Poll

3 features you want the most:

PBR Style material (Disney, Unreal Engine, etc..)
33 (13.4%)
Render-time booleans (cut/slice objects)
13 (5.3%)
Dedicated CarPaint Shader
1 (0.4%)
GPU/Hybrid rendering
34 (13.8%)
Speed improvements
27 (11%)
Cryptomatte
10 (4.1%)
Geopattern
16 (6.5%)
Sketch/Toon shader
7 (2.8%)
Reworking tone mapping (DSLR-style tonemapping)
45 (18.3%)
Interactive rendering in 3ds max viewport (with gizmos, object selection, manipulation, ...)
5 (2%)
Adding own materials to Corona Material Library
4 (1.6%)
Lightmix extended to materials, textures, ...
10 (4.1%)
Curvature map
9 (3.7%)
VR Scans compatibility
5 (2%)
Dedicated fabrics shader
14 (5.7%)
Decals workflow
13 (5.3%)

Total Members Voted: 93

Author Topic: The most wanted feature?  (Read 289504 times)

2020-09-26, 18:37:08
Reply #870

lupaz

  • Active Users
  • **
  • Posts: 543
    • View Profile
Guys, when is VRayAerialPerspective atmospheric effect going to be brought to Corona? Any plans?

2020-09-27, 09:37:28
Reply #871

Yuriy Bochkaryov

  • Active Users
  • **
  • Posts: 51
    • View Profile
    • Home Page
It baffles me that we keep voting up speed improvements every single time instead of new features that make our jobs easier or more efficient. If you have a decent workstation its impressively fast already!

Ummm ..speak for yourself :)  Aannd... what is this "impressively fast" thing that you speak of ? I don't know what it means but it has a very nice ring to it.

there is no way to greatly speed up the rendering by optimizing the render engine, over the years it has already been optimized almost to the maximum
render engine is a set of formulas and algorithms by which the calculation is performed on the CPU, you cannot change the formula to make it work faster, otherwise it will not be correct.
The only way to increase the speed is to buy good new CPUs, AMD now offers good solutions for its money.
To put it simply, you cannot greatly speed up rendering through programming, only by increasing the power of the CPU.
There is also an option to make a hybrid with a GPU, and it will definitely be in the future, since this solution greatly speeds up rendering and this is the future, but it is very difficult and it is a very long process
so if you want more speed here and now - buy new powerful CPUs
so it makes no sense to spend a lot of development time to get a maximum of 5-10% of the real increase in speed, it is better to do something more useful during this time - a new PBR. Parallax, new tone mapping, etc.

2020-09-27, 09:40:37
Reply #872

Yuriy Bochkaryov

  • Active Users
  • **
  • Posts: 51
    • View Profile
    • Home Page
Why can't I see parallax on the voting list?
don't you think it matters?
I saw the result in FStorm it looks great, the detail through parallax is just excellent

2020-09-28, 23:04:07
Reply #873

Jpjapers

  • Active Users
  • **
  • Posts: 1400
    • View Profile

there is no way to greatly speed up the rendering by optimizing the render engine, over the years it has already been optimized almost to the maximum
render engine is a set of formulas and algorithms by which the calculation is performed on the CPU, you cannot change the formula to make it work faster, otherwise it will not be correct.
The only way to increase the speed is to buy good new CPUs, AMD now offers good solutions for its money.
To put it simply, you cannot greatly speed up rendering through programming, only by increasing the power of the CPU.
There is also an option to make a hybrid with a GPU, and it will definitely be in the future, since this solution greatly speeds up rendering and this is the future, but it is very difficult and it is a very long process
so if you want more speed here and now - buy new powerful CPUs
so it makes no sense to spend a lot of development time to get a maximum of 5-10% of the real increase in speed, it is better to do something more useful during this time - a new PBR. Parallax, new tone mapping, etc.

Exactly, the amount of time that passes between major versions of corona isnt enough for there to be much significant increase in hardware capability or code efficiency except by throwing more processor cores at it *Coughs in AMD*.

Speed improvements are great but in the long run they arent sustainable because they become counter productive to spend something like 30% of your dev time trying to gain what could be a 2% speed increase in one area of the software. It could be used to develop a new feature that makes another area of the production process easier and faster and more flexible so maybe the speed isnt as necessary. Like lightmix layers for textures and materials. A small speed increase wouldn't matter there because if you only have to render once to be able to get multiple versions of a render, you're saving time by not having to render over and over again. Theres your speed increase.

We should stop focussing on the speed of a single render and start looking at the bigger picture of the volume and flexibility of work we can produce with one click of the render button.
« Last Edit: 2020-09-28, 23:42:16 by Jpjapers »

2020-09-28, 23:34:24
Reply #874

sebastian___

  • Active Users
  • **
  • Posts: 173
    • View Profile
In theory that sounds ok, but maybe the devs know something we don't know, maybe they can indeed optimize and get more speed still from their code. The "speed improvements" was put on the poll by the devs right ?

