Posts made by panixia

panixia

with the way things currently are set up, we have to basically restart the whole scene parsing process again for each scene if we want to account for visibility changes, which slows down animation renders considerably.

i understand, i suspected it.. but maybe would be possible to add it as an option to tick, so that it is not default, but, if you really need it from time to time, you may try to render the animation properly even if slowly.. i don't know, you have to tell, but i think this could be at least an half way solution..
as we say in italy "meglio feriti che morti" ---translation--> "is better to be hurt than to be died"

very thank for your answering

panixia

@dkendig said:

it will be in the next release: http://screencast.com/t/mqfvvDv390Ev

what about the layer visibility option to support object animation plugins like twilight does? will it be possible in the next VfSU?

panixia

@andybot said:

the fresnel reflection might be a checkbox (instead of having to load a separate TexFresnel)

peraphs the separated TextFresnel is needed for adding it in the perpendicular sub-slot of specular map.. anyway i agree it should be default in reflection, TexNone is good for mirrors only..

panixia

maybe this http://www.holmes3d.net/graphics/subdivision/ could help a bit, even if very basic and contains some greek letters i found it some times ago.. it made me suspect that the original catmull-clark algorithm is intended to work on a quad cage, while the loop subdivision triangulates it all.. i think the loop subdivs is the one behind the existing subdivision SU plug-ins (both free and commercial) it seems that a "quadrified" catmull-clark would be nicer and cleaner.. anyway.. i will search further, maybe if i fiund something nicer, we can open an "official" dedicated thread for this request.. i know is not easy, but i know it isn't impossible for you, man.. you have the technology!

panixia

something like this, maybe?

http://rosettacode.org/wiki/Catmull%E2%80%93Clark_subdivision_surface

http://www.rorydriscoll.com/2008/08/01/catmull-clark-subdivision-the-basics/

panixia

isn't the RT feature also a better sign?
will the "signs" become reality? hope so..

panixia

ah ok, useful tip, anyway its' original model had material tiled very large (0,005 repeat, as already stated) mapped on a mesh with very small subdivisions 5mmx5mm nonquad-quad faces, i'm not sure but i think the problem could be related to the fact that the single facets are to small for the displacement to properly work.. am i wrong?
or maybe the strange behaviour of his mesh can related to the not so clean mesh buiding.. don't know how he build the surface, but i found sort of overlappin edges and so on.. i also tried to turn it quadface whit your brilliant tool, to be sure the mapping was uniform, but it didn't work, it follows diagonals here and there, even if apparently the mesh is, how can i say, a full square-cell grid, with no poles/odd topology.. didn't untderstand why it reacts in strange way..
as a general statement i noticed that the cleaner the mesh, the smooter the displacement will work..

ps. any idea of further developments on quadface tools? loved it! thanks again.. [PLUGINREQUESTMODE/ON]Never planned to do a sort of catmull-clark plug based on quadfaces? That would be ultra-powerful indeed[PLUGINREQUESTMODE/OFF]

panixia

@thomthom said:

@newxieland said:

Think it's back in VfR ... right?

it does.

panixia

can you elaborate the concept?
are you saying if i use 1cmx1cm texture it will render 2,45cmx2,45cm or something like this?? i hadn't noticed it maybe..
anyway i found the wole modeling/scaling/txture repeating thing a lot messy in that model..

panixia

1 your mesh is very messy, i think you should try to keep your modeling a bit cleaner maybe..

2 why the hell you model in mm with a precision of 0,0mm and then put 0.005 in the repeat of your displacement map??? and why model a 20x25cm sea wawe? try to use more realistic and consistent units, so your displace will behave in a more predictable way..

3 next time try to attach the packed vismat, or at least the diffuse/displace/bump/whatever maps

4 when you displace, i would suggest to force your default group material to be the same as the displaced material, this will prevent the grey patches (or whatever your "global" default material is) when the mesh is not 100% continuously mapped

5 i did't changed the subdivision, but i think this is waaaaay overmodeled.. why do you need ALL that subdivisions in the mesh?

anyway i fixed it in some way.. just copied your curves, welded them, scaled them 4x, skinned them with curviloft, checked "make quads", added your original subdvisions to the mesh using quadfacetools-> connect edges, played a bit with trupaint and a decent vismat sea i made wich is attached too..
all simple and clean.. also i uncheked "view dependent" cause i like the displacement to be the same at any viewing distance..

