Render passes/AOVs

[stanland]

Hi,

I am a new Lux user and i'm doing my first personal project test before i start using Lux with commercial use and i have to say that you developers have made amazing render engine, congrats! I have a few questions about render passes/AOVs in LuxCoreRender!

1) How do you combine all render passes in Lux? In cycles you do it like in this manual http://builder.openhmd.net/blender-hmd- ... asses.html

2) Why Bidir engine only supports a few AOVs and will it support in the future all render passes?

[Dade]

2) Why Bidir engine only supports a few AOVs and will it support in the future all render passes?

BiDir uses both eye paths and light paths at the same time. So, while Path uses a single image buffer, BiDir uses 2 with different buffers and normalization strategies (one for eye path samples and one for light path samples).

Given the BiDir Vs. Path core differences, there are few reasons to explain why we support very few AOVs in BiDir:

1) Some AOV doesn't make very much sense in BiDir context given it works in both directions.

2) Even if a AOV makes sense in BiDir context, it would require 2 buffers instead of 1 (i.e. requires a more RAM). And they are more complex to handle.

3) BiDir is (often) used for photo-realism where AOVs are (often) not used. You set up a scene like in a real studio and fire-and-forget the rendering without any particular AOV composition at the end.

4) I'm lazy

It doesn't mean BiDir can not have support for far more AOVs but it is a task with a quite low priority at the moment.

[Sharlybg]

3) BiDir is (often) used for photo-realism where AOVs are (often) not used. You set up a scene like in a real studio and fire-and-forget the rendering without any particular AOV composition at the end.

4) I'm lazy

4) I would say you're extremelly far from that (you probably have some high mind standars )

[stanland]

2) Why Bidir engine only supports a few AOVs and will it support in the future all render passes?

BiDir uses both eye paths and light paths at the same time. So, while Path uses a single image buffer, BiDir uses 2 with different buffers and normalization strategies (one for eye path samples and one for light path samples).

Given the BiDir Vs. Path core differences, there are few reasons to explain why we support very few AOVs in BiDir:

1) Some AOV doesn't make very much sense in BiDir context given it works in both directions.

2) Even if a AOV makes sense in BiDir context, it would require 2 buffers instead of 1 (i.e. requires a more RAM). And they are more complex to handle.

3) BiDir is (often) used for photo-realism where AOVs are (often) not used. You set up a scene like in a real studio and fire-and-forget the rendering without any particular AOV composition at the end.

4) I'm lazy

It doesn't mean BiDir can not have support for far more AOVs but it is a task with a quite low priority at the moment.

Hi, Thank you for detailed and informative answer!

I think that render passes are important to have in LuxCoreRender, no big studio would render animation without render passes and i think it means no animation studio would use LuxCoreRender for bigger projects because of that. For only images i can agree that you don't need always passes.

[Sharlybg]

Hi, Thank you for detailed and informative answer!

I think that render passes are important to have in LuxCoreRender, no big studio would render animation without render passes and i think it means no animation studio would use LuxCoreRender for bigger projects because of that. For only images i can agree that you don't need always passes.

I really do prefer this approach ( being the most accurate possible / and as fast as possible by using caching tech or other stuff to accelerate processing). I'm not a great fan of hyper cheating and heavy post-prod / compositing. But that is it everyone have different mean and needs.

interresting video about the subject here : voice are french but subtitles are english.

https://www.youtube.com/watch?v=1TUP_dGg944

[stanland]

I really do prefer this approach ( being the most accurate possible / and as fast as possible by using caching tech or other stuff to accelerate processing). I'm not a great fan of hyper cheating and heavy post-prod / compositing. But that is it everyone have different mean and needs.

interresting video about the subject here : voice are french but subtitles are english.

https://www.youtube.com/watch?v=1TUP_dGg944

I have to agree with you and i have to say that i'm also preffering this new approach. Thanks for the link, it was very interesting.
Is there any information about how widely used LuxCoreRender is at the moment in CGI industry or how many users we have at the moment?

[cgvirus]

Sorry for the bump but for future reference for google:
Path Render:
(Indirect Diffuse + Direct Diffuse) + (Direct Gloss + Indirect Gloss) + Indirect Specular + Emission = Beauty

[Sharlybg]

Sorry for the bump but for future reference for google:
Path Render:
(Indirect Diffuse + Direct Diffuse) + (Direct Gloss + Indirect Gloss) + Indirect Specular + Emission = Beauty
https://imgur.com/a/L88gTy7

next time will be nice that you embeded your image directly on the forum so we can see it easily

just click on attachements button down when writting your post.

[cgvirus]

Ow. I was searching for it. Thanks!