robcat2075

I didn't digest the entire "hungry renderer" description, but it sounds like hyperthreading as explained by Yves is a way to keep the CPU always working on something.

You might try a reset all settings in A:M then relaunch Netrender. Is that the latest v17d?

-Z-buffered Kleig light shadows may render faster than Ray-traced Kleig light shadows -Sun light shadows are ray-traced only but a distant Kleig light will look similar. -Turning "Cast shadows" OFF on hair will render faster. -Complex material can be baked into a bitmap for faster rendering. -Fewer lights render faster than more lights (mostly true with ray traced lights) -Regular render has anti-aliasing equivalent to 16 pass render, but is much faster.

they ought to be there but you could save a render Preset from the A:M render panel and use that in Netrender.

I looked at the last sample clip. Is it all like that? If you have any ray traced lights change them to z-buffered. If there are complex Materials on the models baking them might help.

Hard to know without seeing it. I could look at it if you want but you'd have to zip up EVERY asset it needs.

to bake hair right click in the chor window and choose "bake particles". the baking files will be written where ever the PRJ was saved. However your cloth hair looks promising too.

Most of those splines over the top of the head could be ended sooner with hooks.

If you turn off "Cast shadows" for hair it will render faster

For what it is worth...FakeAOGPU works for me in 64-bit v17 both as a render time post effect and as a post effect in a Composite However, both FakeAOGPU and FakeAOCPU crash my 32-bit v17 One other work-around for people who want to use 32-bit v17 but also want to have FakeAO.... Render without FakeAO in V17 to OpenEXR with a depth buffer, then use that OpenEXR file in a Composite in v16 to add the FakeAO as a Post effect. Of course this is more efficient for image sequences than single images, but if you tend to use FakeAO in Post anyway, like me, it's a minor commute.

Does this computer have a touch screen? I'd be curious to know if any of the touch interface gestures work with A:M.

For what it is worth...FakeAOGPU works for me in 64-bit v17 both as a render time post effect and as a post effect in a Composite However, both FakeAOGPU and FakeAOCPU crash my 32-bit v17

I'll note that you may have more success with modeled holes if the 5-pointers are from fairly squarish patches and if they come up to a ring around the hole rather than try to be the edge of the hole itself.

So make the object not-green?

OK, you made me install 32-bit v17 again. In a three way race I got v16b 32-bit 7:34 v16b 64-bit 5:57 v17c 32-bit 8:06 v17d 64-bit 5:05 So perhaps 32-bit v17 is a bit slower than v16 edit: I ran another test with 64-bit v16 in the mix and added the result above. So 64-bit v17 is faster than 64-bit v16 but 32-bit v17 is slower than 32-bit v16 In all cases the 64-bits are faster than the 32-bits. It's POSSIBLE that my and Nancy's results for the 32-bit versions are handicapped because we are running them on 64-bit CPUs.

Something is weird happening around the corner of the mouth. It looks like your smile line is connecting to the bottom edge of the lip. I'd have run it down to something below the lip.

The Frequency% is how much of the top clock speed the CPU is running at regardless of whether it is churning through any code. I have a 2.4MHx CPU but right now it is idling at only 1.6MHz. The usage% is how much of the available processing time is spent doing real stuff, regardless of what speed it is doing it at. If you are running One A:M on a two core machine 50% would be an expected number (100% use of 1 core out of two). your results sound reliable. I'd have to install 32-bit v17 to investigate more.

First, I'd find the Resource Monitor (I have no idea what it is called in Win XP) that tells what percentage of maximum frequency your PC is running at . Maybe yours is always at 100%. If it is, it is. But mine throttles quite a bit from 66% to 100%. I have to get 2 A:Ms rendering simultaneously to get it up to 100% (This is not the same as "CPU Usage") I'm not convinced yet that v17 is what has caused the momentary display freezing in my PC If you need your OS for your peripherals then that's a reason to stick with it or start scoping out ebay for a replacement. Used cintiqs can go for not much.

Nested, semi-transparent spheres can make a glow around another sphere that will scale with distance. In this case the spheres have a gradient material on them that makes the edge completely transparent and the center nearly transparent. SpaceOrbs.prj

Try moving the porcelain material to higher in the groups folder than the color material are.

Basically what Gerry did! The 5-pointers are not closing because the center spline at the top has two CPs in the same location. Select it, press teh detach button and you will be able to drag it away and delete it.

You will need the cylinder to be splined so that point where you want to punch a hole presents 2x2 arrangement of fairly squarish patches. Then you can splice in a four-point loop around the center point, break the splines in the middle of the new circle, adjust the biases so it's appropriately round and extrude it through to the inide where you have done the same thing on that surface. A boolean is easier.

I don't know Nancy's exact parameters but I'll note that I've gotten unexpected results when benchmarking with just "one" A:M running. Apparently just running A:M by itself isn't perceived as enough of a panic situation (on my 4 core machine) for the OS to always run the CPU at full speed. Once I did get a result that o newer version was slower than an older version If I run two A:Ms at once that will force the CPUs to run at 100% speed and i will get better apples to apples time comparisons.

First, I would fix the splining so that there are no dead end splines, In the middle of the pic there s a CP that appears to have five splines coming out of it. At best, that is two splines crossing (good) plus one spline ending on that same CP. That will always be bad, porcelain or not. I suspect I see something like that where the leg joins the hip, also. Then we could figure the porcelain out.

Here's a post from the Benchmarks thread with a speed comparison of v15, v16 and v17 http://www.hash.com/forums/index.php?s=&am...st&p=365392 A:M comes with NetRender now so if you have a multi-core PC you can get rendered faster with that also.