Hi yoz,
I have been reading your conversation with marcus. From my point of view, I guess this is nothing to do with which behavior is correct or wrong.I guess what matter is that there is a feature which is useful to the end user (the customers...), I have been thinking about this....
1. From previous version, octane considers what blender exports.
like if the material was assigned as diffuse, or glossy, or specular. Octane reads the same. It is more convenient to assign it on blender by clicking one button.
So we have a head start and focus on the lighting and composition.
2. From current version (v 3.4) most of the materials is considered only as glossy.
so, even if we assign it as diffuse in blender , it returns as glossy in octane, then we have to change the material to diffuse again which when we click the button, it collapses the sliders and we need to collapse again to find where did it go then collapse again to adjust.this process needs more time before giving it a diffuse material.This happens on very first load or during updates or if we add new object with diffuse materials.
I believe, the previous version has this feature and i don't know why this has been removed from the latest version, which I think makes it a setback from the previous version, if there has been complains about it as discussed in the forum, I guess the proper way is to improve it or resolve it and not to remove it. co'z i think its a good and convenient feature not to be scraped from the software.
Well this is just what i think. ur ideals and enlightenment is gladly welcome.
Thanks!! chow!!
Non-offical Blender 2.56 plugin
Hi r0ug3r,
So far I see, the materials are properly set to diffuse from Blender to Octane, but unfortunately this now only happens the first time you export the material, and this is the way Refractive want it to happen.
I liked the approach I could use with 23.5, saying that what was in the internal node graph was updated at import time while what was in the custom node graph stayed untouched (aprt from the bug relinking the nodes to the wrong pins...) - and this being optioanl with the 'relink' option. What happens now is that we can't animate colors anymore for instance, we have to go through animated textures for that. Also changing textures in Blender now forces you to do the same in Octane... or manage them only in Octane (which is a shame as UV tweaking must be done in Blender anyway). IMHO it's better to keep the two synch as much as you can until you go for final render where all materials should be handled by Octane for the best render. It's even better to have the choice...
At the moment the functionalities of the plugin are frozen and I'll move it to v1.00 with less verbosity soon. I'm just waiting a couple of days or weeks to check if any new bug is discovered. I'm already working on the next-gen version of it, but I don't want to talk about the extra functionalities in case I miserabily fail in achieving them
Cheers,
Yoyoz
So far I see, the materials are properly set to diffuse from Blender to Octane, but unfortunately this now only happens the first time you export the material, and this is the way Refractive want it to happen.
I liked the approach I could use with 23.5, saying that what was in the internal node graph was updated at import time while what was in the custom node graph stayed untouched (aprt from the bug relinking the nodes to the wrong pins...) - and this being optioanl with the 'relink' option. What happens now is that we can't animate colors anymore for instance, we have to go through animated textures for that. Also changing textures in Blender now forces you to do the same in Octane... or manage them only in Octane (which is a shame as UV tweaking must be done in Blender anyway). IMHO it's better to keep the two synch as much as you can until you go for final render where all materials should be handled by Octane for the best render. It's even better to have the choice...
At the moment the functionalities of the plugin are frozen and I'll move it to v1.00 with less verbosity soon. I'm just waiting a couple of days or weeks to check if any new bug is discovered. I'm already working on the next-gen version of it, but I don't want to talk about the extra functionalities in case I miserabily fail in achieving them

Cheers,
Yoyoz
Desktop: Ubuntu 13.04 x64 - i7-3770K @ 3.5GHz - 32GB DDR3 - GTX670 2048MB
Laptop: Linux Mint 11 x64 - i7-2860QM @ 2.5GHz - 16GB DDR3 - Quadro 3000M 2GB
Software: NVidia 319.12 - Cuda 4.2.9 - Blender 2.66a
Laptop: Linux Mint 11 x64 - i7-2860QM @ 2.5GHz - 16GB DDR3 - Quadro 3000M 2GB
Software: NVidia 319.12 - Cuda 4.2.9 - Blender 2.66a
my 5 cents:
as long as there is no other possibility to transfer more material/shader/tex settings as the obj format provides,
we need as much flexibility during export from blender (or any other 3dsuite about i dont care:).
so i think the behaviour from 2.3v5 was no good choice for quick mat settings in octane -> eg its much faster to create a simple mixed-material in the side panel.
