Hi,
i have a little question for those who use the TailorTool.
How solid is it? Can it handle very dense models?
The model i am provided with is composed of several parts, but should be treated as one, polycount is about 900.000.
To remodel it for the desired cut-away would be very time consuming, time i don’t have.
thanks.
uwe
Hi, Uwe
Tomas Egger recently tested TT with car model (a lot of groups and like 700K polys)
it does work, but on bigger models be prepared for extremely long render times.
I did 2 very simple scenes, the only difference was I increased the poly count on the model
at 1800 polys it took 14 seconds to render a frame, when I increased it to over 500,000 it took more than 14 minutes. I tried to use it on a model with almost 2 million polys with phong shadows and shaders. I shut it down after an hour and it was still doing the shadows.
I also tried Da Vinci (Northern Lights) it works in a very different way, but I couldn’t get to work on very big models. Instead woking with clip maps it actually creates a model from the child objects. you can preview the boolean object in EI and export it as a fact. That could speed up the rendering, but on very big models I couldn’t get it to create the model in less time it took to just rebuild it.
bw
Hi, Brian W.
The render times looks too big even for our G4 733. A possible reason you use illuminator light (clipping is calculated for each shadow). If your scenes are simple you can archive them and post here, we’ll try to help
Hi Igors - I’m not using any Illuminators, and the simple scene I set up to test only has 1 spot light and no shadows. The test I set up was from 2003 and I think I was on my G4 933, and, EI has been updated. I opened up the old files and tried it on my new G5 QuadCore 2.5 - and EIAS 6.5.2.
All the tests use UberShapes Capsule and Cube (the cube subtracts from the capsule). The cube stays at 6 polygons and to increase the total polys in the scene I increased the density on the capsule. 1 spotlight, no shadows, phong rendering, no shaders (other than TT), no reflections.
results:
no tailor tool, simple clip map on the capsule, 539,000 faces
G4 933 = 14 seconds
G5 2.5 = 5 seconds
tailor tool, 1800 faces
G4 933 = 14 seconds
G5 2.5 = 4 seconds
tailor tool, 539,000 faces
G4 933 = 14 minutes
G5 2.5 = 24 minutes
tailor tool, 1,078,000 faces
G4 933 = NA
G5 2.5 = 2 hours 3 minutes
I can’t upload files here - but I can email the files if you want. I’ve had TT for a couple of years, and I haven’t really used it because of this. I woud really like to if we could figure out why it’s doing this.
thanks,
brian
Hi, Brian W.
Ok, please collect all files, archive and send to mail we’ll inform you with private msg
brian, igors, thanks for your input on this.
these rendertimes are not really what i am looking for. nowadays customers are accustomed to GI lighting or ambient occlusion and i can only imagine how long the rendertimes will be with that turned on. has anyone tried it?
oh, ok. thanks igors!
i guess i have to go the oldschool way via animated clip-maps and seperate render-passes.
any chance tailortool will be updated in the future for RT? is it even possible with rendertime-booleans?
it sounds like a nice thing, if it works, but being limited to phong is a real drawback.
a lot can be accomblished with phong only, but with the availability of fast GI in EIAS and faster computers there will be some need for this kind of thing working with RT (GI).
Is it possible to output a sequence of facts files using TT?
I am thinkin if you can do it, you could possibly then use GI etc on the imported sequence!
or would it be too big tohandle all the models, for say, 25fps?
Hi, antnee (sorry we don’t know your real name)
No, render-time actions/operations like clipping, displacement etc. cannot be saved
We can say nothing about TT upgrade perspectives, sorry. But the “nature” of TT is 100% ray-traced, not phong 
Any RT engine automatically collects information about all volumes intersected by a ray. And all what a boolean plug-in/shader needs is to calculate odd/even of intersections count and set clip if needed. However, there are no chances to have RT info available, so TT implements its own “mini-raytracer” that’s, of course, is much worse, eats a lot of memory and suitable for phong only. That’s a normal way of development
Hi, Brian W.
Thx, we’ve received and tested the project you sent. Yes, the render time is really slow. Unfortunately, we cannot help here, cause every plug-in becomes same slow if it has child group(s) with 1Mb (and more) facets/vertices (check standard Dicer! for example). We are sorry about this host’s problem
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.