0
$\begingroup$

Apologies if this has been asked, I really couldn't find anything on it.

Let's say I want to render a 7200-frame animation, and the goal is to have the final render be at a samples = 1000.

However, doing so directly is committing the PC to a very long render job in its entirety.

Is it instead possible to render the animation, at say, 250-sample increments?

For example, I render 7200 frames at 250 samples. Then I do it again, with some different seed. Then again, 2 more times.

This breaks the render into 4 shorter jobs. Of course, it takes 4x the space, because it's 4 separate EXR sequences.

Can I then composite these together to approach a 1000 sample quality?

Is there some setting or property of seeds, and/or some technique in the Compository node system I can use to essentially "render in sample stages"?

$\endgroup$

1 Answer 1

0
$\begingroup$

What you are saying is just not possible. If you just don't want to commit your cpu for a super long time but still want the quality of 1000 samples, then a better option would be to render the frames at 1000 samples but at a time render only a few frames. You can do this by changing you start and end frames. eg. First day set start frame to 1 and end frame to 1000 then next day set start frame to 1001 and end frame to 2000. Just set them according to your situation. Even if what you are trying would be possible, this is actually faster because CyclesX uses adaptive sampling by default.

$\endgroup$
1
  • $\begingroup$ All good, then, so it's not possible. Chunking by sets of frames isn't really an answer to my question. The premise is to have the entire animation rendered faster, so I could denoise and deliver if I wanted to at the reduced quality level. It's a safety mechanism to prevent a delivery being delayed by rendering at a higher quality and having that process potentially crash or fail, resulting in an incomplete animation by deadline time. But if this isn't possible, then I'll mark this as answered. $\endgroup$
    – Ryth Azhur
    Jun 1, 2022 at 22:25

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .