Understanding DMC Sampler

[3LP]

Yeah sure,

Because I would like to actually see the HDRI in the material editor, I need for most of the HDRI to change the overall mult.
Once you have a decent image which is not completely white, you can actually use that to place you HDRI like you want in the viewport through the environment slot.
If it's all white, you can't visually use it and you need to guess and place through rendering 300 times.

Now because the VrayCam got some exposure, and bark everything down, I need to counter balance the low value of the overall mult with the render mult.

I hope this make sense

I haven't had a chance to look at your scene yet, but I will as soon as I got time. Thanks in advance

Stan

[peterguthrie]

yep, what Stan said.

In general it was a bit of a happy accident with my skies that leaving the intensity at 1 meant that when you use a vrayphysicalcamera you can more or less use 'real world' values like f8, 1/200th ISO 100 and the exposure won't be too far off. So when I start a scene I usually set the multiplier of the vrayHDRi to 1 and set the multiplier of the vraydomelight to 1 as well.

Also on all of my skies I decided to centre the sun (or brightest part of the sky) to take the guesswork out of rotating the sky. When you use the vrayHDRi map you can use the attached diagram that a friend of mine made as a guide.

[Morne]

yep, what Stan said.

In general it was a bit of a happy accident with my skies that leaving the intensity at 1 meant that when you use a vrayphysicalcamera you can more or less use 'real world' values like f8, 1/200th ISO 100 and the exposure won't be too far off. So when I start a scene I usually set the multiplier of the vrayHDRi to 1 and set the multiplier of the vraydomelight to 1 as well.

Also on all of my skies I decided to centre the sun (or brightest part of the sky) to take the guesswork out of rotating the sky. When you use the vrayHDRi map you can use the attached diagram that a friend of mine made as a guide.

[ATTACH=CONFIG]14852[/ATTACH]

I'm lost a bit with this diagram?

[peterguthrie]

I'm lost a bit with this diagram?

Its a plan view. So if you load one of my skies with vrayHDRi and leave the rotation at 0deg then the sun will be shining east to west

[snivlem]

Here's a good read folks,

http://3dlight.blogspot.com.au/

Adaptive Amount= .9 is really handy I think over .85.

[peterguthrie]

Here's a good read folks,

http://3dlight.blogspot.com.au/

Adaptive Amount= .9 is really handy I think over .85.

thanks, will have a read later

[peterguthrie]

This is what I came up with Stan.

As you can see the main speed gain is from the reinhard burn and light cache settings, but then to go from 3 hrs to 33 mins with the new settings is pretty impressive too.



So the actual settings I used were:

DMC Image Sampler

min 2, max 32

clr threshold 0.01

Settings: DMC

Adaptive amount 0.85
Noise threshold 0.005

Lights

generally 80 (use VMC script)
vraydomelight 240

GI

240 subdivs

Materials

generally 80 subdivs (use VMC script to set)
the back grey wall 320
ceiling and right wall 240
other difficult mats (table and chairs) 240

and then the frosted pendant was slowing things down so I set the reflection subdivs to 240, refraction to 320 and the refraction max depth to 2. Obviously this changes the look of it so maybe putting it to 3 would have been better

[3LP]

Nice!
Would it be to much to ask if you could send me the max file alone back in 2012?
I would like to try this out, I'm very impressed!

Cheers

Stan

[GoncaloP]

and then the frosted pendant was slowing things down so I set the reflection subdivs to 240, refraction to 320 and the refraction max depth to 2. Obviously this changes the look of it so maybe putting it to 3 would have been better

I couldn't figure the right amount of glossy refraction but it seams you have nail it. Just wondering, if there is no glossy reflections, what is the number of reflection subdivisions doing in this case?

Thanks

[peterguthrie]

I couldn't figure the right amount of glossy refraction but it seams you have nail it. Just wondering, if there is no glossy reflections, what is the number of reflection subdivisions doing in this case?

Thanks

I must have assumed there were reflections! so to answer your question, nothing

[peterguthrie]

Nice!
Would it be to much to ask if you could send me the max file alone back in 2012?
I would like to try this out, I'm very impressed!

Cheers

Stan

pm sent with a link

[GoncaloP]

I'm still puzzled with the HRDI value... Using a vray camara with Vray sun with the same angle as your HRDI map with defaults 1/1 they produce almost the same amount of light. To be seen in the slot material I was able to figure that multiplying and dividing by the same amount it produces the same final result. But in this case the value used is 25 times stronger than 1, so also 25 stronger than Vray Sun, wich seams a bit odd. I remember that Vray didn't like overbright values much, resulting in a major increase of render time. I would assume that this value is not very reflection/refraction friendly because all the light information that HRDI is able reproduce. Wouldn't be better to increase the camara exposure instead, using a more sensitive film for instance?

[Guest]

There are a few of these threads now and they have all been an interesting read. Although they all seem to arrive at different ideal settings I wonder how much of this will change in V-Ray 3.0, if Brute Force is going to be much faster?

[3LP]

My thoughts are : do we still have to put those settings in.
I mean, I guess, but the whole purpose of progressive is that it's being constantly calculated better and better.
So starting from this, a material after 20 seconds will "look" like it' been render with 8 subdiv, after 5hours, it should "look" like it's been render with 500 subdivs.
I put "look" between brackets because I'm not sure if any subdiv have any real weight here... in progressive anyway... Like RT, no settings, it just calculate and goes...

For the normal type of render, it has been said that BF is quicker, yes, but that doesn't change the base principle of optimizing knowing how the software is working, in this case : DMC subdiv.

Just a side question, does anyone know how AA will be applied to a progressive render? Can we imagine to have a "Cutmull-rom" sharp edge on a progressive render, it means it's refining the AA constantly? doesn't that take more processing power then creating the AA after the whole bucket has been render like usual, I mean, after traditional way of light, GI, etc and AA applied at the end, if it's like that that it's working?

Thanks

Stan

[greysheep5]

probably the universal one is more correct..

how about I try it on this scene http://www.flickr.com/photos/pg/8536...57632942671928 it was a killer to render

hi peter!

you always talk about your or the univeral settings in comparison. could you elaborate one more time what that means (to you)? as i can´t find out what settings those are i don´t know what you are comparing here, are you talking about irradiance map plus lightcache with 1 and 100 min/max dmc settings or something completely different?

cheers,

christoph.