We have all seen the relatively crude facial representations manufactured from the genetics of retrieved DNA (created by Parabon NanoLabs), that produces reasonable but extremely limited results. But by identifying more genetic markers and merging it with artificial intelligence, it may soon be possible to create an extremely accurate age-based 3-D image of a murderer from DNA left at the crime scene, or deposited on the stamp or seal of an envelope.
A recent genetic database of 9,674 Han Chinese individuals has been used by the Chinese Academy of Sciences researchers with the aim to turn crime scene DNA into extremely lifelike 3-D imagery. The model called Difface uses single nucleotide polymorphisms, frequently called SNPs (pronounced “snips”), which are the most common type of genetic variation among people..They can be found in regulatory regions near genes, potentially affecting gene expression.
A recent genetic database of 9,674 Han Chinese individuals has been used by the Chinese Academy of Sciences researchers with the aim to turn crime scene DNA into extremely lifelike 3-D imagery. The model called Difface uses single nucleotide polymorphisms, frequently called SNPs (pronounced “snips”), which are the most common type of genetic variation among people..They can be found in regulatory regions near genes, potentially affecting gene expression.
Each of the 9,674 Han Chinese individuals provided both genome sequencing data and high-resolution 3D face scans. The results were surprisingly detailed. Even subtle traits like the nasion — the dip at the top of the nose — or cheekbone shape could be recreated with remarkable precision.. By analyzing patterns in the DNA, including the segments called SNPs that influence facial traits, the system learned to match genetic markers with physical features like nose shape, cheekbone structure, and jawline. Thus creating a digital face with a high degree of precision and average measurement inaccuracy of only 3.5 mm. This is extremely accurate when you consider the size of the human face. When age, sex and body mass indes are added into the equation, the error dropped to 2.93 mm.
The most obvious application of this tool is its use in forensics. An artificial intelligence tool that can perform such tasks could be used by law enforcement to retrieve DNA from a crime scene and then reconstruct a suspect’s face in absence of witnesses or camera footage. Below is an original scan taken from an individual, accompanied by a generated image using their DNA. Bearing in mind that this technique is in early development and has room for scalable improvement, the results are quite astounding.
The most obvious application of this tool is its use in forensics. An artificial intelligence tool that can perform such tasks could be used by law enforcement to retrieve DNA from a crime scene and then reconstruct a suspect’s face in absence of witnesses or camera footage. Below is an original scan taken from an individual, accompanied by a generated image using their DNA. Bearing in mind that this technique is in early development and has room for scalable improvement, the results are quite astounding.
CLICK THE IMAGE ABOVE TO VISIT ZME SCIENCE FROM MAY 9TH 2025
When the model is eventually expanded using an increased database of larger populations such as America, the aim is to correlate specific markers in the DNA to particular features of the face with reduced margins of error. Think of the DNA like a 3-D map, that can be recognised and visualised by artificial intelligence and converted into human form. We would have nothing to lose by using the DNA retrieved from the hand of Cheri Jo Bates, the DNA from the 1978 letter, or the hair from the October 13th 1969 Paul Stine letter, to produce a 3-D facial image from any of the given samples.
If an image produced in the Zodiac case looked extremely similar to the Presidio Heights sketch or one of the many suspects proposed as the Bay Area murderer, it would be a rather fascinating development. The image formed in respect to the Riverside murder of Cheri Jo Bates could be compared to individuals from the library and any number of suspects connected to the young 18-year-old college freshman. The images created from possible Zodiac Killer DNA by the use of artificial intelligence and then age progressed through the years, may generate an interesting response from the public - particularly from individuals who point the finger towards a person who once lived (or maybe still lives) in Vallejo, San Francisco, or any of the neighboring areas.
If an image produced in the Zodiac case looked extremely similar to the Presidio Heights sketch or one of the many suspects proposed as the Bay Area murderer, it would be a rather fascinating development. The image formed in respect to the Riverside murder of Cheri Jo Bates could be compared to individuals from the library and any number of suspects connected to the young 18-year-old college freshman. The images created from possible Zodiac Killer DNA by the use of artificial intelligence and then age progressed through the years, may generate an interesting response from the public - particularly from individuals who point the finger towards a person who once lived (or maybe still lives) in Vallejo, San Francisco, or any of the neighboring areas.
RSS Feed
