I can't imagine that this kind of evidence would be admissible in court, right? It might be useful to give you some leads but I would hope that no one is convicted on this tech alone.
This is akin to: Police having grainy monochrome security camera snapshot of a suspect, putting the photo in a microscope, and concluding that the suspect has salt-and-pepper hair and a square pixel shaped face.
"fair skin, brown eyes and hair, no freckles, and bushy eyebrows" isn't enough entropy for facial recognition to work. Any entropy that any facial recognition would be operating on here is completely made up out of thin air by the rendering program.
No, but it's more than enough to profile people, which is the real point.
The dna draws the face, the face is back checked for a match, the match has his dna collected and the dna is compared. The process in-between and the face match isn’t the evidence.
I have bad news for you about what is already regularly used in court to convict people, and the scientific validity thereof.
Pretty much everything that makes up "forensics", except for DNA analysis (and even that carries a huge asterisk). Fingerprints, bite marks, blood spatter... the list goes on. Drug tests are borderline - reliable drug tests exist, but police often use tests with much higher false positive rates.
Believe it or not, "police psychics" also exist[0], and are used to manufacture probable cause, so that also fits the definition of "used to convict people".
Forensic psuedoscience is widely known to be junk by modern scientific standards, but courts have stuck by it, in part because "that's how it's been done for decades/centuries", and in part because they have acknowledged that it would be too difficult to convict many cases if forensics were deemed inadmissible.
[0] https://www.cia.gov/readingroom/docs/CIA-RDP96-00788R0001002...
https://www.lawinfo.com/resources/criminal-defense/are-lie-d...