Render Passes Discussion

[v-cube]

Lighting layers, this was already on our road map (not necessarily for 2.1x) but it seems that there's more interest in deep data compositing so we might do deep data first in favour of this one.

I just want to express my hope that lighting layers (which would ideally support user defined groups of lights?) will be supported - for me who does architectural visualisations, this would boost my workflow into another dimension... just imagine you could balance and fine tune your lighting in post!!

please !

Andreas

[geo_n]

99% of the time, I am saving out 32bit PNGs and popping into PS & AE comps. I (and others) want to be able to mask stuff at will and save out PNGs just as they appear in Octane with alpha enabled in RT. It is as simple as that. All of this other talk is just confusing the matter - and I am afraid that this request will just get lost in a sea of mumbo jumbo, again, like it did in beta, and we will still be without ability to mask things.......

When I was saying constant black, I was not hoping to save out a separate b/w alpha image or anything like that. I want a hole where my mask object/surface is in 32bit PNG. I really hope this is cleared up now.

99% of the time we're using exr with ae, nuke, fusion.
The mumbo jumbo you're callling, which is proper implementation of material and object id/tags can be used to create a hole whereever you want even in 32 bit png.
Rendering out a couple of separate b/w alpha with constant black in lightwave is old, legacy technique. Check the cinema4d video or any video dealing with id/tags, multimattes, etc


Its a waste of effort to implement legacy stuff in lightwave because of its lack of proper id/tags.
We have a chance here to modernize lightwave through octane renderer. If they implement proper id/tags in standalone it could trickle down to lightwave.

I wish Greenlaw from newtek forums was here to talk some sense.

[geo_n]

I just want to express my hope that lighting layers (which would ideally support user defined groups of lights?) will be supported

This is big for us as well so emphasis on user defined lighting groups.
That and motionvectors and coverage and I'm set to present octane to my TD to think about using octane in production. Hope these get implemented asap.


From the other thread
Pls add compression, rle, piz, zip, etc. I think we have uncompressed only right now.
The size of exr is 3 times bigger than what it should be.

[FrankPooleFloating]

Marcus/Roeland/Thomas, if anything could be done that satisfies both the EXR and PNG folks in regard to what is discussed above, that would rock. If it ends up that I have no choice but work with EXRs and do all masking through objectID/materialID etc via Passes, so be it. This just seems like a much slower way to work - and I strive to streamline anywhere I can. If I had the ability to mask stuff quickly (things I don't want in render - and want a hole instead) and a save a 32bit PNG that was right on the money and I could drop it into a PS or AE comp, with nothing else needed, I would be like a friggin' kid in a candy store... from what I have gathered since we started talking about matte/masking way, way back, I am not alone. It's all good. I have no doubt something is coming that will make the dream come true... I'm on the edge of my seat to find out what that is.

[stratified]

Something is coming but we are still figuring out what...

Right know there is the alpha matte camp (Lightwave + ..) and the rgb matte camp (Cinema, Max + AE) but we're trying to figure out if we can do something better. If not we can still do one of both.

cheers,
Thomas

[FrankPooleFloating]

Thank you very much Thomas. You da man. I'm sure something better will be great, if you can pull it off.

I'm personally hoping for a scenario where there would be a [x] Zero Alpha (or [x] Mask.. whatever) check box in Diffuse Material, in addition to the [x] Matte - and they would be mutually exclusive (turn the other off - gray other out)... AND have the best materialID masking options that you gents can get into Passes and/or InfoChannels... This would be like winning the freakin' masking lottery.

I might very well need all this masking way more than most of you - because of two very specific things (and completely different, as it turns out) I do in 3d - that requires a shitload of masking - and I really need to be as quick and nimble as possible... But I know for certain that many others have been wanting this for almost two years as well. Hopefully we will all be jumping for joy soon.

[geo_n]

Something is coming but we are still figuring out what...

Right know there is the alpha matte camp (Lightwave + ..) and the rgb matte camp (Cinema, Max + AE) but we're trying to figure out if we can do something better. If not we can still do one of both.

cheers,
Thomas

Thanks. If ever you do only one of both, please go with the proferssional industry standard approach.
Its very rare to create a special matt scene and render out another sequence just to get an alpha matt of an object, when you can get it for free in your first render that is embedded in exrs.
Its a waste of render time just like it was when you had to plug the info channel just to get a particulare pass and re render the whole sequence again. Such a waste and pros don't do it this way since the 90's maybe. Only people who don't understand modern compositing use re rendered alpha matts nowadays.
Outdated technique of masking and render passes in lw.


Although I hope the octane renderpasses could be improved, too, especially rendering out the zdepth takes its own time to render itw own pass which I suspect is a gpu limitation that it can't be rendered at the same time with the beauty pass seamlessly.

[FrankPooleFloating]

Never. Not once, did I say I wanted an alpha matte.

[BorisGoreta]

We just want to force object alpha to 0 in the final render alpha channel. What can be simpler than this ? Many of us are one man band and we need things to be as simple as possible with as few render passes as we can get by ( read just 1 pass ). Otherwise it becomes unmamagable for one man struggling with 6 projects at the same time. That's it. This doesn't interfere with any of the other user suggestions so I don't understand a negative stand towards it. It is just a simple request which would make life a lot easier for many users who understand what this is for. I don't want X amount of render passes embedded in a voodoo exr format for which I need a compositing software which can read it and I need to learn how to use it. I leave this to the pros. For us mortals, 0 forced alpha is a pinnacle of compositing.

EDIT: I don't use compositing because modern renderers like Octane render compositng software obsolete. RAW Octane output looks so good that fiddling in compositing software only makes it worse, this is true for Photoshop work too, I rearly do anything to still renders, before I had 20+ layers and adjustments per still render. Now I don't have any, or maybe one, but I correct that in Octane and have none again. So my point is, please, no compositing, too much hustle. If I really have to composite over live footage I want Octane to give me sequence which I just put on top of original footage and send it to client. This is a perfect workflow. And I get plenty time left for real life.

[FrankPooleFloating]

Thank you so effing much Boris!!.. My bro!

geo_n, unless you work in HTML5 canvas (for instance), with hundreds of 32bit PNGs, you just wouldn't understand... As I said way back: we are coming from different realms with sometimes very different needs. We do not all work in Nuke and AE all day...