And maybe the speed improvements can be not some simple brute numbers or percentage increase by doing some super magic tricks and just increasing all renders by 30% . That doesn't sound plausible indeed.

But instead they can increase the speed by making the render engine even more "smart" . For example the engine should somehow know if most of the image is just blank, or a white wall without much happening in terms of shading, details or reflection and that part should render super fast. Other render engines seem to have an edge in this regard and render much faster if not much is happening on the screen, like maybe for compositing reasons most of the screen is empty.
 Or maybe you do have a minimalist scene with lots of huge walls without details. Corona seems to render these pretty slow.

I think more and more people want to use Corona for animation too, not just for stills. And I think this would help that.

(and I voted for  lightmix layers for textures and materials)

2020-09-28, 23:48:29
Reply #875

Jpjapers

  • Active Users
  • **
  • Posts: 1400
    • View Profile
In theory that sounds ok, but maybe the devs know something we don't know, maybe they can indeed optimize and get more speed still from their code. The "speed improvements" was put on the poll by the devs right ?

And maybe the speed improvements can be not some simple brute numbers or percentage increase by doing some super magic tricks and just increasing all renders by 30% . That doesn't sound plausible indeed.

But instead they can increase the speed by making the render engine even more "smart" . For example the engine should somehow know if most of the image is just blank, or a white wall without much happening in terms of shading, details or reflection and that part should render super fast. Other render engines seem to have an edge in this regard and render much faster if not much is happening on the screen, like maybe for compositing reasons most of the screen is empty.
 Or maybe you do have a minimalist scene with lots of huge walls without details. Corona seems to render these pretty slow.

I think more and more people want to use Corona for animation too, not just for stills. And I think this would help that.

(and I voted for  lightmix layers for textures and materials)

Oh absolutely im not criticising the devs themselves and they obviously know their codebase more than anyone especially what needs work and they probably already know what needs optimising. But im just guessing that a specific speed improvement wouldnt even touch the speed gains you'd get from not having to render multiple times. Its the same for GPU rendering and things like geopattern which would give a speed increase in scene modelling and prep time rather than render time. The speed gains youd get from being able to scale your render speed simply by adding another GPU would probably be vastly more than what you could do with software alone. I just think alot of users get stuck on the speed that they can render one image locally, which in a well made scene is already just as fast as any other CPU renderer if not faster and alot simpler. Voting "speed increase" to me is too vague and pretty much all of the features listed would give some form of speed increase in the pipeline directly or indirectly.

P.S. What youre talking about there is the adaptive engine and corona indeed already has that and will focus less samples on areas such as big blank walls. This is why skies will render clean very very quickly.
« Last Edit: 2020-09-29, 00:13:01 by Jpjapers »

2020-09-29, 01:32:42
Reply #876

sebastian___

  • Active Users
  • **
  • Posts: 173
    • View Profile
About a year ago I did an animation test with Corona. Not sure what version it was but probably already had the adaptive engine.

It had a very pleasant... I don't know what, lighting ? colors ? It was pleasant everything. Even though it was a simple plane, a little bit of corona sky behind and a small animated object. Fixed camera. And I was a little troubled to see that the speed of rendering of just the empty plane ground and the sky was not much different if I added the object in or not. I felt like the speed of the empty scene should have been much much faster.
  It felt stupid for me to wait for the renderer to render 1000 frames which were all identical between them save for the little bit of moving render noise or grain and 25% of the image occupied by the animated object.
  So I managed to animate a small render region just for the animated object and of course that went way faster.

Then I saw here on forum other people with the same problem, I even tried replying that they can also use this method - animating a small render region to follow the moving object, but they were not interested, they were just upset that Corona takes that much time to render a static simple scene or a large plane.

I also tested a few other renderers, I don't remember which ones, maybe even IRay ? Not sure, but some renderers, like one would expect, had a much greater speed increase if you had the animated object in the scene, or not. Also a great speed increase for the portions of the scene without details, if it used bucket rendering you could see the increase in speed.

2020-10-13, 19:07:32
Reply #877

Archdeldesign

  • Users
  • *
  • Posts: 2
  • Architect/Interior visualisation
    • View Profile
    • Behance profile
V-Ray Hair&Fur support, please please please!!!

2020-10-20, 17:51:17
Reply #878

bluebox

  • Active Users
  • **
  • Posts: 241
    • View Profile
Concerning tone mapping that is to be overhauled.

Not sure if you guys came across Raya for Photoshop that works on "luminosity masks" -

I'm not a coder, also don't even try to grasp the science behind tonemappers etc. but maybe this could be implemented internally in some way as the results the guy achieves using this are quite nice and the better the more different exposures of the picture he is able to use.

In Corona VFB we can get unlimited number of exposures. Mixing them in a smart way could help bring back all the detail in the shadows area and also exposed areas of the image without flattening it with Reinhard etc.

@Maru - have you guys thought about this kind of approach ?