Simplest tutorial for passes?

(SandraDau) #1

Kind of a weird question but, does anyone know the best and simplest tutorial on how to do passes? For some reason passes confuses me easily. And I found there are tons of tutorials all about 30 minutes each on this…

(JA12) #2

You didn’t mention render engine. Blender render passes can be difficult to work with, Cycles passes make more sense, but it still takes a lot to explain how they work and the benefits. This is the best one, but find painkillers for the headache before watching it

(SandraDau) #3

lol Thank you! And thanks for understanding the pain of learning about passes! I have tried before and I am trying again… I will have my ibuprofen handy. (easily technically confused person here)…

(sundialsvc4) #4

Part of the problem, I think, is that the term itself is somewhat ambiguous. It might be used, in a generic way, to refer to the workflow-notion of building up a completed shot by building-up individually(?) rendered(?) “layers of data.” Or, it might refer to the “layers of data” themselves … such as a “beauty pass” or a “shadow pass.”

The fundamental idea – the thing most-importantly to be grasped – is that “the finished image” is not “the (one …) finished render!” Instead, you are working in a digital darkroom, methodically building up a finished image through a process that will not be evident to the audience when the finished image is complete. At no point might any of the individual components of which the image is constructed closely resemble the final image that is constructed from it. (It will simply look like magic.)

(sozap) #5

Lol, maybe you don’t have to learn it yet if you don’t really feel the need or if it’s hurting you :smiley:

You won’t find something that covers all in one place because it’s a big topic in compositing. To sum up a bit and help you to better find what your looking for :

First, at the basic level, you can do compositing straight on your render (without passes or renderlayers at all) , like adding glow, vignette , color correction. That may be enough for many renders to get a nicer look.
ex : vignette :

At a higher level, If you want more control, you can separate your render in renderlayers, like one layer for the sky, one layers for the set , and one layer for the characters… That will allow you to apply different color corrections to different elements, or maybe blur the sky and the set to make a kind of defocus effect. After that you put each layer on top of each other.
ex : render layers :

You can take it further using passes, some passes are used just for one thing only. Like Vector/Speed pass is only used to add post-process motion blur, (motion /vector blur filter in the node editor) , Normal pass is generally used to do relighting in the compositor , ID pass allow you to make masks of the objects or material to color correct only one part without exporting as a renderlayer.
The good old Z pass is in the end one of the most interesting pass and can be used to do different things . But that may not be easy to catch at first time.
relight :
ID masks :
Z pass :

There is also passes like Diffuse indirect, diffuse direct , glossy direct ect… these ones are there mainly if you want to change the shading of your render. You recombine all these passes with a special recipe that allow you to lower only the glossiness of your render, or change the diffuse color… In fact it may be simpler to produce a render that is already satisfying…

1st exemple :
2nd , more complete but harder :

After all this you can go crazy, and separate everything into render layers and render passes and turn your render into something completely different.

There is in fact 2 extreme ways of doing renders, you can do all the work in the lighting/ shading trying to make thing act as natural and good as possible there, output just one or two layers and do just some little tweak and color correction in the compositor.

Or you can get a bit lazy in the lighting and shading and make a lot of passes and renderlayers to get the final look in the compositor.

Of course you can go control freaks and do both.

At first I was using a lot the second solution, it’s true that you can turn a crappy render into something good with a lot of compositing cheats. But now I try to get good lighting and shading first and do less cheats in compositing.
Many compositing tricks are useful if, for example your working on a project and want to change a lot of things without redoing renders. If not, generally these tricks tend to keep us away from the very basics issues in our scenes. If your render isn’t good, great chances are that the lighting is not well set , or maybe it can be the shading. But crappy renders in general tends to have way better shading than lighting.

If you’re self taught you may come to learn lighting in the end , but in fact it’s the best thing to learn if you want to improve your renders, then comes shading, but with good lighting and very simple shaders you can do a lot of good stuff still.

If you’re really serious about all that and want to learn more, I suggest you these two books :

It’s easy to read and covers all the basics and also it’s could be a good read even for professionals. But it’s more about compositing in general , even if it covers 3D compositing.
This one is about lighting and shading, it covers a bit of compositing too, but if you want to improve your renders it’s easy , simple and well explain.

These two book aren’t about blender or a special software but explain well the theory behind all this , after that you can apply it to any software.

Good work ! from the technical pain comes great skills !

(SandraDau) #6

These comments are great.! And thanks Sozap for all this information. What you say at the end about how important lighting is, I really feel that too, as that is another issue I am having… so many things to work on, it’s always never ending. Anyway I will try to go through with these. Thanks for putting them in order.

(RayVelcoro) #7

ty very much for pointing out this tutorial!

(JA12) #8

SandraDau and RayVelcoro:
note that the tutorial is quite old, and you don’t have to use switches for the layout organization anymore. Reroute nodes have labels, and imo they’re more clear than switches for the layout. Might need to change the default colors for those in user preferences -> themes to have more contrast.

Node Wrangler addon is a must if you work with nodes. It was first made by Bartek and later Greg Zaal helped to extend it. The addon can be used to handle nodes more efficiently, and the myriad of functions include creating and labeling reroute nodes for sockets and between connections.

(RayVelcoro) #9

I saw the Reroute node come across in a other tutorial I saw recently. But thx for pointing it out, otherwise I’m afraid I could not commemorate it. I use Node-Wrangler a fair bit of time already, very helpful add-on indeed.

The tutorial above helped me to get familiar the benefits of OpenEXR and how to approach the different channels of it in the compositor. If u know more recent in-depth on that I Would me very pleased.

(sundialsvc4) #10

One motivation for a multi-pass workflow is that you might not have time to try to render something perfectly, all in one go, and then to wait however-many hours to see if you did. … … and then to do it all over again, because (of course) you didn’t.

It can be very advantageous, sometimes, to break the total problem down into smaller parts which can then be addressed separately. Both lighting and shadow can be major issues which are very difficult to get just right.

In addition, the workflow puts controls into your hands during the compositing stage, which by definition is much less computationally expensive. For instance, if a shadow-pass output tells you where the shadows are, you can exercise compositing controls to determine what they are: how dense the shadows are, how much they fall off, to add tint to the shadow areas to mimic the effect of blue-sky, and so on. (You can also simplify the scene and therefore the calculations … deciding which shadows are “important” and which ones are either distracting or not likely to be missed.) A lot of render-time is burned up by shadows!

This, then, is a more iterative approach which does not demand that every component of the frame – let alone the entire thing – be “perfect.” With planning, you can ensure that each thing which you do “spend the time necessary to render” can be used.

(RayVelcoro) #11

Thanks for this contemplation.

One more question, when i drop my rendered openEXR’s in the compositor building myself the desired node-tree with the old tutorial in mind. Does anyone here perhaps have a template file for composing the openEXR’s righ away with all its passes willing to share?


(sundialsvc4) #12

I don’t think that any such thing exists – or, really, could exist. It depends entirely on your project.

Each render, to MultiLayer OpenEXR files, becomes a source of all of the layers of digital information that were (laboriously …) produced during that render, and that’s it. What you do with all that data is up to you … and you should work this out ahead of time.

For instance, since you know that this is a “frame per file” format, first render one frame from all of the various sources. Now, build a “noodle” network that transforms these into your intended one-frame output. There can be no “template” for this, since every project is different. But, do not spend hours rendering “hundreds of frames” before you work out what you’re going to subsequently do with them!