Also i think that those automatism like atm(only mat import on 1st export) are no good and should be manual cotrollable but at least better than in 2.3v5. in my competition scene there are ~hundred materials. just to much work in 2.3v5 to connect everything in the nodes to keep them save from overwritten by blender rexport.
at the end we need more control over the matset already in blender, but a change from obj to rib, colada or i dont know what, will take a while imo.
so we have to squeeze all flexibility out of the obj thing..
to explore the best way for export i am missing some more info what gets udated when? where is the uv-mapping stored) (directly in obj?) what information could be included in the mtl? what does the ocs contain and on what has blender "write access" compared to octane?
if i understood it correct, blender writes only an ocs, when there´s not already a same named one.
ocs written only at first export -> but what does that ocs include at that point additionally to an obj/mtl file?
how is the material linked to the mesh in octane?
may i change names of objects and materials in blender after the first export or will this mix everything up in octane after reoad?
didnt tried it yet because of nyrc:)
quote yoyoz:
"At the moment the functionalities of the plugin are frozen and I'll move it to v1.00 with less verbosity soon. I'm just waiting a couple of days or weeks to check if any new bug is discovered. I'm already working on the next-gen version of it, but I don't want to talk about the extra functionalities in case I miserabily fail in achieving them
"
cant wait for your news
is there any cancel animation option in sight?
regards dave
as long as there is no other possibility to transfer more material/shader/tex settings as the obj format provides,
we need as much flexibility during export from blender (or any other 3dsuite about i dont care:).
so i think the behaviour from 2.3v5 was no good choice for quick mat settings in octane -> eg its much faster to create a simple mixed-material in the side panel.
Also i think that those automatism like atm(only mat import on 1st export) are no good and should be manual cotrollable but at least better than in 2.3v5. in my competition scene there are ~hundred materials. just to much work in 2.3v5 to connect everything in the nodes to keep them save from overwritten by blender rexport.
at the end we need more control over the matset already in blender, but a change from obj to rib, colada or i dont know what, will take a while imo.
so we have to squeeze all flexibility out of the obj thing..
to explore the best way for export i am missing some more info what gets udated when? where is the uv-mapping stored) (directly in obj?) what information could be included in the mtl? what does the ocs contain and on what has blender "write access" compared to octane?
if i understood it correct, blender writes only an ocs, when there´s not already a same named one.
ocs written only at first export -> but what does that ocs include at that point additionally to an obj/mtl file?
how is the material linked to the mesh in octane?
may i change names of objects and materials in blender after the first export or will this mix everything up in octane after reoad?
didnt tried it yet because of nyrc:)
quote yoyoz:
"At the moment the functionalities of the plugin are frozen and I'll move it to v1.00 with less verbosity soon. I'm just waiting a couple of days or weeks to check if any new bug is discovered. I'm already working on the next-gen version of it, but I don't want to talk about the extra functionalities in case I miserabily fail in achieving them

cant wait for your news

is there any cancel animation option in sight?
regards dave
- Mint 10 64bit nvidia drv 260.19.29/cudatoolkit3.0 intel q6600, 4gbRAM, GTX470 1,2GB
- Mint 10 64bit/ Win7 64bit nvidia drv 260.19.29/cudatoolkit3.2 amd X6, 16gbRAM, 2x GTX580 3GB
->Octane 2.44/ Blender2.5x
- Mint 10 64bit/ Win7 64bit nvidia drv 260.19.29/cudatoolkit3.2 amd X6, 16gbRAM, 2x GTX580 3GB
->Octane 2.44/ Blender2.5x
uv-mapping is in the obj file. We're supposed to control the u/v/w scale from mtl but this is not recognized by Octanedave62 wrote:to explore the best way for export i am missing some more info what gets udated when? where is the uv-mapping stored) (directly in obj?)
- Only materials are in mtl.dave62 wrote:what information could be included in the mtl? what does the ocs contain and on what has blender "write access" compared to octane?
if i understood it correct, blender writes only an ocs, when there´s not already a same named one.
ocs written only at first export -> but what does that ocs include at that point additionally to an obj/mtl file?
how is the material linked to the mesh in octane?
- Blender doesn't write into the ocs. ocs is build by Octane based on info from obj+mtl, with material info in mtl ignored if same material already exists in ocs.
