Cleanup noise in Elements: Reflection and Shadow - Also LOOONG render times

[svetlozar.draganov]

I think everyone at some point has thought about the magic "best render" button - but the main idea behind Vray is to be adaptive/fast and at the same time to produce the best quality.

[Morne]

this is what I don't understand. Vlado all the time say NOT to use that "magic" LWF button as its not magic, but instead use affect "dont affect colours".
2 or 3 years ago when ChaosGroup was doing a demo in South Africa, the guy from Chaosgroup that was here also said to just use that button and ignore the max gamma.

Now I ask you, why the conflicting info? Also, if you use the "linear" magic button and NOT the "dont affect colours", wont this mess up your LWF exr and vrimg output so that when you open in another app, its all messed up? (I can test it myself, but my pc is busy right now)

no answer to this yet?

I tried this on a real scene with lights and cameras, and the one with the lwf button ticked came out flat, grey and low contrast. I used the settings from your youtube video and so far I don't agree with the use of the lwf button. Am I missing something?

ALSO, in the same scene using the 2 different methods, in the one with the LWF button ticked I got a strange HUGE black SPOT in my reflection and rawreflection elements, which I didnt get in the normal "correct" lwf method.

Any hints or further comments about the "magic lwf button"?

[svetlozar.draganov]

We would like to see your render setup - could you please send us both 3DSMax and Vray Gamma setups on different files at vray@chaosgroup.com
Don't forget to mention which version of Max and Vray you are using.
A few screenshots of 3DSMax and Vray Gamma renderings would be very helpful too.

Regards
Svetlozar Draganov

[trixian]

As far as I can see in your video, you are not really seeing the correct result, but the linear data. Usually we want to take this rendered output into say Nuke (or other comp suites), and it then usually want to be read as linear and given the 2.2 \srgb LUT. Adjusting materials and lighting while seeing the linear result is exactly what we used to do in the olden days, and makes it really hard to adjust thing the right ammount. Should you not enable "don't affect colours" and look at it with the VRay framebuffers srgb LUT?

[svetlozar.draganov]

As far as I can see in your video, you are not really seeing the correct result, but the linear data. Usually we want to take this rendered output into say Nuke (or other comp suites), and it then usually want to be read as linear and given the 2.2 \srgb LUT. Adjusting materials and lighting while seeing the linear result is exactly what we used to do in the olden days, and makes it really hard to adjust thing the right ammount. Should you not enable "don't affect colours" and look at it with the VRay framebuffers srgb LUT?

In the video is shown how to set-up Vray Gamma if you are not using compositing and the render from Vray is your final product.
If you need to send the result from Vray to compositing software like Nuke - the setup is different - I mentioned about these 2 setups early in this topic.
Here is the link : http://www.chaosgroup.com/forums/vbu...748#post550748

Let me know if you have more questions

[zeronove]

you are not really seeing the correct result, but the linear data.

trixian,
may I ask you why are you saying that? ...he has VRFB gamma set to 2.2 and burned into image (don't affect colors is unchecked) ...afaik to show the linear image gamma should be 2.2, don't affect color checked and sRGB unchecked, isn't it?

[joconnell]

Svetlozar,

Many thanks for the detailed reply to my questions - so it would pretty much seem that problem is one of efficiency? So the problem or slow downs come in with the wrong settings making vray loop through the levels of AA and dmc sampling and terminating later, rather than AA sampling being more efficient than DMC for example? It isn't that AA or DMC rays are quicker or slower than one another, it's more that using the wrong settings is causing vray to work through more levels to get a clean result. Of course it's going to be tough to know exactly what settings to use each time without testing, I'll take some old scenes and apply your method to them! Maybe it might be worth doing some kind of script to automatically prepare a few different methods of AA vs samples as a test bed? Like for example one would be all materials and lights at 256 samples with aa 1,4, another with 2, 16 AA and 32 samples, and a last one with 1,100 and other samples at 256?

