Planetside Software Forums

Has anyone run across this.

Where the rendering stopped is when the errors started. It claims 21 errors however, there is one error and two errors repeated X 5.

Let me know if it's me.

My Graphics card is a nVidia (PNY) GeForce GTS 250 1024MB XLR8 Performance Edition.

---

One of the objects in the scene doesn't have the same filepath as when you loaded it, I imagine.
The first error says 'cannot locate file'. Did you move any models that were being used in this scene to a different location on your pc?
Go through your objects until you locate the object reader that isn't finding the file and reassign the correct filepath.

I see the traceray error often, it never seems to have any impact on the scene, it usually appears for me when I delete certain types of nodes from the node network, and not really at all when rendering.

Thank you. I'll check it all out.

Appreciate you're response.

Marc

Actually, this usually indicates a memory-related issue. It's reporting the issue multiple times because it occurs multiple times as it tries to continue rendering the scene. I would suggest posting your render setting details here, and watching the Task Manager next time you try to render the scene. Since you're on a 32 bit version of Windows, TG2 won't be able to use more than 2GB of memory. If it gets too close to, much less above that, you'll get errors like this, or even a crash. Running a 64 bit version of Windows can help even though TG2 is not yet a 64 bit app.

- Oshyan

Oshyan, I just added two gigs to my computer, making it 3.5 gigs. (I don't know where the 1/2 gig went) I was having population difficulties when they exceeded 2 million of one object. I have not seen this since the RAM upgrade.
Is the population instancing memory handled differently than the other memory use?

Here are my Render settings.

---

One thing that stands out right away - having GI surface details turned on is rarely necessary. While it may not be connected to your errors, it will slow your renders down. Recommend you look at the sticky on render settings in the Terragen 2 Discussion forum. (Microvertex jittering is also unnecessary unless you're using water).

Domdib,

I'll read the post and incorporate your suggestion.

Thank you.

Marc

Your render settings look fine really, aside from the GI Surface Details issue which was already mentioned. So I'm left to wonder how big the populations/objects (in terms of polygon count) are in your scene and how much memory your system has.

- Oshyan

I did have an .obj that was part of the problem and corrected it. Thanks for the heads up.

I did some major optimizing of Vista today, turning off unnecessary services and startup programs. Also, I created my page file on an external drive and made both min and max settings 1.5x the amount of RAM on the box. Then I went into the power settings and changed my cpu settings to min 20% and max to 100%. Vista by default will show 2 cpu's even if you are running a quad core. Most people are not aware of this. If you type msconfig under the Start button then go to the Boot tab and open Advanced options you can check if it's using all your cpu's. Did a bunch of other stuff like deleting orphans in the registry and too many other things to list here but, the box is smoking. I mounted a fire extinguisher on my monitor. There is a noticeable performance increase. I'm a Mac guy and never have to mess with this junk till I bought my new HP.

I'll do a render tonight and keep you all posted if your interested.

Marc

Try IOBit's Advanced System Care. It does all the things you did by hand. There is also a program called Game Booster, by IOBit. This turns on and off unnecessary stuff. Both are free. 8)

I finally got my scene rendered but, I got another ERROR. This may be due to an XFrog export issue. I'm just guessing.

The image may be viewed in the Image Sharing section under the topic Bigfoot.

Marc

---

If you are not too far time wise in a render, stop the render. Go to the objects populator and then the Seed Tab. Hit randomize. When you start the render over, this error should be taken care of. Very seldom does this error effect the render.

njeneb,

Thanks, I'll give it a shot.

Marc

Hi,

Quote from: MGebhart on October 27, 2009, 10:05:15 AM
I finally got my scene rendered but, I got another ERROR. This may be due to an XFrog export issue. I'm just guessing.

This isn't actually an error, it's a warning. That's what the yellow warning icon signifies, as opposed to the red error icon. It isn't a great warning message, I'll admit. This particular warning is not really something to worry about. It's actually there more as a notification to the developers. IIRC the problem its mentioning doesn't seem to effect the final image, but it does signify some internal state has got a bit out of whack and we've been trying to figure out why.

Regards,

Jo

Thank you Jo.

You guys at Planetside are always on top of things and respond fast.

I really enjoy this forum. Everyone is helpful, respectful and encouraging, unlike some of the other multimedia forums I belong to.

Thanks a ton.

May I borrow this thread as I have encountered the same error messages. First time it crashed whole Terragen, second time Terragen survived but the render was stopped (error messages in the attached images). I am now trying with a little less objects. This is the first time I am trying to do something with my new machine (dual CPU, 24GB RAM, Windows 7 Ultimate) but it is also a new scene so it is difficult to say, which one to blame. First I was running with 6 cores, second attempt was done with 4 cores. I'll let you know what happens with my third attempt.  

---

Antti, this may be the result of some memory allocation fault. Try using the Pre-Allocate subdivide cache. This will grab the memory the program needs. You may have another program running which does not handle memory correctly; it may be overwriting memory T2 has called for.
The reason I think this may have happened is one of the errors. The one about the subdivide cache. It's a guess, so...

Thanks for the suggestion, I will try the preallocation thing as the third attempt was not successful (got the runtime error quite early on).

Reducing the subdivide cache (in the advanced tab in full render) helps too. It is in the error mesage.  If it is 400 try 300 - 200 or lower too.

Kadri.

This is definitely a memory-related issue. I assume your Windows 7 is 64 bit, so you should have up to 4GB for TG2, in practice about 3.5GB. The thing is the more threads you use, the more memory is needed for render caches (the cache size is split among each thread). I use up to 8 render threads with pretty complex scenes and it works well, but you need to keep the cache size down. 100MB per thread is usually fine, but in some extreme cases you need to reduce to 50MB per thread. It will warn you about this as a potential performance issue, and sometimes it *will* make things a bit slower, but you'll have a better chance of succeeding. The one thing I do see a lot is people assuming that setting a larger render cache is a good idea, for performance or whatever, but this is definitely *not* a good idea, especially in complex scenes.

In the future when TG2 has a 64 bit binary, memory issues like this should be a thing of the past on high-memory systems.

- Oshyan

Thanks for the helpful information. Today I managed to finish the render (see picture (http://antti.pictures.fi/kuvat/Terragen2/syksy5.jpg/full)). Preallocation of subdiv cache did not resolve the problem (another crash) but then I changed the size of subdiv cache to 100Mb and I also divided the render to two crops that I ran simultaneously and finally combined together. By the way, even though I set 100Mb for the subdiv cache, Terragen said that the "size was automatically increased to 200 Mb so that each thread has 50 Mb" (I was running with 4 cores per process). Is this OK or is there a way to force the 100 Mb setting if I encounter more serious problems with some other render?

Edit: Now I noticed that Oshyan was talking about memory per thread and not about the whole cache so I was running now with 50 Mb per thread which definitely was a working solution

As you noted from the error message, 50MB per thread is currently the minimum.

- Oshyan