Author Topic: Comparison between 3DS Max 2023 and Cinema4D 2023  (Read 2756 times)

2023-05-16, 11:15:52

MircoTosti Studio

  • Active Users
  • **
  • Posts: 98
    • View Profile
Good morning lads,

out of curiosity I did a comparison test between the C4D and MAX render engine and I noticed some really substantial differences on which I ask for your opinion (same scene, model, render settings, light, same camera parameters, materials, etc. etc.. scenes perfectly identical in all respects):

- render times are significantly reduced in 3DS Max, from 95 seconds to 72 seconds ... Corona in 3DS Max is 25% (and more) faster than C4D.
- the rendering of shadows and lights is totally different, in 3DS Max the light bounces are much more calibrated and even inside the rooms a completely better light reaches.

Can anyone explain to me why this is all happening?
« Last Edit: 2023-05-17, 10:32:12 by MircoTosti Studio »

2023-05-16, 13:24:45
Reply #1

TomG

  • Administrator
  • Active Users
  • *****
  • Posts: 5434
    • View Profile
I am not sure those differences in the shadows are different GI calculations (because those are shared between both versions) - may be more to do with color calibration type stuff.
Tom Grimes | chaos-corona.com
Product Manager | contact us

2023-05-16, 14:02:26
Reply #2

maru

  • Corona Team
  • Active Users
  • ****
  • Posts: 12714
  • Marcin
    • View Profile
To make it easier to reliably compare images, please either attach them separately (see attachments here), or put them in some kind of slider like this:
1 - https://corona-renderer.com/comparer/qNDZ08
2 -  https://corona-renderer.com/comparer/qA6bKR

One major difference I can see is that one image is darker than the other. Other than that, they are very similar in terms of noise and quality.
The difference might be caused by:
- different tone mapping
- different secondary GI cache calculation
Another thing is that the clouds have a different pattern, and they affect environment lighting, so this can explain the difference in brightness. I suspect the scene setup between the two platforms is not identical.

Also, we would need to know a bit more about your tests, for example: Which exact version of Corona for 3ds Max and for C4D did you use? What is your CPU model? Were you doing some additional tasks in the background while rendering? (for example browsing the internet) Are your CPU temperatures within the acceptable range so that there is no throttling?

And lastly (and maybe the most importantly), rendering for just a bit over a minute does not give us a good understanding of real rendering performance. Using a pass limit (for example 100 passes) and then comparing the render times to reach this number of passes would be more accurate.



I think the "OMG" is not really applicable here. :)


Marcin Miodek | chaos-corona.com
3D Support Team Lead - Corona | contact us

2023-05-16, 18:04:13
Reply #3

MircoTosti Studio

  • Active Users
  • **
  • Posts: 98
    • View Profile
Hi guys, thanks for your fast reply.

Excuse me if I allow myself to make some clarifications because perhaps you have not paid attention to the posted images and what is written inside them:
- one situation is with "shadows active and affected" by clouds..
- one situation is with "shadows unaffected" by clouds..
- the mapping of the tones and the colors are IDENTICAL.. don't say that the color changes the tone or anything else, because I wrote from the beginning that all the settings and the parameters are the SAME, the scene, the materials and the settings are all perfectly identical ..
- in the situation where the shadows are not influenced by the clouds, there is always a notable difference in yield and times..

Both versions (3dsmax and c4d) are with the release 3 apr 2023 (as stated in the upper part of VFB).
Both tests were done by rendering ONLY the scene, without doing anything else otherwise it wouldn't make sense to make a comparison.
The machine is equipped with a Ryzen 7950x processor and 192GB of DDR5 ram, with the operating temperatures of the cpu during the rendering phase are constant at 69°/70° with 100% use for both 3dsmax and c4d.

I'll immediately do the test render with 100 passes and post both the times and the yields. Let's get to the bottom of it now!

Question: If the motor is the same, the version of the crown is the same, how can it calculate the secondary GI differently to such an extent that it changes the rendering of the scene?

2023-05-16, 18:10:46
Reply #4

maru

  • Corona Team
  • Active Users
  • ****
  • Posts: 12714
  • Marcin
    • View Profile
Clouds do affect scene lighting. They are like an HDRI. Even if you disable their shadows, they still affect lighting, reflections, refractions, etc.
As you can see here, the clouds are different in your 3ds Max and C4D images:
https://forum.corona-renderer.com/index.php?action=dlattach;topic=40037.0;attach=183234;image
https://forum.corona-renderer.com/index.php?action=dlattach;topic=40037.0;attach=183236;image

If the clouds are different, then the lighting is different = you will not get identical results.

