERROR_CORRECT_Q,
    box_size=10,                                    # size of each little square
    border=4,                                       # clear space. do not reduce.
)
q.add_data("https://example.com/your-target")
q.make(fit=True)
q.make_image(fill_color="black", back_color="white").save("code.png")
PY

# or straight from the command line
qrencode -o code.png -s 10 -m 4 -l Q 'https://example.com/your-target'

2. Choose the smudge tolerance on purpose

# L  about 7%   smallest code, least forgiving. fine on a clean screen.
# M  about 15%  the usual default
# Q  about 25%  good for print, or if a logo will sit on top of it
# H  about 30%  most forgiving, largest code
# printing, or putting anything on top -> Q or H

3. Keep the clear space, and keep it dark on light

# clear space : at least 4 blocks of plain background on every side
# polarity    : DARK blocks on a LIGHT background
# contrast    : real black on real white. not charcoal on cream.

4. Keep the target short, and never point at something you cannot change

# prefer
#   https://yourdomain.tld/x
# over
#   https://yourdomain.tld/some/deep/path?lots=of&tracking=junk

# point it somewhere YOU can redirect, so the destination can move
# without reprinting anything

5. Place the real code on top, do not regenerate it

python3 - <<'PY'
from PIL import Image
art  = Image.open("poster.png").convert("RGB")
code = Image.open("code.png")
panel = Image.new("RGB", (code.width + 40, code.height + 40), "white")
panel.paste(code, (20, 20))
art.paste(panel, (art.width - panel.width - 60, art.height - panel.height - 60))
art.save("poster-final.png")
PY

6. Scan it, from the real distance, with a real phone

# the checklist
#   a plain phone camera, not a specialist scanning app
#   at the size and distance a real person will use
#   printed, if it is going to be printed
#   one dark room and one bright room
#   and confirm it opens the RIGHT place

python3 -c "
from PIL import Image
from pyzbar.pyzbar import decode
print(decode(Image.open('poster-final.png')))"

Related reading: my assistant told me it had done the job, and it had not is the same lesson about confident output that nobody checked.