On 8/25/24 12:37, Cousin Ricky wrote:
> I just accidentally destroyed a 14 hour render, by accidentally
> repeating the render command--within seconds of it completing, so no
> backup.  Don't you hate when that happens?
> 
> I'm using GNU/Linux, where file undeletion requires a 3-credit semester
> course to learn how to do it.
> 
> Do you with the 3 week renders ever have that problem?

I've read through this discussion on looking in on the news forums today.

Off the top of my head, I think situation on accidentally re-starting a 
long render isn't all that dire(*) these days - as long as you don't panic.

Prior to v3.7 and smp / threads, the image file was written as the image 
pixels were rendered; an accidental re-start of a long render was more 
of a problem because the output image file was almost immediately 
written to after the render phase started. Today, the image and state 
are stored internally and written to the output image file only after 
the last block of the image completes.

On Linux / Unix at least, the reality is even after the fat finger 
render phase starts and opens the output file, the prior image file will 
exist until the file handle is closed at the end of the render phase(**).

Let the accidental re-render run (or better pause it) until you've 
copied / moved the existing image file somewhere safe - then abort the 
fat finger render.

Bill P.

(*) - POV-Ray versions using a gui might make me a liar, if the gui does 
something with the image output file itself on the re-render.

(**) - Given most all default file buffering schemes.

Post a reply to this message