Gamma 2.2 comparisons between Vray and Mray

[ARTECONI-CGI]

Hi all,

I am spending this not too busy day at work (after an apocalyptic deadline) to run some comparison tests between vray and mental ray behaviour in 2.2 color space.
I am pretty sure it is an interesting topic for most of you so it will be nice to get your comment and impressions about it.

What I did is lighting the same white scene both in vray and mental ray trying to achieve a similar result.
I have already done this test in past and noticed that the 2 rendering engines deliver a pretty different result.
At the time I have noticed that while mray keeps his shading consistent throught the 256 grayscale range, vray seems to lift up the blacks and open the shadows.

I have tried to recreate this situation today and got some interesting result.
The goal was to check how close the 2 rendering engines matches without having to use tricks and post color correction.

The scene has been lit by 5 spherical photometric lights with soft shadows (for vray and mray) keeping the same paramateres wherever possible.
I didn't use a linear workflow but I have tried to use the same reinhard/highlight burn value on both of them set at 0.05 and used a mental ray arch material with a white value of 0.7 on both the situation software for the material override.
The only value that has been changed to match the amount of light is the multiplier in the vray color mapping tab. This should work only as a mutiplier for the value of the light so, technically shouldn't be considered a tweak.

The first test has been run with the following settings for mental ray.
Final Gather on with 3 bounces - No Global Illumination
Gamma & Lut panel: all on at 2.2
highlight burn: 0.05

This is the straight output from the render:



The second test has been run in vray using irradiance map + light cache with the following settings.
Gamma & Lut panel: all on at 2.2 (except the output value at 1)
Color Mapping Reinhard
Burn Value: 0.05
VFB on

This is the straight output from the render:



And this is the render after I have applied a curve in photoshop to match the mray output.



You can see how much the first 2 renders are different. It seems like that I had to compensate the 2.2 curve in photoshop to get back to a render closer to the mental ray one.
I agree that the final gather doesn't push the light as far as the irradiance map + light cache solution but overall I keep on thinking that the mray solution looks a little more photorealistic and believable than the vray one in that case.
I know as well that the 2.2 setup is meant to be color corrected in post but I wonder why do I get such a big difference between the 2 rendering engine.

Now I want to show you a further test I did with vray in gamma 1 instead of 2.2.
I didn't change any other parameter except for the gamma and input/output set back to 1 and a value of gamma 1 as well in the reinhard color mapping.

This is the straight output from the render:



This output is really close to the mental ray one and seems to be more correct to my eyes. But the Gamma settings for vray are set to 1 meanwhile they are set to 2.2 in mental ray.

A last final test with the LWF option on available in the color mapping and the gamma set to 2.2.
Gamma & Lut panel all off.
The result is closer to the render above.

This is the straight output render:



Again the render looks closer to the mental ray one.
I am not saying that mray is correct or better I just wonder why they're so inconsistent.
I'd like as well to run a comparison with Arnold to check which one is closer to which.

I'd love to hear your thoughts and opinions about it.

Kind Regards,

Giacomo.

[MoonDoggie]

You're using LWF in the first test but then applying a Reinhard color curve to the result in both tests? Perhaps MR doesn't use the same algorithm for Reinhard color mapping?

A better test would definitely be to leave the color mapping at linear 1.0 1.0 gamma: 2.2 and do the tests.

[grantwarwick]

You're using LWF in the first test but then applying a Reinhard color curve to the result in both tests? Perhaps MR doesn't use the same algorithm for Reinhard color mapping?

A better test would definitely be to leave the color mapping at linear 1.0 1.0 gamma: 2.2 and do the tests.

OP obviously doesn't understand the word "linear"

[vlado]

I didn't use a linear workflow

If you did not set display gamma to 2.2 in the 3ds Max render settings, the mental ray photographic exposure will still apply gamma 2.2 to your output behind the scenes. That immediately makes any comparisons invalid and it's why you had to fiddle with gamma 2.2 curves in Photoshop.

but I have tried to use the same reinhard/highlight burn value on both of them set at 0.05

The mental ray photographic color mapping and the V-Ray Reinhard mapping are completely different. Burn value of 0.05 has different meanings in each of them. This also makes any comparisons somewhat pointless. I agree that the mental ray mapper gives a more photographic feel to an image (that's why it's called "photographic"), whereas V-Ray simply tries to avoid burn outs - nothing more, nothing less. We have not particularly tried to match a photographic look with it.

I agree that the final gather doesn't push the light as far as the irradiance map + light cache solution but overall I keep on thinking that the mray solution looks a little more photorealistic and believable than the vray one in that case.

The mray one is also incorrect - regardless of how it looks to your eyes. One can play with curves to get a good image, but you can't compensate for the missing light. It just goes to show that you can't really trust your eyes to figure out if an image is correct or not - just whether it looks better or not. Of course, if a nice image is all you want, that may be perfectly fine.

Best regards,
Vlado

[ARTECONI-CGI]

Hi guys,

Thanks for the reply.

Colin I didn't use a linear color mapping but a reinhard with gamma set to 2.2.
The reason is that in an ideal world I would render linear and then tonemap it in post with nuke or AE.
The reality is that I work for archviz, mainly interiors and there's never time to do things properly, so, most of the time I rely on a exponential or reinhard color mapping.
As well I already did a test in past with the linear workflow and wanted to try an approach I can use at work.

Vlado,
1- The set display gamma is set to 2.2 in the first 3 images. So a proper linear workflow except for the reinhard tone mapping. In the last 2 render I have set it back on purpose to 1. The goal for those 2 images was to reach a closer look to the previous one.
2- I didn't know that the mray and vray color mapping were different. I have assumed that they were similiar because both of them control the burning somehow so I gave it a go. What vray colormapping do you reckon is the closest to a photographic look when using a 2.2 setup?
3- Overall I am happy with vray, but sometimes I struggle to get a photographic look getting a flat image lacking of shadows. (blacks / shadows / midtones / bright / highlights)
Can you suggest me the best workflow to get a photographic look in vray?

Thanks again,

Giacomo.

[Morne]

3- Overall I am happy with vray, but sometimes I struggle to get a photographic look getting a flat image lacking of shadows. (blacks / shadows / midtones / bright / highlights)
Can you suggest me the best workflow to get a photographic look in vray?

Thanks again,

Giacomo.

Stick to full linear workflow. Make your shaders a bit darker than what you're use to. IE if you have a white object then dont use 255,255,255 for your rgb. Try something between 200 and 235 for whites.
You'll get much better contrast this way and less light burns

[ARTECONI-CGI]

Hi Morne,

Thanks for the suggestion.
This is my usual workflow but In the last years I have largely experienced with the exponential and reinhard as well.
Yes using darker colors helps but I am always looking for the perfect lighting and I have always the feeling that it lacks something.

I mainly refer to architectural interiors that I consider the most difficult situation for lighting.

Regards,

Giacomo.

[Morne]

Stick to real world lighting values. Use IES lights from the lighting manufacturer where possible. Think like a photographer and add fill lights where needed. If you don't know what a fill light is, then just start by adding a big vray light just in front of your camera to simulate a camera flash. Once you use to that, move the light to the side a bit so your scene gets light from a side angle. Depending on the scene you can use sometimes 3 or 4 fill lights.

Also look into 3 point lighting for some tips on fill lights