sneak

[pedrofelipe]

well not really everyday but at least a development blog updated every weekend would be highly appriciated. Then at least we know what to expect and we have something to be waiting for
There is another company, which I'm a constumer too (TheaRender), that does it and it works really great.

Pedro

[radiance]

I am not sure if a blog is needed, it might be overkill right now,
but we will surely keep posting previews of new features in development.

Radiance

[Mateuet]

@Radiance: You could try Twitter.

All the information is in the forums but you really have to know in which particular threads it is allocated in. the information is there but not clearly.
I think that as a customer/Octane-addict (yeah you created a hungry bunch of GPU addicts! ) You guys are doing great but sometimes is a little frustrating to have to spend so much time on the forums to have an idea how the development is going.

Maybe by releasing the information in a more central manner you could facilitate the access to good info to all users. As an example you could tweet posts you think are important to the community, that reveal some extra importance but not yet deserve a sticky.I think that could be very helpful and release some pressure some users have been creating on your team.

Eitherway keep it up!

Kind regards Mateuet

[n1k]

Any info on cuda 3.2 implementation to utilize all cuda cores in gtx460 chip?

Cheers,
n1k

EDIT

I just read in another thread that next release will tackle this issue. Hooray

[radiance]

Yeah, 3.2 is working fine, but, nvidia had made quite a few changes and as a result the multi-gpu support is broken. It works but there is no speedup as things don't run %100 parallel anymore, so i'm currently rewriting the multi-gpu code, which is quite a laborious task unfortunately.

Switching from 3.0 to 3.2 gives an additional 15-20% speedup on fermi architectures (all of them), and for a yet unknown reason a slight performance loss on pre-fermi (GTX200 and lower).

I wish nvidia would'nt change things constantly in between new cuda versions.

Radiance

[steveps3]

well not really everyday but at least a development blog updated every weekend would be highly appriciated. Then at least we know what to expect and we have something to be waiting for
There is another company, which I'm a constumer too (TheaRender), that does it and it works really great.

Pedro

Every weekend before Christmas you mean? Hang on, let me count them..... 1.

[radiance]

We need to be careful as not to give away too much of our developmennt roadmap and new ideas to competitors, so we can only post previews of new features like this one a few weeks or days before a new release.
I know it's not what you would like to hear but it's nevertheless important for us as a company...

I agree that some more transparency regarding newly implemented development would be a good idea, but i'm not sure if the amount of news would be enough to dedicate a special blog/twitter/news page to it, hence doing it with forum posts.

I will think about it some more.

Radiance

[FamilyGuy]

radiance - about opacity and 2D objects:

Can you make such a material, which will automatically add an alpha channel for opacity?

Now if I want to make 2D tree I have to add map to diffuse and to opacity. If I have 20 people and 10 species of trees - it's a lot of work. It would be cool if during the mapping process Octan automatically search for alpha channel in a *.tiff file and add it to the opacity. You can call it 2D material objcest or somehow.

[radiance]

radiance - about opacity and 2D objects:

Can you make such a material, which will automatically add an alpha channel for opacity?

Now if I want to make 2D tree I have to add map to diffuse and to opacity. If I have 20 people and 10 species of trees - it's a lot of work. It would be cool if during the mapping process Octan automatically search for alpha channel in a *.tiff file and add it to the opacity. You can call it 2D material objcest or somehow.

That would'nt work in the concept of nodes, imagine an image texture connected to the diffuse slot, how will it connect to an alpha channel input ?

Radiance

[matej]

Can you make such a material, which will automatically add an alpha channel for opacity?

Hah!, this whole time I was under the impression that the 'alphaimage' load mode serves exactly this purpose - ie. that it loads the image in RGBA mode, so that you can connect it to diffuse and opacity at the same time (the diffuse channel would read RGB data, opacity would read A data).

Note to myself; read the manual