oh and in addition, tried to render your model adding a simple random displacement map and it took forever to render in the ugly dicontinuous way, using merely 60% of CPU on i7, now it renders fast as the hell..

hope it will help

samplewawe fixed.rar

panixia

i hope layer change will be supported in the so waited new version..

panixia

@valerostudio said:

go to each frame and render individually

not suitable at all to my purposes

panixia

just forget to link the video...
http://www.youtube.com/watch?v=n83qNwwFvZ0

panixia

i'm searching here and there to find a workaround to render moving objects in vray.
the general opinion is that it is not possible, due to the fact vray doesn't support visible layer update in batch rendering.
i read about this here: http://forums.sketchucation.com/viewtopic.php?f=323&t=33417

then I found this video and it shows that SU-animate can do the job!
i wonder how it is possible if the problem is on the vray side????

anyone tried this? or maybe the new VFSU version (1.49.don'trememberexactlyi'venotvrayonthispc) solved the layer-visibility-update issue??

i know twilight can do the job easily and i tried it a couple of time far ago and it worked, BUT the boss has vray on his machines at work and (most important) i don't like twilight at all, so I would like to stay with the VFSU solution as much as possible.

can anyone help me please? i did't find any topic about SUanimate in this forum..

grazie

sorry for my bad english..

panixia

thanks tomhtom for the useful "adding edges" trick. you are plain genius.
oh and very thanks for quad face tools. it actually made my life a little bit better!

panixia

will the new VFSU be RT????

panixia

good, great exposure for sure.. but IMHO not the best looking ramp renders ever, to be honest.

panixia

i did a slight variation to the proxy to obtain a quad-mesh most confortable for texture mapping, faster selections and further editing and it works fine for me.. hope it can help someone else for mapping leather, fabrics, etc. on it.
quadface tools rocks!

[xxl-img:2cc3ed4y]http://img705.imageshack.us/img705/4575/02cubettiimbottititest.jpg[/xxl-img:2cc3ed4y]

panixia

Hi everybody! Recently I did this render (godzilla not included =D )
I actually don't like their project so much, but they like it and this is what they want.. and they lead.. so..
Anyway I'm not posting it in relation to the fact of it being beautiful wich is absolutely not (IMHO) but I'm considering the heavyness of the render itself..
I'm not sure this can be considered an heavy sketchup+vray render, but I can ensure this is pretty heavy for me.
I mean, till a short ago, I was not able to complete such heavy renders, due to the fact the render often crashed cause the 32 bit lack of ram issue..

Actually here you have a lot of high poly geometry, 737 components repeated in 10700 instances or so + about 900 groups, tons of lights + invisible lights,
about 140 materials with many bumps, transparency, displacements, blurry reflections, quite high render engine parameters, lens distorion, hi-res etc..etc..

It took me here on sketchucation reading interesting discussions about handling those lack of memory issues whith some tricks and workaround I didn't know and most of them worked really nice,
for example the fact that you should subdivide a mesh for better displacement performance, or that component modeling is the key,
or that you can smartly place png elements for vegetation and characters instead of modeled ones etc..
anyway I think it could be also better generally optimized with other triks I don't understand well like the whole saving exr, saving maps and channels stuff..
or maybe using ies lights (I'm not able) could save some memory instead of using tons of hidden spot/rectangle lights to produce the light effects they wanted on walls..

I think if would be useful if we post our most heavy works to discuss the proper general workarounds to fight agains the big bad memory limit issue..

[xxl-img:11fi66xb]http://img545.imageshack.us/img545/470/godzillab.jpg[/xxl-img:11fi66xb]

[xxl-img:11fi66xb]http://img842.imageshack.us/img842/7746/renderpiantanotturno.jpg[/xxl-img:11fi66xb]

[xxl-img:11fi66xb]http://img254.imageshack.us/img254/4526/aereacortilenotturno.jpg[/xxl-img:11fi66xb]

[xxl-img:11fi66xb]http://img585.imageshack.us/img585/1489/plasticobig.jpg[/xxl-img:11fi66xb]

[xxl-img:11fi66xb]http://img232.imageshack.us/img232/6707/aereasudnotturna.jpg[/xxl-img:11fi66xb]