[svetlozar.draganov]

Svetlozar,

Many thanks for the detailed reply to my questions - so it would pretty much seem that problem is one of efficiency? So the problem or slow downs come in with the wrong settings making vray loop through the levels of AA and dmc sampling and terminating later, rather than AA sampling being more efficient than DMC for example? It isn't that AA or DMC rays are quicker or slower than one another, it's more that using the wrong settings is causing vray to work through more levels to get a clean result. Of course it's going to be tough to know exactly what settings to use each time without testing, I'll take some old scenes and apply your method to them! Maybe it might be worth doing some kind of script to automatically prepare a few different methods of AA vs samples as a test bed? Like for example one would be all materials and lights at 256 samples with aa 1,4, another with 2, 16 AA and 32 samples, and a last one with 1,100 and other samples at 256?

Every scene is different and there is no general rule whether Image Sampler or DMC will be more efficient.
We will be glad to see the results of your tests

[trixian]

Well first of all, the video is dealing with consistency of colours between view port, materials, and render output. He is selecting colours that are displayed "non gamma corrected".
Not quite sure how, but if he is achieving a match between a linear colour swatch (not affected by Max gamma since it is off), and his vfb rendering, then there is something fishy going on in the "VRay gamma". My point is that his colour selection is in linear space. It seems you either select colours in linear where you cant really see the difference on dark tones, or gamma corrected, where you have the same problem in the lighter tones.
Anyway, this issue of consistent colour is completely moot, as once you turn on GI, add lights of different colours (specially the VRay light in dome format that displays like an omni in the view port) this "match" is long gone.
I also tried to set up a simple scene with the suggested VRay Gamma setup, without max gamma, with GI on and using the "don't affect colours" check box while viewing it in the VFB with the sRGB LUT enabled.
What I don't understand is how the "Linear work flow" checkbox can give results that are so far off compared to using max's gamma in conjunction and still be "correct". Are neither of them correct, and we left with educated guess work, or is there some secret sauce I somehow have never discovered?

[kosso_olli]

I cant get this to work correctly, either. When I follow the steps from the video, I get an image that is much darker then what I am used to. Have to admit that the resulting images are much cleaner and less noisy. But as said, the images I get are much darker.

[svetlozar.draganov]

Is it possible to send us your scene - we will try to help you with this setup

[Dman3d]

Well first of all, the video is dealing with consistency of colours between view port, materials, and render output. He is selecting colours that are displayed "non gamma corrected".
Not quite sure how, but if he is achieving a match between a linear colour swatch (not affected by Max gamma since it is off), and his vfb rendering, then there is something fishy going on in the "VRay gamma". My point is that his colour selection is in linear space. It seems you either select colours in linear where you cant really see the difference on dark tones, or gamma corrected, where you have the same problem in the lighter tones.
Anyway, this issue of consistent colour is completely moot, as once you turn on GI, add lights of different colours (specially the VRay light in dome format that displays like an omni in the view port) this "match" is long gone.
I also tried to set up a simple scene with the suggested VRay Gamma setup, without max gamma, with GI on and using the "don't affect colours" check box while viewing it in the VFB with the sRGB LUT enabled.
What I don't understand is how the "Linear work flow" checkbox can give results that are so far off compared to using max's gamma in conjunction and still be "correct". Are neither of them correct, and we left with educated guess work, or is there some secret sauce I somehow have never discovered?

Yeah what he said

[PDI]

Hey Morne,

I had a look at your scene made a few changes and managed to render out the following.



I rendered it at double the image size you set (ie. 2560x1440). The render time is 36min. I did cheat by using 3 machines to render, so lets multiply the time by 3. This gives us 108min.
If you were to render this at half the resolution as specified you could halve the time. So 108min/2 = 54min.

Some of the problems I found as discussed above is the over bright areas at the end of the tunnel. I turned the exterior light down to avoid this. I will try a test render with subpixel mapping on to see if that helps.
This is a great thread and I have had the same problems as Francesco Legrenzi with glass objects and render time. My pet hate is glass chandeliers.

I find rendering 1/3 art, 1/3 science and 1/3 magic...

[Morne]

My final complete scene I managed to render at 3000 wide with DR in about 4 hours. There was a LOT of glossy surfaces in the scene. The slight noise that remained in the scene was acceptable as it actually contributed to making it look more realistic. At the end of the "tunnel" there was a glass facade with various different shades of tinted film. I mostly used Dimitry's method, with some minor tweaks. It's a little slow to render, but I'm happy with the results as it saves times endlessly tweaking stuff.

Unfortunately I can't post my final render.

[raymarcher]

Sweet 'thread'.

Could another solution be a "reduce noise" node on problem elements in composting of render elements?