SOMETIMES there are security cameras. And sometimes the crook is wearing a mask.
But sometimes justice comes down to whether a witness can remember important things about the bloke shoving a double-barrel shotgun in their face - things like the shape of his chin.
A picture may be worth a thousand words, but a good identikit could be worth 10 years in jail for the offender.
Every year, police officers at the Forensic Science Centre in the city reproduce about 200 images of suspects, based on recollections from the victims and the witnesses of crimes.
Technology called Facial Automated Composition and Editing (FACE), introduced to South Australia Police in 2000, allows officers to build a person's face by focusing on five main components - chin, mouth, nose, eyes and hair.
Crucially, they have no subjective input in helping victims or witnesses recall information and work only on building the image from information given to them. If two people have witnessed an incident, they will be asked to provide descriptions independently.
The process, for some, can be relatively straightforward, while others can spend hours trying to remember every possible detail. Senior Sergeant First Class Ian Crammond, who has worked at the centre for 20 years, said the time taken to produce an image didn't necessarily influence how close a final resemblance might be.
Sgt Crammond said emotion played a big part in recollections.
"It's not unusual for two different witnesses at independent times to have quite a different recall of the same event," he said.
"That can either be positive to the process or it may be that they are under so much stress that they can't recall a lot of detail."
To determine whether a victim or witness is suitable for the FACE program, they must first confirm to the case investigator that they are able to provide enough detail.
"If they can describe all five (components), that's suitable. If they can only describe four, that's still suitable. Once they can only describe three, it's not," Brevet Sergeant Tony Rowe said.
Victims and witnesses are led through the five main components on a database.
They must choose the most suitable sub-category for each component, then view about 20 images. Once they have short-listed about half a dozen, a decision is made and the face starts to take shape.
"It is purposely limited to a range of chins. You could theoretically put thousands on there, but that would just become confusing," Sgt Crammond said.
Elements such as skin colour and facial features can be adjusted to suit and once the victim or witness is satisfied the image is a suitable likeness, it is provided to the investigator for circulation.
Sgt Rowe said identikit images would never perfectly replicate the real image of a person.
He said achieving a likeness of at least 75 per cent was the aim.
"It's an old story, a picture tells a thousand words," he said.
As Sgt Crammond points out, the identikits are just one step of an investigation.
The FACE technology has come a long way since the days of sketch artists and physical components on transparent slides.
A police spokesman said there were no statistics kept on how many identikit images had led to charges being laid against suspects.
How victims put police in the picture | adelaidenow