The Rabbit Hole

troy_s

00:14

Hello @Mareck

How are you?

Mareck

nice thanks

2 days of reading most of you're sharing

troy_s

LOL

Sorry to hear that.

Apologies.

Mareck

nice ^^

troy_s

How have you been finding it?

(I'm hoping it is mind ripping, because it was such a huge mental leap for myself.)

Mareck

yes my brain is upside down

the more i learn the more i want to know about that

troy_s

00:17

The remarkable thing is that when I first started learning about the division between the scene referred and display (output / device) referred domains, I was sort of shocked, but the few things I already knew were suddenly illuminated with more clarity.

In truth, most imagers rarely stop and ask what is going on when you twiddle / swizzle an RGB knob. The moment they think on it, the moment questions come up.

(Which is a great entry point.)

@cegaton had a similar experience I think via two questions on SE.

It's very rewarding watching people get flipped upside down and then land with a much better grasp of what is going on and the importance of the details.

@Mareck did you see the slide presentation I did up for one of the local Blender meetups?

Mareck

You probably answered a question i ask myself for some years now.
How can i achieved more profond lighting? ^^

Yes i've seen it ^^

And i've seen to late that you had a blog.

troy_s

Yes. Quit posting on it. Lost interest in bickering over dumb things.

Have you started mucking with wide dynamic range renders?

(There are quite a few folks using them now, the general sentiment is much like Henri said:

@troy_s @BlenderNation Well... I can't go back to sRGB anymore for ligthing! :D I still use sRGB for the material and shading process.

Tweeted by HenriHebeisen on May 12, 2016 at 7:17 PM in reply to troy_s

(I've started to scheme on a version two of the pack given the first was really a technical demonstration for Bassam.)

I'd like to generate a contrasted Look that is suitable for material work.

Mareck

Next part, i need to do a pause of reading about this and start making a real render, then go deeper in the knowledge

troy_s

To start, it's worth just sort of mucking with diffuse based objects like a still life of cube / sphere / etc.

Mareck

Really thank you for your time, you put a lot of it in your answers.

troy_s

00:25

That will demonstrate how much more indirect bouncing you get immediately.

Well I learned from @cegaton, whose answers are ridiculously thorough.

But I also realized that SE is turning into a very good place for folks to research subjects.

(and I got frustrated with a few questions that had answers that were outright incorrect to misleading)

Mareck

Yes i've already realised that, so i quesion i have, is that the things that make most Cycles renders not as realistic as Vray's ones?

troy_s

(From personal experience, 99.5% of the trouble I had learning about colour management and pixels, light, alpha, etc. was all because of really crappy misinformation.)

I'm not terribly familiar with VRay's format or layout. It wouldn't surprise me if they had a custom tone-mapping thing applied.

Have you read the slides?

Mareck

Yes, really nice too

troy_s

Because that covers sort of what Alex Fry covers (both links in the Wide Dynamic Range post)

Basically Alex has some excellent demonstrations in his video.

Mareck

Seen the Alex Fry conf too ^^

troy_s

00:30

But his Cameo interview has the most informative paragraph that sums up everything. This one:

Mareck

Not already read all of his website .

troy_s

The other important feature of ACES is that it provides a robust, well tested, high dynamic range, wide gamut, display transform. In general, most display transforms used “in the wild” are either simple 1D curves, or complex show specific 3D transforms created using the time-honored "twiddle the knobs on the DI system" method. Each have their own issues. Simple 1D transforms, like Nuke's default Rec709 OETF are well understood, and invertible, but begin to fall apart as colour and intensity get high. The seemingly simple issue of clipping once values move above 1.0 leads to all sorts of sub…

Those two paragraphs are so filled with wisdom that an imager understands almost immediately.

That's a huge concept to grasp.

The video has that lovely demonstration with Guardians where the sky has the same value as the wing of the owl.

It's a very insightful read.

What's more exciting is watching imagers grab onto the concepts and start using them. Some of the renders I have been seeing are incredible.

Mareck

Yes so much nice information.

I'm the next one ^^

troy_s

Not only are the renders more rich with indirect bounces and interesting nuances, but the imagers are taking more care and forcing themselves to save the shot, and then grade it.

Mareck

What do you mean by save the shot?

troy_s

00:33

I'd love for Blender to have more colour friendly pipelines though, so this is part of the goal; get more imagers seeing the differences and adopting the techniques so that there is more pressure to get the things fixed.

As you probably have seen, grading should be considered a separate creative act. That is, grading a shot is separate from lighting the shot.

So when you render a shot, physically saving the EXR as a "photo" of the scene, and then taking that EXR into an application or Blender to work on it for grading is a huge step towards a refined and nuanced look development.

There's a lot of secret goodies I've tried to prepare everyone for, but not burden them with just yet.

Mareck

Yes nice way, i've tried to talk about that in a French chat but i didn't have any feedback, it's a pitty

troy_s

For example, the ASC-CDL is an industry standard transform, which means that if you sit down at a Resolve or Baselight station with CDL knobs, you will feel at home and get 1:1 with Blender assuming you have your inputs correctly set up.

But more importantly, for your own work, the ASC-CDL values can be captured into a CC file, and you can then make your own Looks that show up in that Looks drop down.

Ideally, when Cycles is ready for it (example developer.blender.org/T46860) I could prepare a proper ACES pipeline.

Have you tried any simple renders yet?

Mareck

yes a single diffuse green cube ^^, but even with just that I've felt the difference

troy_s

Try putting diffuse checker objects with saturated values on them.

Near each other.

You should see a good deal of amazing indirect lighting and such that wasn't present in the sRGB default transfer curve.

Mareck

and I've tried to compare render that i'm working on with an without -10 +6 range, but they'aren't lit properly so i think it's better to make a complet scene set up for that workflow.

troy_s

00:43

The interesting thing is how much of an impact the wider dynamic ranges impact your work. They increase the light levels so renderers behave differently, they actually capture much more light on the highlight side, so you see different things in the result, and on top of all that, imagers start to actually grade their work and pay attention to the shadows and highlights.

It's very difficult to do an A / B test.

The ranges of lighting and the nuances in materials (strongly encourage you to chase down @jtheninja on Twitter as that peep knows a heck of a lot about how renderers work internally.) results in completely difficult to compare results.

Mareck

I've tested the blue shadow tint and yellow highlight tint, it react completly differntly, and was so cool

I'm i right? The -10 +6 range is for sRGB convertion of wide dynamic range, and the looks with desaturation for the washed out color hightlight?

troy_s

Well the basic view is the -10-+6.5

The various "sharps" sort of give you a ballpark idea as to what you might do with a grade.

(There are some issues that need working out in the looks for the next version - aligning middle greys for example so that the power curves don't futz middle grey points in the images)

the readme has more information.

Mareck

I've alredy read it, so the best is to do it manually, but how do i make the washed out color, is it automatic?

I need to sleep and let my brain rest, one more time thanks and i'll comme back ^^.
See you

troy_s

Ok.

Ping here if you have questions.

Use the

@ name

Mareck

Ok

troy_s

00:57

It will ping whoever you ping on mobile too which helps to get an answer.

The "washed out" is basically the generic mapping.

To make it more contrasty, roll it through the ASC-CDL node with a power and a slight reduction of slope.

Play around with it. Shouldn't take much effort to get a handle on it.

Good luck and hopefully you will find it rewarding.

Transcript for