Hi guys,
my website is officially live and to help promote the site I’ve released lesson 1 for free to give you an idea of what to expect from my course.
http://www.grantwarwick.net/mastering-vraynew-page/
Hey Grant
Thanks for the lesson. I was quite curious as to what you show and teach in those builds and training sessions…
I do however find few issues with it…
1. as 1st lessons shouldn’t you take care of topic as to which DMC sampler to use? Subdivision or Adaptive? There is big difference between them and u should use them in different situations…
2. By setting up Adaptive amount in DMC settings to .9 you are basically telling vray " I’m nob take care of all the settings for me" - reason why u had to go up to 150/180 samples on lights and shaders. Just lower that to 0.5 and ull see that you will have more control over scene and see results as expected not as vray adjust them. In some situations when I was at 0.9 with even with 1000 samples I could not clean some noise…
3. When you using LC with Store direct light. You actually ask GI/LC to take care of some math over direct lights which mean that DMC settings has affect over ur lights and sampler - making it more complicated and so on.I think its also reason why ur light pass gets darker when you use GI. - but I’m not 100% as to this point tbh… - this is wrong now ! I remember that its darker when you use GI because of the MIS on lights and so on that kick in…
Other than that very nice 
Thanks for share.
Thanks for your feedback, I knew I’d get something from you haha 
1: The DMC Sampler is the most commonly used AA engine and this was the very first intro lesson. I am trying to make it simple for beginners as well here as we all know it’s very easy to get lost with when to use what so as I point out in the video it’s a quick run through and later on I will push adaptive subdivision.
For me personally though, it’s rarely rarely used, maybe it’s the similarity from job to job but it’s not often I have scenes which need more evenly distributed samples
2: As far as I’m aware, dropping to .5 is saying no matter WHAT, each pixel needs 50% of scene subdivisions, it doesn’t matter if there is noise or not just shoot 50% into the scene… So I don’t agree with you here but would love to be proven wrong. If anything this is more the way Arnold behaves where there isn’t the adaptivity Vray has which I’d consider that being the noob route 
So for example on jobs that need large subdivs everywhere that’s the only time I’ll go low on the adaptive amount.
Would be great for Vlado to confirm this once and for all.
In future settings, I dropped the noise threshold to .005- .007 to fix the issue of needing like you say 1000 samples in some scenario’s, making detailed tuts aint easy so as I also point out to customers, there will always be updates to the methods use
3: I didn’t know that so thanks!
Sureee u just being biased towards DMC adaptive
DMC Adaptive have more random samples distribution than dmc subdivision. DMC adaptive is like search and destroy he looks for details where as subdivision is for grid based smooth details… - if I’m not wrong so if you want evenly distributed samples then u might be using wrong engine 
- its also the reason why dmc subdivision can get pretty perfect smooth light with 12-16 samples
2. I think the same thing. If you have 0.5 set up and u cast 100 samples from light. The 50 will be cast definitely and 50 will be based on DMC settings threshold. I just found out from my experience that there is a lot of trouble when it comes to use 0.9+ adaptive value to control both light,gi,refl,refr(IF I’m not wrong LIGHT samples are not included in that equation and they are ONLY affected if you are using GI with store light !) quality so I tend to go for 50% as then I can predict the result. 0.9 its like Russian roulette I never can get done anything with it !
3. yea there was post by one of developers showing it in action… I’ll see if I can find it and post here.
PS here is the post : http://forums.chaosgroup.com/showthread.php?76752-Light-Cache-Question&highlight=store+direct+light
You don’t find you are oversampling parts of your scene with a setting of .5?
Maybe there needs to be a sample rate for scene subdivs as well as AA subdivs?
I haven’t experienced this issue of russian roulette, I find it really easy to work with actually! I have to squeeze every penny from the renders mostly (we only have 3 render boxes atm hahahahaa 
)
I would love to be able to go lower and just forget the settings as much and focus on the work but atm I can’t do that.
How many render boxes do you use at work? I think our patience levels must be very different huauhaua
If you are really interested I can put you on the subscription as you are one of the guys who knows more than me and can probably clear up some grey areas.
Thanks, look forward to seeing it haha for a second I thought the info might be in the documentation so I’m glad to know it was only shown in a forum post lol! Don’t make me look bad haha 
Hey
Well when you work with MR you know how to deal with samples. U dont type 9999 u just go slowly. I do the same with Vray. I’ts bit difficult to explain. Lets say I have light. And then the light has 5 meters range. Now 0.5 adaptive means that 50% 0-2.5 meters of that light range(ITS NOT REALLY WHAT HAPPENS!!! :X but should illustrate…) are at 50 samples and the other 2.5 - 5.0 meters are adaptive - now it does not work this way but its good example I think. Now because of that I know that 50% of brightness and part of light will have good sampling and the gray/dark area which I don’t care too much about are taken care by DMC sampler.
