PDA

View Full Version : LW Renderer vs Modo Renderer



Over
07-09-2017, 09:36 PM
50one and hrgiger, I created this thread so we don't derail the LW Next one.

This thread is about the comparison in render time between Modo and LW.

I decided to use a heavier object, pretty much everything is default render settings except 50% radiosity added, camera and light in the same position in both, resolution 1080p. I assume that you're both better experts in this than me, so you can advice any other settings that can be changed to make the comparison more oranges to oranges.

Pictures below.

LW 2015.3

137327

Modo 11.0v3

137328

Results:

LW: 6.19 seconds

Modo: 26.82 seconds

Specs:

CPU: Core i7 [email protected]
RAM: 16GB @ 3000Mhz
GPU: GTX 1080Ti

Chris S. (Fez)
07-09-2017, 09:41 PM
You sure radiosity is switched on in LW?

Over
07-09-2017, 09:42 PM
Yes, I know, itīs dark, but itīs 50% for sure.

Chris S. (Fez)
07-09-2017, 10:08 PM
Keep the black background but add a white environment for radiosity so we can compare grain and such.

hrgiger
07-10-2017, 01:10 AM
Again, default doesn't mean anything because they're different for both applications and each has unique controls for render settings.

Unless you're posting a scene file with render times and a given quality of render, you won't get a honest comparison of render times because you can't match settings between LW and Modo.

gerry_g
07-10-2017, 04:09 AM
yeah I'm calling BS on this too, have seen times go either way by a large margin and as stated complexities and differences between the two make any meaningful comparison problematic to say the least, the light is more ambient/global in the Modo one and it looks like you haven't set up your glass material correctly

jasonwestmas
07-10-2017, 08:05 AM
Bucket renderers (modo) tend to save time (and memory) with much more complex scenes fyi. A bucket renderer is going to win the the bang per buck competition for larger scenes. Not including setup time which is an important factor as well.

Over
07-10-2017, 09:15 AM
yeah I'm calling BS on this too, have seen times go either way by a large margin and as stated complexities and differences between the two make any meaningful comparison problematic to say the least, the light is more ambient/global in the Modo one and it looks like you haven't set up your glass material correctly

Yeah, sure, it must be that... OH WAIT.

All I did was take an object made in LW, setup the same layout for both, add 50% radiosity and press F9. Yes I know that the glass material is not picked correctly by Modo (gotta guess why), yes I know the light is more ambient/global in Modo, but is all I have for you. Setup as said above, F9 and times. It seems youīre kind of used to call BS everywhere because it canīt be a lack of knowledge of my part, it absolutely has to be BS.


Again, default doesn't mean anything because they're different for both applications and each has unique controls for render settings.

Unless you're posting a scene file with render times and a given quality of render, you won't get a honest comparison of render times because you can't match settings between LW and Modo.

Iīm sorry, but why does people buy Octane to Render? Itīs not because it is fast and also the result looks very good? What does architectural diferencies has to do when comparing different render apps? As far as I know your desicion is based on time and looks, not if this is programmed like that or like this. I think that how fast you can setup your render scene also counts.

I will post the render output with times if thatīs what you want.



Keep the black background but add a white environment for radiosity so we can compare grain and such.

Ok, will do.

hrgiger
07-10-2017, 09:28 AM
I don't know how else to explain this.

Modo handles Antialiasing different than LW. It handles shading and samples different than LW. Modo is a bucket renderer and LW is not and handles memory differently. Both have different importance sampling. Saying you set things at default render settings, added radiosity, doesn't make them the 'same scene'.

jwiede
07-10-2017, 09:37 AM
50one and hrgiger, I created this thread so we don't derail the LW Next one.

This thread is about the comparison in render time between Modo and LW.

I decided to use a heavier object, pretty much everything is default render settings except 50% radiosity added, camera and light in the same position in both, resolution 1080p. I assume that you're both better experts in this than me, so you can advice any other settings that can be changed to make the comparison more oranges to oranges.i

Without the scenes, there's no way to judge how similar or dissimilar the render settings are between the two apps. Please provide the actual scenes/objs/textures used.

Settings such as how radiosity is "added" in the two apps can yield quite different results (f.e. MODO's renderer behaves differently depending whether using progressive rendering or not). Obtaining comparable sampling, AA, etc. in both requires different settings in LW vs MODO. Your saying "defaults" isn't enough info by itself.

