(http://img186.imageshack.us/img186/8198/bugtb4.jpg)
Phylloxera, is this accompanied by any error...like the one I just mentioned about skipping a render function?
Maybe it's because you aborted the render?
Cause I got the feeling that the order in which parts of the image are being rendered is somewhat weird in the last version.
"...the order in which parts of the image are being rendered is somewhat weird in the last version." That's a key to understanding what's going on, if we can help. I've noticed the same thing, but I'm not sure it's the way it should be happening. Every time I have a render go bad, the regular order I'm accustomed to in a render takes on new avenues and goes all over the place.
Quote from: nikita on April 30, 2008, 10:05:48 AM
Maybe it's because you aborted the render?
Cause I got the feeling that the order in which parts of the image are being rendered is somewhat weird in the last version.
That information to take into account, I reported on the render l'll see!
Quote from: calico on April 30, 2008, 10:13:50 AM
"...the order in which parts of the image are being rendered is somewhat weird in the last version." That's a key to understanding what's going on, if we can help. I've noticed the same thing, but I'm not sure it's the way it should be happening. Every time I have a render go bad, the regular order I'm accustomed to in a render takes on new avenues and goes all over the place.
I kind of dismissed that as being what core was doing what job.. A guess.
Mandrake, I was thinking the same thing until it started breaking down. Just trying to figure out what it might be.
When reporting bugs please let us know what version you are using. In this case I assume it is the unstable alpha. It is known to fail to complete some parts of the render, most likely when it encounters memory allocation problems (which are reported in the Errors and Warnings dialog). If no errors were reported by Terragen, then it is probably a different bug.
Matt
Nikita, Calico,
When rendering with more than one thread, more than one bucket/tile is rendered at the same time. With many threads this will look completely different to what you're used to in previous versions :) If a thread fails to complete a bucket (e.g. due to an error) it will usually move onto the next unstarted bucket.
Matt
@Matt - If the newest Alpha renderer (multi-threaded) doesn't break down (no errors), will the Pre-GI and subsequent rendering look the same as the previous version? Or will the paths all change and go in various directions?
The order in which buckets are started is the same as before (top right to bottom left, in rows). But some buckets will render much more quickly than others, so it might look a bit strange if it gets in a situation where a bucket that is further along the sequence is more filled than an earlier bucket. There's no voodoo magic going on :)
Matt
I've encountered some thing that looks like that in the alpha, but once the render moves on a bit the issue seams to resolve its self and that error as far as I can tell is no present in the final image.
Regards to you.
Cyber-Angel ;D
I tested again, still a bug, not of error message announced (version used: TG2 1.9.88.3)
It is necessary to practise a small crop to solve this problem!
Thanks Phylloxera - it's interesting that this happens without reporting an error.
I'm currently working on making the multi-threaded renderer more efficient with its memory with a large number of threads. I've also fixed a memory leak which was leading to high memory use after repeated preview renders. It's possible that these changes will reduce the number of scenes that experience these problems. We'll try to release an update soon.
Matt
Quote from: Matt on May 01, 2008, 09:47:52 AM
Thanks Phylloxera - it's interesting that this happens without reporting an error.
I'm currently working on making the multi-threaded renderer more efficient with its memory with a large number of threads. I've also fixed a memory leak which was leading to high memory use after repeated preview renders. It's possible that these changes will reduce the number of scenes that experience these problems. We'll try to release an update soon.
Matt
Thank you !
Would it be advantageous (or possible) to mark and report buckets that fail and then present an option to retry them, maybe on a different core than the one it failed on?
Of course the main goal is to get no render failures, but for testing purposes it may be good to see if a failed bucket would ever complete or not.