As to this question:
Quote
Question: If the motor is the same, the version of the crown is the same, how can it calculate the secondary GI differently to such an extent that it changes the rendering of the scene?
The UHD Cache is biased and by default it is recalculated from scratch for each frame. You can get different UHD Cache pattern (different brightness, some hardly visible splotches) even if you re-render exactly the same frame multiple times. If you want to get rid of this factor, you can switch to Path Tracing as both the primary and the secondary GI solver, but then you will not be using the default settings any more.

Some minor differences between the rendered frames are expected. I am not saying that it is exactly what we are observing here, but it could be it.
Marcin Miodek | chaos-corona.com
3D Support Team Lead - Corona | contact us

2023-05-16, 22:21:12
Reply #5

MircoTosti Studio

  • Active Users
  • **
  • Posts: 98
    • View Profile
Not being able to understand where to load the images to make the visual as your "compare" I am attaching both the clean renders and the screens of the relative VFBs.

As requested I rendered the scene with 100 passes:
- the sky has no clouds and all the same parameters.
- both renders are path traced as requested.
- same material (all settings).
- same cameras (all settings including color mapping).
- everything in the scene is identical.

Comparison:
- tests show that shadow quality is markedly different.
- the tests show that the times are not at all identical, in fact we have 3 minutes less in 3dsmax than in c4d.
- the rays and all the other calculations are almost similar as seen also in the screenshots.

No other programs were used during the render tests. I hope it will be useful for you..

 Thanks guys
« Last Edit: 2023-05-17, 00:24:08 by MircoTosti Studio »

2023-05-17, 09:17:25
Reply #6

maru

  • Corona Team
  • Active Users
  • ****
  • Posts: 12714
  • Marcin
    • View Profile
Thank you for testing this.
Sorry for the confusion about the image comparisons. Only Corona team has access to our slider tool. You can use some online tools (like this https://web-toolbox.dev/en/tools/image-compare-slider) or I can take your images and upload them to our tool if you wish.

What exact tone mapping settings are you using for your images? Could you share a screenshot?

Other than that, here are some ideas what may be causing the differences, but I will get in touch with our developers to learn more and will share the results with you:
- There are differences between color management in 3ds Max and Cinema 4D (3ds Max is not color managed). This can be affecting the final appearance of the saved image.
- Even if both platforms support the same features, the way those features are implemented may be different. This can cause slight differences in rendering performance.
- You are comparing two daily builds. They might not be in synch with each other in terms of features and fixes (e.g. some fix may be already included in C4D but not in 3ds Max, or the other way around).
Marcin Miodek | chaos-corona.com
3D Support Team Lead - Corona | contact us

2023-05-17, 09:29:07
Reply #7

romullus

  • Global Moderator
  • Active Users
  • ****
  • Posts: 8780
  • Let's move this topic, shall we?
    • View Profile
    • My Models
Could it be that CforMAX and CforC4D has different default settings for something like cache precision, or something like that. I remember back in the alpha days Corona used to store these settings in a text file, which user could change at his own discretion and use as defaults.
I'm not Corona Team member. Everything i say, is my personal opinion only.
My Models | My Videos | My Pictures

2023-05-17, 09:35:23
Reply #8

maru

  • Corona Team
  • Active Users
  • ****
  • Posts: 12714
  • Marcin
    • View Profile
@Mirco - could you share both scenes with us at https://support.chaos.com/hc/en-us/requests/new ?
Marcin Miodek | chaos-corona.com
3D Support Team Lead - Corona | contact us

2023-05-17, 09:52:27
Reply #9

Stefan-L

  • Active Users
  • **
  • Posts: 503
    • View Profile
One thing to mention from color: have you set the corona preferences to NOT use the c4d color management? if you have the default on corona c4d uses the c4d management whi9ch is slightly different, try turning that off to get more similar results as in max.

overall c4d has a different scene and material handling, the 3 min arent 25% slower as you said in first post, but yes a tiny bit, which might come due host app differences i think (render scene export, and the way materials and shader work, are quite different in c4d compared to max and probably a tiny bit slower).

in max you cannot do anything in the app while work, in c4d we can fully work even on the same scene, while rendering. this alone costs a tiny bit of speed due different cpu  threading. i rather prefer the c4d way in that case, even if that has a slight render time cost.

cheers
Stefan
« Last Edit: 2023-05-17, 09:58:46 by Stefan-L »

2023-05-17, 10:31:31
Reply #10

MircoTosti Studio

  • Active Users
  • **
  • Posts: 98
    • View Profile
Thanks guys all for the replies!!!

