View Full Version : Mray/ "-fb_virtual cache" parameter for huge print res
02-28-2011, 09:53 AM
I need to render huge picture (cca 10k high and 7K wide) and mray do not want to render it....so I guess its limit of virtual framebuffer size/memory...also tried in batch render
I went through the mr manual and found an interesting rendering parameter when using mray from command line.
mr -fb_virtual (on/off/cache) ...the default is on which store whole fbuffer into physical memory and the cache option should write every render bucket/buffer on hard drive so no extra memory needed > as manual says it should be used for extra large framebuffers in terms of pixel counts...
However whenever I try to use it in cmdline it doesnt recognize this parameter so I d like to know if anyone used it before.
Maybe its only in mray standalone but this info was in the mray for maya docs so I guess it should work in Maya batch command!
Im running Maya 2011 64bit, Win7 (all maya spacks installed)
Thanks for any info!
02-28-2011, 11:34 AM
For rendering large images first of all you should disable "Export Post Effects" in Render Globals, otherwise Maya will crash after 100%.
You may switch buffer mode via mel command:
setAttr mentalrayGlobals.frameBufferMode 2;
0 = off, 1 = on, 2 = cache
02-28-2011, 01:35 PM
Hi! Thanks for your input Pavel!
Well Im not sure if I used it correctly I put the script into pre render MEL scripts in RenderGlobals (I have tried different types (prerender mel, prerender layer, prerender frame) but every time my batch render crashed :(
here is a bit of process output message (just before the crash occurs - it doesnt start to render at all)...must be something with the size of fbuffer cos when I submit render at 80 percent reduced size the render goes normal and finnish....
here is a bit of a progress msg:
"RCI 0.2 progr: begin intersection preprocessing
RCI 0.2 info : using scanline for eye rays
RCI 0.2 info : using BSP2 for secondary rays
RCI 0.2 progr: end intersection preprocessing
RCI 0.2 info : wallclock 0:00:00.00 for intersection prep.
RCI 0.2 info : allocated 2152 MB, max resident 2152 MB
RCLM 0.2 progr: lightmap vertex run, shader 0x73c: misss_fast_lmap_maya1 on instance 23:0
PHEN 0.2 progr: misss_lightmap_write_init: Got shader init
PHEN 0.2 progr: misss_lightmap_write_init: Got instance init 309441048
PHEN 0.2 progr: misss_lightmap_create: starts w. tags 1757 and 0
RCLM 0.6 progr: lightmap vertex run, shader 0x73c: misss_fast_lmap_maya1 on instance 13:0
RCLM 0.8 progr: lightmap vertex run, shader 0x73c: misss_fast_lmap_maya1 on instance 10:0
Fatal Error. Attempting to save in C:/DOCUME~1/smedge/LOCALS~1/Temp/smedge.20110228.1426.ma
--- Attaching to process stderr stream ---
// Maya exited with status 1"
Thanks again for your kind response!
02-28-2011, 05:53 PM
You can also change your framebuffer cache mechanism under the miDefaultOptions. (You can select this in the select by name box by typing: miD*) Under the mentalrayGlobals look for the Extra Attributes rollout at the bottom. You will find Frame Buffer Mode. Use this.
I am not sure that will conflict with what Pavel mentioned.
It's also possible your lightmap is huge. It's based on screen resolution and the container itself for a lightmap may have an internal limit. I am not sure of that but might be worth investigating.
Almost forgot: turn off Scanline (for everything) Select raytrace. This will help your memory consumption.
02-28-2011, 08:28 PM
I'm pretty sure there is a cap for the framebuffer of 2 GB
why not split the image with -reg flag
03-04-2011, 11:51 PM
great tips Bitter!! I can think of the lightmap which is res dependend ...Im using misss_fast shader in there so the default lightmap size is expression driven probably causing a lot of memory consumption....also the scanline tip is interesting!!
Ill give it a try tommorrow (I need to finnish about 3 10k renders during weekend!) I hope much that it will help.
Thanks again for these tips!
03-05-2011, 06:31 PM
Great tips which also worked!! Thanks much Bitter I owe you a beer :beer:
Short info: my render was 9933x7016px which caused expression driven misss lightmap explode to 14500px height in full 32bit float.....read huge eating of ram
also switching to pure raytrace mode helped.
03-05-2011, 06:31 PM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.
vBulletin v3.0.5, Copyright ©2000-2013, Jelsoft Enterprises Ltd.