Le_Forgeron wrote:
> Le 24/10/2009 06:28, Kirk Andrews nous fit lire :
>> "An access violation exception was generated at address 0x00000001400491C4"
>>
>> Anybody know why this might happen?  I've been getting it a lot.  The size of
>> the render (say, 2400x1200) and number of included files (both images and POV
>> files) seem to increase the probability of the failure.
>>
>>
> A bit more of details, please ?
> Does it happen at parse time, or render time ?
> Have you any additional time ? (photons, radiosity)
> 
> So far, you're like a car driver reporting "the more kilometres/miles I
> do at once, the more likely I got a problem."; you do not show the car
> to the mechanics' (minimal reproducing scene), and you do not even tell
> the company if it's a tire/gazolin/engine problem, or may be just a side
> window which cannot be operated. "it make noise sometimes".. nobody is
> officially a psychic here, it will be very difficult to get more details
> if you do not provide them.
> 
> You assumed size of render can influence that: what are the
> investigations you made to affirm that ? (you might be true, but so far
> you have provided no proof or even the detail of a study which might
> lead to that conclusion). Same for others points.

My apologies for the sparse information--it was late and I was 
frustrated that my scene had failed to render once again.  It seems that 
as soon as I'm ready to finish a scene and crank up the settings for an 
overnight render, this happens.

I had successfully rendered the scene at lower resolutions, although not 
all of the quite the same for things like radiosity, media, number of 
trees, or resolution of height_fields.

It parses fine; it seems to always happen about 5% of the way through 
the render itself.  There doesn't seem to be anything particularly 
unique in the image at the point at which it stops.  In fact, sometimes 
it will render fine with the camera one way, but if I rotate the camera, 
then the access violation will occur.  Yet, it is not random--that is, 
if I leave the scene the same way, it will happen 100% of the time.

The access violation code number seems to be consistent every time.

---

I'm using beta 34, 64 bit version on a 64 bit Vista system with 2 cores.

Current settings:  +W2400 +H1200 +fn16 +a0.2 +WL0 -RVP +SP8 +EP8

In the scene are several high-resolution height_fields, ~160k POV-Trees, 
low-grade radiosity and media, a low-quality area light, and an infinite 
plane.  The height_fields are created using functions made from tga 
files from World Machine.

I use a averaged object pattern to generate a simple proximity function 
that governs several textures on the height_fields.

Post a reply to this message