Since I use both software, I'm pleased to investigate this "aspect" both as a personal doubt and as a possible solution to be adopted at a working level when it occurs.
As far as the question of color is concerned, I believe that this does not affect the calculation of the shadows of the scene: if I compare the results of the two renders, the color of the sky seems very very similar to me, instead what differs are the shadows which are much more contrasted and strong in c4d.. i guess this is due to the calculation of the light which affects in making the process of 3dsmax speed up compared to c4d.
Anyway, as soon as I have a moment, I'll send you both scenes but you can easily do a test using any scene outdoors and setting exactly the same settings in both software as I did, nothing in particular.
As for the time, to be picky, going from 13 minutes to 16 minutes results in +24% ehehehe :) :)

2023-05-17, 10:46:33
Reply #11

Stefan-L

  • Active Users
  • **
  • Posts: 503
    • View Profile
well i am a user too, i am not part of corona or chaos:)

the tone mapping i think can in  deed affect the shadow parts of an image, it is worth testing both in the same setting(so c4d tonemappng off), best exported both as 323bit exr without any tone mapping applied.

for speed: as said i think it is mainly the differences in the host apps - but i rather have c4d while i continue to work on the scene., c4d has lower thread priority which is a good thing overall(more balanced).
if there really is a code issue of course that affects the differences i hope the corona team can improve this.
« Last Edit: 2023-05-17, 10:50:15 by Stefan-L »

2023-05-17, 10:54:19
Reply #12

maru

  • Corona Team
  • Active Users
  • ****
  • Posts: 12714
  • Marcin
    • View Profile
Anyway, as soon as I have a moment, I'll send you both scenes but you can easily do a test using any scene outdoors and setting exactly the same settings in both software as I did, nothing in particular.

Sure, but we are interested in checking your scenes specifically, as the issue appears in them. :)


the tone mapping i think can in  deed affect the shadow parts of an image

I may sound like a broken record (I keep writing this on the forum once in a while) but this is a very important piece of information. Things like the difference between gamma 2.2 and sRGB gamma result in differences mostly visible in the shadows.

But on the other hand... in Mirco's comparison, the same image is darker in the center (shaded areas in the center of the image) and at the same time brighter in other areas (e.g. the bottom part). So this is really confusing. :)
Marcin Miodek | chaos-corona.com
3D Support Team Lead - Corona | contact us

2023-05-17, 12:25:49
Reply #13

MircoTosti Studio

  • Active Users
  • **
  • Posts: 98
    • View Profile
Guys I made the render with tonemapping off (file 1.jpg).
The scene approaches that of 3dsmax, but at the same time it always remains distant from the point of view of the shadows and their strength.
I am attaching this test render made following Stefan's instructions

-update-
Maru sorry but I have only now seen your answer, in part I have already provided these new elements.
I know you repeat yourself like a broken record, I always pay close attention to advice and by often reading the forum I always treasure everything you provide. At the same time, if I notice problems, I can't help but point them out and I always try to provide as much information and tests as possible, both to give a more accurate picture of what I'm experiencing and to clear any doubts.
« Last Edit: 2023-05-17, 12:47:02 by MircoTosti Studio »

2023-05-29, 20:29:19
Reply #14

smckenzie

  • Active Users
  • **
  • Posts: 24
    • View Profile
I had the same confusion but between VRay and Arnold. In the end it was the default tone mapping that is applied by Arnold in C4D  (ACES SDR Video), it applied this even when C4D is in the regular SRB mode for color management.

So in C4D with Corona, I turn off the ACES transform in the VFB and it's all good.

2023-05-30, 15:44:00
Reply #15

maru

  • Corona Team
  • Active Users
  • ****
  • Posts: 12714
  • Marcin
    • View Profile
@Mirco - We are still interested in looking into this. Have you shared your scenes with us somewhere? If not, could you please submit a new support ticket here and attach your scenes? (or some other ones where the issue is reproducible)
https://support.chaos.com/hc/en-us/requests/new

Thanks in advance!
Marcin Miodek | chaos-corona.com
3D Support Team Lead - Corona | contact us

2023-05-31, 05:25:50
Reply #16

wadeheaton

  • Users
  • *
  • Posts: 2
    • View Profile
regarding the speed difference.I would say corona is much faster in MAX compare to the C4D version.