Rendering In Passes - Useful?

Hello, all:

Some recent conversation with key Blender developers raised an interesting question. I am currently pursuing the idea of permitting rendering in passes as a part of Blender’s render engine. The question is, who would use this, and how?

Please take a moment to answer the poll, but also to post a brief note on how you would make use of this feature and specifically how it provides features which are not currently possible (or are unnecessarily difficult) in Blender itself.

Thanks for your time,

-Bischofftep

I think it would be useful for some of the more advanced users who understand the principals behind this option. For the beginner, it would be a lot of hassle, IMO.

I would like to see it integrated, but I am not sure that it would be used properly, nor used a lot.

That’s just my feeling, but I would still like to see it added as a render option.

BBgDM

Pretty much echoing BgDM… probably would not be recommendable as default behavior. It would definitely make my life easier in a couple cases. Maybe having a “render all passes” option set as default and let advanced users go in and pick n’ choose? The only question, then, is if there are pre-assigned passes or if the user configures that on his or her own. I’d lean toward the latter. It complicates things, but I’d rather have more control with complication than ultra-simple with no control.

I’d use this - mainly to render Zbuffs and Alpha passes for compositing work.

I find think it would be easier to get a render or animation close, and then have the ability to tweak it in the compositing stage. Instead of looking for that perfect render. Besides if you render in passes you have OPTIONS. Your render isn’t an all or nothing thing.

nik

I think it would take rendering a step up. It would be reaaly useful to render in passes and have tha passes set up automatically in the Video Sequence Editor for tweaking.

Alexandre Rangel

Sounds good. I’m all for it.

If I could render out the various passes (spec, norm, dif, alpha, rgb/cmyk channels, more…) I would also hope that real ZBuf would also be possible. Very helpfull when compositing and can sometimes speed up rendering as the final composite is done seperate. It also allows for external manipulation of different channels to enhance or produce different effects. you could do some fun stuff with reflections or specularity with a post process that then gets combined back into the final rendered image/animation. This is something that is available in almost every industry standard app and should be looked at. Granted we’ll be able to get some of this with Yafray integration but most industry apps have the capability internally as well as with external renderers. So I vote yes!

I would most definitely use this feature. Please please please, I want to render in layers!

I agree. Zbuf would be the most important.

I already use Blender’s layers for rendering, but in more complex models I’ve found only having 20 layers takes some shuffling. Multiple passes (and everything already mentioned already) would help. Of course, more options are always welcome and they generally make Blender more powerful… as long as they doesn’t replace ease of use.

As cool as this would be, I don’t think it is absolutely necessary or on the top of the list of things we need.

Thanks for starting up this discussion Bischofftep, because I think it is a vital one if Blender ever hopes to become a serious (production level) 3D suite. I am absolutely for integrating the ability to render in passes, and some of my thoughts on the issue can be seen on this thread on ambient occlusion by Little_Cube. Anyone interested in multi-pass rendering should read this thread for some basic concepts.

Firstly the terms multi-layer and multi-pass rendering are sometimes used interchangeably. I like to think of multi-layer rendering as rendering the scene in parts with different objects in different layers and then compositing them in post. This itself is a very useful technique to have total control over your results, no matter how frowned upon it seems to be by amateur 3D ‘purists’ - the fact is all professionals use compositing and post-processing. Your render is not sacred - live with it.
This can already be achieved to some extent using Blender’s layers, but I think any further tweaks to the code/interface to better facilitate this kind of work flow is always welcome.

Multi-pass rendering involves rendering the various visual attributes of an entire scene in passes, and then compositing them together to get your final result. The resons this is done, include speed, better control over the results, and the reusability of data, amongst others. This is much needed in my opinion, and it is also the technique which works best when it is supported and built into the software. The different passes can be manually created with settings, but that would defeat the purpose of a smooth work flow.

I’m including some links here. I’m sure there are many who don’t totally understand the significance of this technique, and there are others who would like to know more about it. I hope this list of sites provides enough raw material for every level of interest.

