Why Corona render GI diffuse is much weaker than other Renderers ?!

[thacmac]

I think this is both problem of energy loss and  problem of the core GI algorithms.
Actually, All the shader of Corona have problems, but in this post I will start with Diffuse GI lacking. I wonder if the developers will rewrite the core in the future?

On this images below, only 1 direct light ( sun ). other renderers have much more wide GI range.

[davetwo]

you will need to share the scene so others can check parameters I think. It's not the sun thats the issue per se - the look I will depend on if it's reflecting the sky (visible or not) and what the light has to bounce off on the floor. And how there parameters change between the other tests.

[Stefan-L]

Corona has one of the best GI i know from the current render apps.

Vray has more bounces per default(100) thats true, but when rendering Corona interiors one see the default 25 bounce in Corona produces a similar if not better look.
(by the way vray at 25 bounces looks also same as vray with 100 bounces. so it seems the extra bounces have only minimal effect)

from your tests you seem to use a very different tone mapping than for the others, and it seems you yet miss some knowledge on corona setup (see your leave material)

best as Dave said share your scene;)

[thacmac]

Corona has one of the best GI i know from the current render apps.

Vray has more bounces per default(100) thats true, but when rendering Corona interiors one see the default 25 bounce in Corona produces a similar if not better look.
(by the way vray at 25 bounces looks also same as vray with 100 bounces. so it seems the extra bounces have only minimal effect)

from your tests you seem to use a very different tone mapping than for the others, and it seems you yet miss some knowledge on corona setup (see your leave material)

best as Dave said share your scene;)

No, it's not. Corona only look good if it don't put beside V-ray. In this particular case. As you can see, V-ray is the real best renderer for tree scenes. Its deep scattering light through the foliage very evenly and that make the scene look very dreamy and fanciful mood. Corona is not beautiful like that, its contrast not good at all, its GI itself is also lost.



I update some more renderers result of the post. Actually, all the shaders of Corona are weaker than other renderers in both quality, speed and memory using. But on this post, I only start with diffuse, the next post would be volume, SSS, hair, transparency, performance, workflow and manything else.

If you want files, I have .c4d ( + fb and alembic ) files : https://drive.google.com/file/d/1brgvBp2l0843Z3oIqomoZy_890VyJWhy/view?usp=sharing
 But I guess you've only used 3DSMAX and never used c4d. So I will talk about the parameters for you to set up yourself. For the comparison to be unbiased, the conditions for consistency between the software are:
1. Keep the original input texture
2. Always set fraction value of translucency at 55%.
3. Tweak the light so that the color and brightness of the translucency color and direct diffuse match the V-ray render image :
4. do not use any LUTs

The remaining thing is to admire the difference in indirect diffuse between the two or more render engines

[davetwo]

Sorry but theres clearly something wrong with your setup.

I copied the scene into a new file, and the default result from corona is attached (left with default tonemapping including aces ot, right with the aces ot turned off). Take a look - it doesn't replicate whet you showed.

Edit - I see that you set your exposure to -2.8 in the VFB. But replicating that still doesn't create the harsh blacks shown in you example. It actually looks pretty much as per the v-ray version you attached unless you have the aces transform turned fully on. This sems to be a tonemapping issue not GI.

[thacmac]

Sorry but theres clearly something wrong with your setup.

I copied the scene into a new file, and the default result from corona is attached (left with default tonemapping including aces ot, right with the aces ot turned off). Take a look - it doesn't replicate whet you showed.

Edit - I see that you set your exposure to -2.8 in the VFB. But replicating that still doesn't create the harsh blacks shown in you example. It actually looks pretty much as per the v-ray version you attached unless you have the aces transform turned fully on. This sems to be a tonemapping issue not GI.

 So I confirm that this is error issue with ACES tone mapping. After the turn off ACES, even though its scattering is still not as deep as V-ray level, but it looks a lot better than the previous version.

[davetwo]

I dont think you can conflate exposure and tonemapping settings with gi depth in a scene like this TBH. No-one expects the default tonemapping and exposure to match between different render engines. A feew small tweaks in the VFB and you could match pretty easily IMHO.

[thacmac]

I dont think you can conflate exposure and tonemapping settings with gi depth in a scene like this TBH. No-one expects the default tonemapping and exposure to match between different render engines. A feew small tweaks in the VFB and you could match pretty easily IMHO.

I disagree, because
- If there is interference by tweaking in the VFB (except exposure ), or composition, this will no longer be fair.
- Even if you make adjustments in VFB or composition, 2D-based adjustments cannot have the same impact as light transpot algorithms for 3D. For example, when you try to adjust contrast and brightness to brighten dark areas. This affects the entire image, and makes the parts appear washed out and stuck together, ugly and unnatural, while the darkest areas remain dark. Normal areas and highlights are neutralized together. Also, it creates stains and noise from those dark areas. This is far inferior to images produced by pure raytracing algorithms.
Anyway, my purpose in this post is fulfilled, which is to find the error of ACES, thank you for that. :D

[Nejc Kilar]

I'm not really 100% sure what is happening here - from my understanding you are comparing f.e V-Ray with Corona and you left all the VFB related setting at their default? If so then yes, you will get drastically different results because Corona has ACES OT enabled which adds a filmic curve to your image (darker blacks, "softer" highlights etc...) whereas V-Ray doesn't do any of that when you just open the VFB. V-Ray has a Filmic tonemap layer which you can use to get somewhat of a similar effect to what ACES OT does in Corona. For the sake of simplifying things even further, think of it as comparing an image with a LUT applied to an image without it.

And even then that wouldn't be a proper way to compare GI results - this is why comparing render engines is hard.

Ideally you'd have all the settings matching but because each renderer works a little bit differently that is hard to do. Sure, you can set both renderers to do unbiased pathtracing but then you still need to consider V-Ray has its Max Depth parameters set differently than Corona. There's loads of things like that at play potentially.

It could be that I misunderstood you and you have actually taken all of the above into consideration in which case we can, of course, deepen the debate here :)

[Stefan-L]

"No, it's not. Corona only look good if it don't put beside V-ray. "

well, from what and how you put things, it seems you yet lack some render or at least corona knowledge yet, so why not better ask for help if you have an issue?
that is normally better than making such statements right away. (the way you put it it comes more as a troll like comment. please note i am a user myself and not part of chaos)
there is a reason why many of the best archviz studios world wide use corona render as their first choice. it does have outstanding render and gi quality.

i have deep knowledge of both engines, (used vray probably as the first vray user on c4d) overall, and i love vray and corona as well. both are great, also in gi.
as you might know both are developed by the Chaosgroup team, they share their knowledge and technology.

Both use in fact pretty similar GI methods (like BF/LC), with only slight differences.
stronger differences are in the used render settings, the way materials settings work, ray depths etc., and even more in default vfb tone mapping settings.

what you see are the tone mapping differences that vray and corona have in their default settings. it seems also you have a not so well set leave material in the corona tree version. probably too high ior and too much transmission. vray in default is pure neutral, where corona has the today very common new ACIS style active as default , which many like, but can be too dark i agree, if used as full 1.0.

best share both scenes that we might be able to help you