Even little details like whether the object was saved with LW surface definitions and then imported into MODO, versus imported without surfaces and surfaced native in MODO, can introduce significant bias in the comparison. In order to yield a "neutral" comparison, either both need to import and deal with interpreting "foreign" surfaces (via FBX or whatever), or both need to receive optimal-for-them native surfacing from scratch.

1. Please provide both scene pkgs.
2. Please tell us what (if anything) was loaded versus imported into LW (and from what format), as well as what was loaded versus imported into MODO.

If you can do that, the folks here can give a much better assessment of what changes (if any) are needed to yield a "neutral" comparison.

ActionBob
07-10-2017, 10:46 AM
Just at a glance, and the time given for the render, I am going to venture that the Lightwave version does not have radiosity enabled. The shadows on the lightwave render are quite harsh and dark without any discernible bouncing going on. Yes, a white background (even if background is excluded from radiosity calculations) would better show the difference in shading in the shadow areas.

It has been some time since I have used LW's render engine for radiosity solutions (Have Octane / other renderers) that I can't remember, but seem to recall, you need to enable radiosity in LW in two places. There is a check box and something else.

Anyway, I am suspect of the render time on the LW image (Not that I am a fan boy of Modo - haven't touched and and no desire), but it seems something is off.

-ADrian

erikals
07-10-2017, 11:31 AM
something is off.
it is.

the LightWave render has way fewer bounces than the Modo one, so no wonder Modo renders slower.

look at the shadow area in the LightWave render, totally black, it shouldn't be.

hrgiger
07-10-2017, 11:45 AM
I don't think its bounces. By default they're both set at 1 so if he didnt' change it, that wouldn't make the difference. I'm guessing he's using LW's interpolated radiosity which is fast but not very accurate. Modo by default has many more samples on shading and ray recursion than LW. And I'm trying a monte carlo side by side myself using a luminous polygon and Modo is much faster with making the settings as similar as I can think to do (there's no way to make them match exactly) and LW produces a lot of dark areas where Modo Illuminates them. I'll put something up as soon as LW gets done rendering.

MichaelT
07-10-2017, 11:59 AM
That scene can't be set up correctly in LW.
Made a quick scene for comparison (still not entirely correct btw)

137333

hrgiger
07-10-2017, 12:09 PM
That scene can't be set up correctly in LW.
Made a quick scene for comparison (still not entirely correct btw)

137333


Didn't really think he was going for correct, thought he was just doing for a similar scene rendered in both apps. Although, I don't think he's doing that really either.

erikals
07-10-2017, 12:14 PM
I don't think its bounces.
true, i should've written "bounces"

MichaelT
07-10-2017, 12:17 PM
Well, I wasn't after "correct" either :) Just that the scene can't be right. There is no radiosity to speak about in it.

hrgiger
07-10-2017, 01:35 PM
I'm not really finding consistent enough results to think that I have a reasonable comparison between the two renderers. With both render outputs set to srgb color space, I'm getting very dark areas in the LW renders that that I can only remedy by increasing bounces.

Both renders used only one bounce monte carlo, 8 passes of AA, 1024 rays fired. I disabled importance sampling for both, disabled interpolated in LW and Irradiance caching for Modo which makes them both Monte Carlo. 50 minutes in LW and about 10 in Modo. Like I said, I don't think this is even a good comparison because I can't be sure there aren't more settings that are skewing these so I wouldn't put much faith in them.

Neither of the scenes used any lights. I deleted the light and environment/visibility settings in Modo and disabled all the light display visibility in LW. I also removed all the ambient light from LW, Modos is off by default (as it should be) so the only illumination in the scene is a single luminous polygon.

I'll attach the .LWO object for the scene, people are free to test it out themselves if they own both Modo and LW. You'll need to make the mirror box in Modo reflective and the light top polygon Luminous in Modo.

137334137335

you may need to rename the attachement to a .lwo137336

erikals
07-10-2017, 01:42 PM
50 minutes in LW and about 10 in LW
what? :)

hrgiger
07-10-2017, 01:46 PM
what? :)

Typo. You can see the screenshots.

MichaelT
07-10-2017, 02:58 PM
In any case.. modo have a better renderer (currently, until perhaps whenever LwNext is real) because both have the same (original) author, and it is self evident that he would not make a lesser renderer the second time around.

prometheus
07-10-2017, 03:03 PM
it is.

the LightWave render has way fewer bounces than the Modo one, so no wonder Modo renders slower.

look at the shadow area in the LightWave render, totally black, it shouldn't be.

