News - Serious number changing bug found in some Xerox copiers

[HEXUS]

Internal compression software often causes numbers 6 and 8 to be substituted in copies.

Read more.

[NecronomicoN]

That's pretty scary when you think about it. Where are you Saracen, could use some commentary.

[miniyazz]

Just.. wow.

[mtyson]

Xerox has said a software patch will be issued to address this problem in the coming weeks.
The BBC reports that Xerox "acknowledges that referring to the affected mode as "normal" on the devices' selection screens needs to be reconsidered".

[Odeas]

You'd think that'd get picked up during testing by Xerox unless they were using it on size 20 fonts.

You can bet companies that use things like invoice scanning for archiving etc have some nasty desicions to make.

[aidanjt]

Why would they bother with lossy compression at all?

[MrJim]

Eh? I thought copiers used a raster imaging process, like a 'photograph', err...as in 'photocopier'. Or does this story refer to OCR?

[kompukare]

Eh? I thought copiers used a raster imaging process, like a 'photograph', err...as in 'photocopier'. Or does this story refer to OCR?

Nah, they were being far too clever here. It scans the original, converts it to JBIG2 and the prints. Problem is that a high compression levels JBIG2 looks for repeating segments in the picture and tries to only encode them once. So it might see a '26' once decide it will be encode once but then when it comes across '78' it finds that the '26' is close enough.

Don't know how this slipped by Xerox. I mean it's even mentioned the JBIG2 wiki:
http://en.wikipedia.org/wiki/JBIG2#Disadvantages

This is from the guy who spotted it:
http://www.dkriesel.com/en/blog/2013...he_causes_edit

[Unique]

i thought my arse looked a bit funny when i copied it at the office xmas party

[Butuz]

That's why I buy Ricoh!

[MrJim]

Nah, they were being far too clever here. It scans the original, converts it to JBIG2 and the prints. Problem is that a high compression levels JBIG2 looks for repeating segments in the picture and tries to only encode them once. So it might see a '26' once decide it will be encode once but then when it comes across '78' it finds that the '26' is close enough.

Don't know how this slipped by Xerox. I mean it's even mentioned the JBIG2 wiki:
http://en.wikipedia.org/wiki/JBIG2#Disadvantages

This is from the guy who spotted it:
http://www.dkriesel.com/en/blog/2013...he_causes_edit

Ah...thanks for the explanation

[killer7xx]

Maybe this was the cause of the world recession