Actually this is only way forcing a material update. Mixing issues are supposed to be fixed with 24.3.dave62 wrote:may i change names of objects and materials in blender after the first export or will this mix everything up in octane after reoad?
didnt tried it yet because of nyrc:)
I'm sorry man, but you'll have to waitdave62 wrote:cant wait for your news

This is supposed to be already fixed with latest version. If not then I should go for a 'Brain Academy' session...dave62 wrote:is there any cancel animation option in sight?

Have fun!
Desktop: Ubuntu 13.04 x64 - i7-3770K @ 3.5GHz - 32GB DDR3 - GTX670 2048MB
Laptop: Linux Mint 11 x64 - i7-2860QM @ 2.5GHz - 16GB DDR3 - Quadro 3000M 2GB
Software: NVidia 319.12 - Cuda 4.2.9 - Blender 2.66a
Laptop: Linux Mint 11 x64 - i7-2860QM @ 2.5GHz - 16GB DDR3 - Quadro 3000M 2GB
Software: NVidia 319.12 - Cuda 4.2.9 - Blender 2.66a
he yoyoz,
thx for your detailed explanation!
have a nice day
dave
thx for your detailed explanation!
have a nice day
dave
- Mint 10 64bit nvidia drv 260.19.29/cudatoolkit3.0 intel q6600, 4gbRAM, GTX470 1,2GB
- Mint 10 64bit/ Win7 64bit nvidia drv 260.19.29/cudatoolkit3.2 amd X6, 16gbRAM, 2x GTX580 3GB
->Octane 2.44/ Blender2.5x
- Mint 10 64bit/ Win7 64bit nvidia drv 260.19.29/cudatoolkit3.2 amd X6, 16gbRAM, 2x GTX580 3GB
->Octane 2.44/ Blender2.5x
hey yoyoz.
really love your exporter, exspecially the animation part..
some idea: normally the exporter uses the render settings of the subsurf modifier set under "render" not "view".
which is correct at all. when you globally lower the subsurfdiv under "simplify" (in blender 2.5 below global gravity settings) - your script is exporting the lowerd subdiv settings, which is also correct and real great to quickly reduce polycount while sped up export. for the final render you push it up again if enough vram available..
now while animating we always have to voxelize which is very time consuming on heavy meshes...
-> couldnt the subdiv level depend on the distance between object and cam? maybe the max value in the modifier with "render" for close ups and a low value with "view" for far range shots. have you access during export on the "view" settings of the subdiv mod?
i hope this is understandable ->cam z-depth dynamic scene-subdiv-setings: near distanced object have high values, far distanced lower. all frame to frame dynamic.
what do you think?
really love your exporter, exspecially the animation part..
some idea: normally the exporter uses the render settings of the subsurf modifier set under "render" not "view".
which is correct at all. when you globally lower the subsurfdiv under "simplify" (in blender 2.5 below global gravity settings) - your script is exporting the lowerd subdiv settings, which is also correct and real great to quickly reduce polycount while sped up export. for the final render you push it up again if enough vram available..
now while animating we always have to voxelize which is very time consuming on heavy meshes...
-> couldnt the subdiv level depend on the distance between object and cam? maybe the max value in the modifier with "render" for close ups and a low value with "view" for far range shots. have you access during export on the "view" settings of the subdiv mod?
i hope this is understandable ->cam z-depth dynamic scene-subdiv-setings: near distanced object have high values, far distanced lower. all frame to frame dynamic.
what do you think?
- Mint 10 64bit nvidia drv 260.19.29/cudatoolkit3.0 intel q6600, 4gbRAM, GTX470 1,2GB
- Mint 10 64bit/ Win7 64bit nvidia drv 260.19.29/cudatoolkit3.2 amd X6, 16gbRAM, 2x GTX580 3GB
->Octane 2.44/ Blender2.5x
- Mint 10 64bit/ Win7 64bit nvidia drv 260.19.29/cudatoolkit3.2 amd X6, 16gbRAM, 2x GTX580 3GB
->Octane 2.44/ Blender2.5x
Hi Dave,
LoD is something that might work with 2.56a but it won't be possible with final release. We get the transformed mesh with parameters saying if we want to apply modifers (yes or no) and which version we want (preview or render). Applying or not the 'simplify' settings on a per-mesh base would mean writing into the Blender context during rendering. This may be done with some workaround wth 2.56a, but no more in some recent builds (this changes from time to time), and from the info I have it will probably not be allowed in final release.