Actually ..it should be totally black like that, if you use the backdrop only radiosity that is, but you are orrect that if you use montecarlo or final gathering, then the diffuse surface should light up the plane much more, on the other hand it also depends on how much diffuseness there is, lower values would probably yield equal darkness, so this leads me to think he simply used the backdrop radiosity model, and since all is black in the backdrop, you will most likely not see any difference at all wether radiosity is on or off.

with some sort of ambient light value like default 5, you need to check use ambient occlusion ofcourse, or it will look horrible...generally you may actually not want ambient light value at all, but in some exterior renders it can be usefull since we have ambient occlusion.

But it is too little information..what light is used, what radiosity model is used and how is the surface settings.

erikals
07-10-2017, 03:09 PM
Actually ..it should be totally black like that
not totally, as it should receive some light from the plane itself. the LW version is too dark, the Modo version is too bright. (imo)


But it is too little information..what light is used, what radiosity model is used and how is the surface settings.
yep, too little info.

though like others say, LWnext will change the game, in some way.

prometheus
07-10-2017, 03:22 PM
not totally, as it should receive some light from the plane itself. the LW version is too dark, the Modo version is too bright. (imo)


yep, too little info.

though like others say, LWnext will change the game, in some way.

No..I donīt think so, havenīt you tried using radiosity in backdrop only radiosity mode? with total blackness in the backdrop/environment? backdrop only means it will not fire off any rays based on luminosity or diffuseness in the material..if you somehow do, please let me know, and turn of ambient light as well, use one distant light only.

gerry_g
07-10-2017, 03:45 PM
but everything that could possibly throw light into the scene is contained within the box, so why concern for BG Radiosity

erikals
07-10-2017, 04:10 PM
For "Correct" result in LightWave MonteCarlo non-interpolated is the way to go.

(even then it won't be 100% accurate, but closer)

prometheus
07-10-2017, 04:12 PM
but everything that could possibly throw light into the scene is contained within the box, so why concern for BG Radiosity

Donīt understand what you mean here? concern for BG radiosity? why not ...itīs about clarifying what it does and doesnīt, though I personally wouldnīt use it that much, but if you do want faster radiosity and not bouncing rays, but still getting decent renders, backdrop radiosity is what you can use..If you have any kind of color intensity in the backdrop color, or any image or gradient, unless you have that..it really doesnīt matter if you turn on radiosity or not, it wonīt make a difference as newt once said :) Radiosity do not come out at night or complete darkness.

so what do you mean by everything that could possible throw light in to the scen is contained within the box, thereīs no box here, and there isnīt anything that can throw light in to the scene which radiosity would take in account, so what do you mean?

I talked about that radiosity mode, since I thought that what he maybe used, and since erikals mentioned it shouldnīt be so black, but it actually should if you use backdrop radiosity mode, I canīt see no way how radiosity in backdrop mode would fire off rays to lighten up the plane just as it is....Unless there is a color value more than zero in the backdrop color, and unless there is a fog or ambient lighting.

It would however be best to talk about it when we now the facts, what radiosity mode he was using etc.

djwaterman
07-10-2017, 04:24 PM
Mine rendered in seconds, I used Louis De Mont's GI settings, I actually halve these settings for professional renders.

137338137339

gerry_g
07-10-2017, 04:28 PM
was talking about box bench mark scene used by hrgieger above not the original aeroplane scene, only thing illuminating it is a luminous poly on the ceiling bouncing off the internal walls of the box, ps you have a great presentational voice, don't know why you think it's odd

djwaterman
07-10-2017, 04:34 PM
Of course my render was interpolated, but all my renders are interpolated, I've never done non interpolated renders ever, apart from animation I don't see the major benefit.

By the way, the Modo render looks like the walls have reflective fresnel surfaces.

prometheus
07-10-2017, 04:38 PM
was talking about box bench mark scene used by hrgieger above not the original aeroplane scene, only thing illuminating it is a luminous poly on the ceiling bouncing off the internal walls of the box, ps you have a great presentational voice, don't know why you think it's odd

Ah..ok, misunderstood since you didnīt refer to any specific post, just right after mine.

Thanks for that, well ..I know I can speak decently..when I am alert and in mood for it that is, or if Ivé been speaking a bit before I go at it, but this was late night and not prepared, and the very first one..so when you start to think about how to present things best and in a different language, It may not flow naturally so some words comes in the wrong way, like " speaking of the software Krita and the sharp maskīs, where I should have said .."the masks will simply be too sharp" and it turned out.."the masks will be too sharp ...simply" :)

Thanks for the feedback on the voice, appreciated.

prometheus
07-10-2017, 04:41 PM
Of course my render was interpolated, but all my renders are interpolated, I've never done non interpolated renders ever, apart from animation I don't see the major benefit.

