Thanks, I eventually found it there. Seems that the helpfile is mapped to the classic version of adobe acrobat -- the first computer i tried viewing it on didn't have classic installed -- so it got all confused. i'm still trying to figure out why eovia would choose to embed the pdf inside the application, rather than making it available to say copy to another computer so you can read it in on-screen on one machine while working on another one. and in typical eovia fashion, the manual lacks many of the specifics that would make it useful -- like details on how the network render works -- does it treat mulit-processor machines as one or two nodes; is there a performance loss when a really slow computer is in the render chain; is there an advantage to having machines of equivalent speed or processor type; etc. now i have the jolly task of experimenting to figure this stuff out on my own. but hey, at least they threw the feature in. i suppose that's better than nothing. -scott

Attached Link: http://homepage.mac.com/scottspringer/resources/discussion/water_composite.html

Unfortunately, if you're not lined up exactly, you may see a gap between the water and the boat hull. Which would ruin your illusion.

I think the simplest, fastest, almost no-brainer way to solve this problem is to use compositing. If you're making a still image, it's perfect. If you are making an animation, it's a wee bit trickier, but still doable.

http://homepage.mac.com/scottspringer/resources/discussion/water_composite.html

Follow this link to see a quick example of what I'm talking about.

Using previous models I built, I was able to animate, render and composite this animation (using Photoshop of all things!) in about 2 hours of total work.

Most of the time was spent screwing around w/ Photoshop (for some reason my copy of QuickTime Pro stopped exporting otherwise I would have used that.

I thought it said in the Carrara manual that the size/scale of lights and cameras is irrelevant in the 3D environment. If you're using a point light, the light should eminate from a single point, i'm guessing the center of the the light. I may be wrong, because I'm also guessing, but I don't thing changing the scale will affect this any. -Scott

One thing to keep in mind when using the transparency channel to mask away unwanted geometry is that the geometry is still there. If you apply certain post-render effects to your scene, you may see the outline of the original plane. Also, if your lights are set to use soft shadows, the cast shadow will be that of the actual geometry - regardless of the transparency. Until Carrara supports clip mapping (using the alpha image to actually cut-away the geometry) You might want to consider tracing your alpha in illustrator and then making a spline object out of it. Then you could stick your image on it and it should fit perfectly. - Though you only need to go to this legnth if you have problems with the first method. -Scott

pinklet, dammit, i just made a mess on my keyboard!!! finally they release upgrades for machines that really need them. anyway, about my dual 1.25 -- while it was a rev1 and still boots in OS9, you can't use SMP in OS9, so for Carrara I would stick to X. Depending on how fast the G3 is when/if I upgrade it, I'm thinking about using it to model scenes while my dualie renders. As far as the e-mac is concerned, yes the price and power of it is tempting, but the lack of dual monitor support relegates it to a render box. and if that is all you use it for, you're probably better off with an original imac, since 3D makes the G4 useless. overall, it has never been cost effective to upgrade processors. you would get a much better deal buying any of the quick silver macs than you would spending it on an upgrade for a beige box. that being said, i'm still gonna have to get that Ghz G4 - (when i checked it was priced at $599), but look at the fans on that thing!!! i'll bet it will rival the noise of my dual1.25. Maybe the G3800 is a better idea after all. hmmm. -scott

OK dude, check this out...

I also have a beige G3 266, and despite having a dual 1.25ghz, all i can think about at night is pimping out my beige beast.

i think that the 800 Mhz G3 card is what you and i both want. -- particularily if you're going to be doing 3D on it, which gets no benefit from Altivec (the thing that makes a g4 so much faster than a g3). So pop that puppy in.

Then get at least 512 MB of RAM in that thing. Apple says that the early beige machines can only take a max of 3x128 chips for a total of 384 MB, but you can get away with 3x256 for 768. There is some spec you need to make sure the chip meets (which i forget, but will get to you shortly) or you can just buy RAM that is certified for all Beige G3's. (A 256MB stick should run about $30)

then decide whether this is a render box, or a machine you want to model and manipulate scenes on. if you want to actually work on scenes, your next big bottleneck is the graphics card. if you're just rendering final images, the calculations are all done on the CPU, so you don't need to blow your dough on the GPU.

most of these beige machines shipped with a piddly 2MB of integrated ATI Rage 128 video acceleration. I would recommend buying the 4MB upgrade chip ($29.99), and then also picking up the ATI 7000 PCI graphics board ($119) This board sucks by today's standards but is the only Mac-compatible board that I have been able to find that has a PCI interface. Since we don't have AGP, we're kind of SOL.

