problem when rendering large image
i got a problem when i try to rendering a bigger size image around 12000 pix x 9600 pix, i divided them into like 48 small images to render.
everything look fine when i batch render them, but problem start to come out when i check the images, some of them is not even complete, just a few pixel get rendered.
some of them are fine.
is there anything i can do to fix this problem? is it the cpu problem or my command line prb?
here is the line i write:
render -x 12000 -y 9600 -reg 0 2000 1200 2400 -r mr -im p07 -of jpg filename.mb
need help asap, many thanks
It may be a memory problem. What are the specs of your system?
I've never rendered a sequence of images that were so hi-rez before, only still images at that size. Am I correct in assuming you are rendering out an animated sequence? I don't know why else you would be rendering out a batch, but that is a really high resolution for any animation. Your explanation is a bit confusing. It also sounds like maybe you had just one big image that you were trying to render out in pieces that you could assemble later. Is that it? I'm not clear about your process. Some of those pieces are not rendering out completely? What exactly is the problem you have when rendering out the whole image?
How complex is the scene and what render features are you using?
Do you have the same problem rendering with the Maya renderer and the Mentalray renderer? Have you tried rendering to a file format other than jpeg, perhaps something not compressed?
Are you getting any error messages in Maya when it fails to render some of the frames or does it say it's complete even though you can see it isn't?
As far as your command line goes, I'm not sure if you got anything wrong there or not. I usually have most of that stuff already set in the scene's render globals and camera settings so I my commands are pretty minimal. Have you tried just running the batch render from inside Maya rather than the command line to see if that works?
thans for the quick response. i am doing a architect rendering for the printing purpose. you are correct i try to render them in small pieces then assemble them later in photoshop. since when i render this size in maya, maya crashed.
i have 140000 tris in the scene. and my system is :
intel core2 DUO CPU T9400 @2.53GHZ
nvida quadro FX 770M 512 MB
i have tried to output the images as tiff or bmp but all have the same problem. i have rendered one smaller version (6400x 4800, i divide them into 12 pieces) of this file with the same command and the outcome is great without any problem.
also it always said it is completed even though the image is only half done, here is a sample what happen
this is killing me since it takes all the time to render but nothing come out can be used.... >.<
Hmm... I wish I had a good answer for you but it just sounds like your computer, though powerful, is being overpowered by this render at that resolution. Without knowing the specifics of what you are rendering (raytracing, soft shadows, global illumination, refractions, sampling qualities, other render globals settings, etc) I can't say what you could do to optimize your render. It's possible that you have some things set in a way that is costing you a lot of render processing without benefit and that could be either turned off, reduced, or altered to lighten the load while still giving you the final result you want.
Can you upload a complete image rendered at a much lower resolution so we can see what it is you are trying render at hi-rez? And maybe list how many lights and shadows along with their settings, the render global settings and options you are using (render type, sampling, global illumination, final gathering, raytracing depths, etc.) That info will tell a lot.
It's quite possible that you have it all optimally set but maybe there are things you can do that will really lighten the load and even make it so you can successfully render the whole hi-rez image in one pass without compromising quality.
hi thanks peter.
here is the image i try to render
i have create the physical sun and sky along with one arealight and one directional light , so all together are 3 lights.
i use only 1 sample level with multi-pixel filtering set to MItchell
i only use final gathering and raytracing and for FG my setting is accuracy 32, point density is 1, point interpolation is 16.
for raytracing i just keep everything default reflection 2 , refraction2 , max trace D is 4, shadow is 2 , refl and refr blur is 1.
also the resolution i set is only 72, i have optimized the scene several times and i have around 30+ materials and texture applied to this scene.
hope this info will help
Thanks for the additional info and the picture.
Are all of the cars, plants and people just 2D textures placed on planes? From what I can see it looks like a very simple scene that shouldn't be that heavy of a render. The only place in the shot where I can see any appreciable show of soft shadows and reflected light is on the FG building on the left. At full resolution the BG buildings' lighting might still look just as good rendered without doing final gathering or any global illumination. It can be faked cheaply with some simple lights acting as subtle reflected light.
There is nowhere in the image that I see a need for any of the glass to have refractions turned on because no refraction effect would be noticeable at all with flat glass panes looking into a building in a shot like this. I would also reduce the reflection levels to 1 because there is nowhere in the shot where you would see or notice a reflection in a reflection.
The global reflection blur level probably doesn't matter because I don't think you are using it on any of the shaders in the shot. If you are, don't because I don't see where it would apply or ever be noticed in this shot.
Because we are looking in from the outside only, I would make it so none of the window glass surfaces cast shadow, just the window structures. It's a lot of calculation that probably won't give you anything noticeable. It's not that glass doesn't cast shadows in real life; it's that you would never be able to notice it in this shot.
Is there anything else in there that is casting or receiving shadows that you will never see or that is having calculations for reflecting light that you wouldn't ever notice? There probably are a number of things, especially for a still image.
All those things might reduce the processing enough to get the render to work, maybe even in one pass. You could also break this down into FG and BG and composite it in Photoshop. That would also give you some flexibility to make changes without having to render everything all over again. You could even leave some things such as the 2D people to be comped in Photoshop entirely.
I hope these suggestions help. I think you have a lot of options here.
BTW, how big are you planning on printing this?
i will try what your suggestion and see, thanks a lot for your help and i really appreciate it. i need to output the image to a 12000x9600 pixel images , so a lot puzzles to assemble.
I was just wondering how big you plan to print this. At that resolution you could print it at 40"x32" at 300 dpi which really would be a lot more dpi than you need for something that big because people don't usually view large format prints at nearly such a close distance as a small print being viewed from less than two feet away.
I only ask and point that out because you may really be able to solve the whole problem just by rendering your image at a much smaller resolution. Again, it's about optimizing things for what people looking at it will actually be able to perceive in the context that it will be shown.