By the way, the Modo render looks like the walls have reflective fresnel surfaces.

Might be easire to avoid flickering with non interpolated, with my new machine it now seems acceptable in speed as well.

erikals
07-10-2017, 04:45 PM
by the way, don't forget the Gerardo / DPont denoiser trick, might be of use >
http://forums.newtek.com/showthread.php?146003-Gerardo-Estrada-DPont-Denoiser

hrgiger
07-10-2017, 04:56 PM
Mine rendered in seconds, I used Louis De Mont's GI settings, I actually halve these settings for professional renders.

137338137339

Well I could make a quicker render in Modo as well but the point was to make a scene in lw that was similar in setup to Modo to compare the two. Personally I don't see a way to make a comparison that makes a lot of sense. In the end as always it's going to come down to personal taste. If people want speed they should go with octane.

TreyX
07-10-2017, 05:32 PM
:foreheads

samurai_x
07-10-2017, 08:00 PM
The quality and complexity of this scene is too low.
There are ton's of free scenes on the net better suited for stress testing renderers.

Over
07-10-2017, 09:01 PM
Well, It seems I was all wrong.:foreheads Thanks jwiede and prometheus for the explanation.

I was getting an all LW object(no textures), loading it in Modo and then rendering it. I was changing only radiosity and ambient lighting in the tests I done and LW ends up faster always, but it seems that many more things need to be tweaked to have a more equal result, if thatīs really possible, as Modo seems to do more calculations by default. Iīm using interpolation in radiosity settings in LW, so that speeds up the rendering.

Keep in mind that I design aircraft parts and only recently started in this "art" world as a hobby, it turns out that Iīve been able to get good buck with my hobby.:hey:

Iīm using LW for about a year now and I think I know enough about it and decided to test Modo, so been with it about a week.

erikals, as prometheus said, when you turn radiosity at only 50% intensity with a black background and no ambient lighting, the result ends up like if no radiosity was enabled at all.

I will put my last tests anyway.

137342 137343

137344 137345

Can somebody tell me why the shadows area in Modo is so noisy?


The quality and complexity of this scene is too low.

I hope youīre not referring to object complexity.

Over
07-11-2017, 09:41 AM
Ops... sorry, mean jwiede and hrgiger (Iīm not able to edit my last post)

gerry_g
07-11-2017, 10:57 AM
when you turn radiosity at only 50% intensity with a black background and no ambient lighting, the result ends up like if no radiosity was enabled at all.

think you need to look more deeply, Modo has fresnel on all surfaces by default @ 100% and has a very high specular value compared to LW so you get way more of an ambient look without doing anything to the settings, still your black background is your problem, you need to go put something in there like an HDRI spherical image to make a realistic comparison, when you have things set up similar for both you will usually find Modo way faster.

prometheus
07-11-2017, 12:07 PM
Well, It seems I was all wrong.:foreheads Thanks jwiede and prometheus for the explanation.

I was getting an all LW object(no textures), loading it in Modo and then rendering it. I was changing only radiosity and ambient lighting in the tests I done and LW ends up faster always, but it seems that many more things need to be tweaked to have a more equal result, if thatīs really possible, as Modo seems to do more calculations by default. Iīm using interpolation in radiosity settings in LW, so that speeds up the rendering.

Keep in mind that I design aircraft parts and only recently started in this "art" world as a hobby, it turns out that Iīve been able to get good buck with my hobby.:hey:

Iīm using LW for about a year now and I think I know enough about it and decided to test Modo, so been with it about a week.

erikals, as prometheus said, when you turn radiosity at only 50% intensity with a black background and no ambient lighting, the result ends up like if no radiosity was enabled at all.

I will put my last tests anyway.

137342 137343



137344 137345

Can somebody tell me why the shadows area in Modo is so noisy?



I hope youīre not referring to object complexity.


donīt have modo installed but have tried several versions, for the noisy shadows, I suspect that is equal to lightwave..you may need to go to that light and check how many samples it has, equal to go to the lightwave render panel and raise light samples up when using area, spherical and other lights.
I think those samples should be in modos lights settings, or go to render settings and check there, it also matters how much aa in general is set.

The question of comparing the speed is very difficult, especially if you go with default settings in lightwave an compare to default in modo, what would probably be more accurate, that would be to use the lowest settings possible in aa and light samples, but even then it would probably not be a fair comparison.