The reason you want to drop 30 beans on the 4MB upgrade is so you can run 2 monitors, which in my opinion allows you to work much faster.

If you still have money after that, consider getting TempoTrio ($179) (USB2, FireWire & ATA133 all on one card) This gives you lots of options for storage, without using up all the precious PCI slots. If it is too much money for you, consider just a regular ATA133 card ($84) Add a fast hard drive to that, and you'll get much faster virtual memory access. Carrara uses alot of virtual memory, so this should improve render times noticeably. Don't bother with a RAID card. I don't think you'll add any real world speed increase with it for the kind of work you'll be doing.

If you are sending files back and forth to another computer, drop in a 10/100 ethernet card ($15-30) into the last remaining PCI slot, so that you aren't restricted by the 10base-t of the built-in ethernet. You'll be very happy you did.

So these are the upgrades that I've been looking at for my G3. It totals about 750, but it is just so tempting to plunk money into that machine... Anyway, I hope that this helps you make your decisions. Please let me know how any of this equipment works out for you if you do choose to upgrade, and I'll do the same for you when I get some of this stuff.

-Scott

I've also experienced strangeness in the shader preview with really large objects, or ones with wacked hot point locations. my solution is usually to track out really far, and use zoom to come back in. sometimes this solves the distortion of the geometry. like the others, sometimes a reloading of carrara is necessary.

Also another thing to keep in mind when working with transparency in carrara is that soft shadows do not respect the transparency channel. only raytraced shadows do.

this may not be at play in your case (since we haven't seen the image yet) but i thought i'd atleast give you another possibility to investigate.

-scott

i used to use infini-D as well. and for the first month of usig carrara 1.1 i swore that i would still probably do most of my work in infini-D. i was basically just used to infini-D. Once I got going with carrara, i never even thought about going back.

back in highschool when i first got into computer graphics (on a centris 660av) i was using Aldus SuperPaint. about a year later, my dad bought a scanner that came bundled with photoshop 2.5. It took me about another year before i was actually using photoshop.

i think you'll have a similar experience when ever you upgrade to a more powerful product. when you know how to get results in a lower-end product it can often be faster than stumbling around in a more powerful one.

at least that's my take on it.

-pix

last time i checked, the alpha channels (g-buffers) weren't anti-aliased. so if you are using post render effects based on them (depth of field, light cone, lens flare, etc.) you will probably see aliasing in your final render that wouldn't be present if there were no post-render effects.

kix is right that carrara has long had poor anti-aliasing. I generally render my images 2 or 3 times larger than necessary and then scale down in photoshop. This will soften your g-buffers too.

-pix

for saving png, gif, jpg files in photoshop, "file: save for web" is the nicest way to compress files.

it allows you to have lots of control over the compression settings. You can look at multiple compression settings simultaneously, and save compression settings.

-pix

thanks for all the comments.

a note to PAGzone, I couldn't agree with you more about macs being a more elegant solution. and I have dual 1.25 right now that is pretty fast at rendering, but my needs in computing have changed, and I find myself needing portability.

so I'm getting a powerbook, selling the dual 1.25, and grabbing a reasonably priced pc to do the rendering on.

this way i can still have desktop power when i absolutely need it, but most of the time i will be still using a mac.

about new macs -- i can't wait. but i'm really excited about the machines that will eventually be coming out with IBM powerpc chips. that won't be for a couple years. sigh.

anyway thanks for all the advice.

-scott

Geeez. Of course i'll post it. (provided i finish it) still have much modeling to do. don't hold your breath. : )

Thanks for the tip on quasi-clipmapping. i guess i was afraid to try that earlier cause i'm splineless.

oh dear, i didn't just say that did i?

-scott

with these kind of things, it always helps if you post an image.

but let me try anyway...

shadow catchers are used in conjunction with composited live footage or backdrops. basically you model a primitive object of something that appears in your backdrop photo, line it up in the right place, and set it to be a shadowcatcher. it will render invisible, except for the shadowed area, which will appear to fall on the objects in the backdrop.

hopefully that makes sense.

since you've modeled the actual room, you don't want to be using shadowcatchers. objects in carrara will always accept shadows, as long as the objects or lights that would create said shadows are set to "cast shadows."

hopefully this helps produce more desireable results.

-pix

I don't have 1.1 anymore, but dupe with symmetry seems to work fine in 2.1 in both the assembly room and the metaball modeler.