I hope that Refractive will find a way to multithread the voxelization, but it's probably more tricky than it sounds.
Yoyoz.
LoD is something that might work with 2.56a but it won't be possible with final release. We get the transformed mesh with parameters saying if we want to apply modifers (yes or no) and which version we want (preview or render). Applying or not the 'simplify' settings on a per-mesh base would mean writing into the Blender context during rendering. This may be done with some workaround wth 2.56a, but no more in some recent builds (this changes from time to time), and from the info I have it will probably not be allowed in final release.
I hope that Refractive will find a way to multithread the voxelization, but it's probably more tricky than it sounds.
Yoyoz.
Desktop: Ubuntu 13.04 x64 - i7-3770K @ 3.5GHz - 32GB DDR3 - GTX670 2048MB
Laptop: Linux Mint 11 x64 - i7-2860QM @ 2.5GHz - 16GB DDR3 - Quadro 3000M 2GB
Software: NVidia 319.12 - Cuda 4.2.9 - Blender 2.66a
Laptop: Linux Mint 11 x64 - i7-2860QM @ 2.5GHz - 16GB DDR3 - Quadro 3000M 2GB
Software: NVidia 319.12 - Cuda 4.2.9 - Blender 2.66a
sometimes my english doesnt serves me well..yoyoz wrote:...We get the transformed mesh with parameters saying if we want to apply modifers (yes or no) and which version we want (preview or render)...
does this mean you have access to which one gets applied to the mesh during export? depending on the z-depth of cam it would be exactly what i mean. forget about the simplify option, just leave it disabled.
i imagine an input xx (option): more distance than xx take "preview", less dist than xx take "render"
yes and additionally multithreaded voxelizing

but of course i´ve no plan if you say it would be unpossible (or just able with crippled workarounds) in final 2.5x, than i´ll of course believe:)
thx dave
- Mint 10 64bit nvidia drv 260.19.29/cudatoolkit3.0 intel q6600, 4gbRAM, GTX470 1,2GB
- Mint 10 64bit/ Win7 64bit nvidia drv 260.19.29/cudatoolkit3.2 amd X6, 16gbRAM, 2x GTX580 3GB
->Octane 2.44/ Blender2.5x
- Mint 10 64bit/ Win7 64bit nvidia drv 260.19.29/cudatoolkit3.2 amd X6, 16gbRAM, 2x GTX580 3GB
->Octane 2.44/ Blender2.5x
Hi Dave,
this applies to all modifiers, including particle instances, mirrors, curves, lattices, etc... it's nothing or everything, so very risky if mesh as modifers other than subsurf.
But the idea was nice!
Lionel
this applies to all modifiers, including particle instances, mirrors, curves, lattices, etc... it's nothing or everything, so very risky if mesh as modifers other than subsurf.
But the idea was nice!
Lionel
Desktop: Ubuntu 13.04 x64 - i7-3770K @ 3.5GHz - 32GB DDR3 - GTX670 2048MB
Laptop: Linux Mint 11 x64 - i7-2860QM @ 2.5GHz - 16GB DDR3 - Quadro 3000M 2GB
Software: NVidia 319.12 - Cuda 4.2.9 - Blender 2.66a
Laptop: Linux Mint 11 x64 - i7-2860QM @ 2.5GHz - 16GB DDR3 - Quadro 3000M 2GB
Software: NVidia 319.12 - Cuda 4.2.9 - Blender 2.66a
Not sure if this is a plugin issue or not, but I can't get turntable animations to work properly. The camera is rotating around its own axis rather than around the axis of the object I am picking with the focus picker. Using 0.93 and 2.56a.
Anyone else experiencing this? Or am I just doing something dumb?
Anyone else experiencing this? Or am I just doing something dumb?
Ubuntu 9.10 x64 | GTS 240 | 260.19.44 drivers | 3.0 Toolkit | Dual-Core 2.4 GHz | 4Gb| Blender 2.56a/latest Yoyoz plugin
WinXP32 | GTS 240 | 266.58 drivers | 3.0 Toolkit | Dual-Core 2.4 GHz | 4Gb | Blender 2.56a/latest Yoyoz plugin
WinXP32 | GTS 240 | 266.58 drivers | 3.0 Toolkit | Dual-Core 2.4 GHz | 4Gb | Blender 2.56a/latest Yoyoz plugin