Introductions to multi-pass rendering
http://www.renderaid.com/infoDetail.aspx?p=d&id=1028&sid=1003&f=t]@B%3EXI
http://www.thegnomonworkshop.com/tutorials/passes_layers/passes_layers.html

How MAYA handles multi-pass rendering
http://www.technolumiere.com/software/renderpassdocs/RenderPass_Tutorial.html

Multi-pass rendering in XSI
http://www.iro.umontreal.ca/~roys/softimage_xsi/files/docs/html/tutorial/renderin.htm
This is a must see for those who would like to work on this feature. While I have never used XSI, from the description and screen shots, the XSI interface to handle multiple passes seems to be the best I have come across - a simple menu to create passes based on standard models (diffuse, occlusion etc), user defined settings, and even layers (thus integrating multi-layer and multi-pass rendering under a single interface). This is exactly the sort of thing Blender should aim for.

Multi-pass Rendering Surface Setup For Lightwave 7.5(and Maya)
http://www.g-3d.com/Newer_AAC_3d_multi.html
A good tabulated description of the settings and features of the different passes. This should be useful for anyone wanting to work on this feature, and also for anyone wanting to fake the passes with Blender settings until (hopefully) we get native support through the interface.

Multi-pass walkthrough/tutorial
http://cedar.he.net/~jackal/lightwave/multipass/
A step-by-step practical guide to how a multi-pass image is put together.

I hope these links are, in the least, good starting points for understanding this subject. It is true, as some have mentioned, that this feature will not be of interest to the casual dabbler/hobbyist. However, I think Blender has already gone way beyond being just a tool for dabblers in CG; it was always meant to be a production tool. As long as the multi-pass features are invisible to anyone who doesn’t need them, I don’t think it should be an issue. This is a feature that is essential if we want the majority of professional users to not look at Blender as something you play with before you switch to Maya for real work. I myself wish to use Blender forever, and never want to switch to anything, irrespective of the importance or complexity of the project I am working with. For that attitude to be duplicated on a mass scale, multi-pass rendering is a must.

ok multipassing i like the osund of.

however i would love an option for export buffers to file.

e.g. shadow buff, zbuff, edge detect buff, and the lovely crazy coloured buff that you see when rendering. infact i would love an option to see nearly ever single buffer there was for a scene,becasue afterwards these can be very helpfull for edgedetectng in PS and otehr effects.

another feature i’d like to see in the rednering engine is to save the images as XCF in layers, so a single file could be output with all the buffers and any layers e.g. scenes that have been compositied as well as effects plugins that were put over the top.

this would give ultimate control over an images look and feel.

Alltaken

Rendering in passes would be nice, 'nuff said.

I totally agree with you on that one… I use a lot of buffers and junk when I make games… (Zbuffers are depth maps etc)

However I would like to see this option taken further… possibly one could set a buffer as a material and render… :slight_smile:

WOW! Raytracing! Multipass rendering! Now we are getting serious. Whats next, motion capture? You guys are awesome!

Did anyone see the “translucent fish” effect on the making of Finding Nemo? I have been trying to think of a way to duplicate this effect in Blender, and multipass rendering would be the answer.

BTW I am a professional using Blender exclusively for all my 3D work!

So am I. Blender combined with Photoshop, After Effects, and FCP I haven’t found anything I couldn’t do if I wanted to.

I was just wondering what the status of multipass rendering in Blender this was?

Unfortunately, it doesn’t exist.

There is the belief among at least one highly-placed coder that Blender should be sufficient for any task involving its output. If you want to composite something using higher-than 8 bit values, do it in Blender itself!

The idea that taking a 16 bits per channel file, or a series of files rendered in passes, and using it in another program, seems to be unheard of.

I apologize if this comes across as spiteful: that’s no my intent. It’s just that I see great value to 16 bits/channel and multipass rendering, and it has no support from the coders. Alas, I am woefully inadequate to the task of doing it myself, so I just grumble about it. %|

-Bischofftep

yesyesyes. i’d love to see that!!! no, not because it’s funky having that, it’s a really usefull feature for compositing, post production etc. let all them buffers (color, reflection, normals, z, shadows, reflection etc.) be seperately available. this would be again a major leap forward towards more professional blender usage.

marin

Definetly! I totaly agree with Samir.
A few months ago I didn’t know what it was good for until I read Digital Ligthing and Rendering. It seems that makes compositing very easy, but furthermore, you can render your complex scenery once, and than render moving objects in passes. It saves you a lot of render time it seems and you have total control in the final outcome