I think I would love to have a AA min max/ sample min max/ lights min max/refl min max/refr min max/ and everything else I can have min max/ amount and so on I’d spread it all in to singular control points fur full on best ever control I could have
I have 1 render nodes… so trust me if it renders longer than 10 mins I start running around room screaming - I abuse amazon when I need to, god bless 10 render slaves on demand
Sorry for making you bad if I did:x I keep watching and reading posts on forum from all u great guys. Every tutorial I read /watch I always find things I dont agree with and so on. I think I should just make my own damn tutorial and stop ruining other peoples topics!
hhahaha yeh it’s all very confusing! Not everybody is gonna have the same workflow, I used to use Adaptive subdivision before I understood DMC! Because it gave me faster/cleaner renders. I was obviously messing around with wrong settings in DMC and gave up. this is a few years ago now 
I’ll chat on skype, I was only joking about making me look bad Of course I don’t expect to be perfect but the feedback I am seeing from subscribers has only been positive, Nobody so far has complained about this method taking longer but if you are much more used to using Adaptive sub you will have the little secrets I don’t.
In lesson 3 I used this same method on Peter Guthrie scene also and haven’t heard from the CG guys to try adapative subdivision.
Are you doing cars a lot? This is one example where I happily can swap to adapative subdivision if it’s a glossy car paint.
anyways, I am heading to sleep and we can talk about it! I’m more of a scientist mindset than a religious one so I’m not stubborn to change if the render times/quality speaks for itself.
Grant, looks good.
Can we get a higher res version of that image showing all the materials rendered?
Was going to do my own, but… time etc
Is the download section up and running yet?
If one wants to do speed tests and such would it be technically possible to convert your materials to other engines? I got the impression they are at least somewhat procedural?
Ah yes, the classic drug-dealer trick, give the first ‘hit’ away for free and they’ll be back as customers!
I watched episode 1 and immediately bought the subscription, although the materials alone are worth the sub fee. Looking forward to seeing what you come up with next
Hey Grant, great tutorial…thanks!
Just wondering is it important to do test renders following your method at the final output resolution?
I’m not sure, but am I wrong to assume that the image sampling is linked to the amount of pixels being output?
I know from experience that you can dial down the GI settings a bit when doing very large renders, but not sure about the other sampling (lights, mats, DMC)
Cheers
Great stuff Grant! Really appreciate you putting out the first video for free, hope it translates to a lot more subscribers!
I’ve had a chance to put some of these approaches to sampling into practise recently and can definitely see the merits of being able to tweak each noisy aspect in isolation. That said, in the tests I’ve done between this and my normal approach (8 subdivs across the board, image sampler tackling pretty much everything) I’ve only noticed a marginal saving in render time (2 mins down to 1:50 for example) which is arguably negated by the time it would take me to refine the separate subdivisions for each light, material…etc.
I still need to do more tests as there’s every chance I’m missing something (or just stuck in my ways). It could also be that I’m used to having to add noise to images to help with artefacts/banding when things hit CMYK, so I’m less fussed with noise in my renders. I’m curious to know if this is something you start tweaking straight away or do you use rough settings and then clean everything up once you start taking things to final?
awesome! I did want odd add that your Twitter share is to long; nobody can RT it. You should keep your Twitter post below 120 characters, which will alow for the RT and persons @handle.
Is Grant on Twitter? Ignoring that, what you’re describing is a manual retweet which is generally considered bad practise, it’s better to use the RT button which allows the original tweeter to keep track of shares/favourites and allows for the full tweet to be seen in its original context. Unless I’m missing something and there’s a tweet link somewhere that’s over 140 characters
It’s been fixed It was the Vimeo share button, which had a long line of hash tags, which would have required me to edit until I got under the 140 characters.
Aha! Sorry Bobby!
don’t let it happen again :)[quote=“Steffen Dünner, username:Steffen_Dunner”]
Aha! Sorry Bobby!
[/quote]
Yep, uploading them for you right now.
Download section is up and running, did you get the email? Some people asked but it turned out they simply missed it in their inbox
Everything but bercon maps and the functionality with the custom falloff curves is currently translatable to Maya. I’m currently working on some tests before we lose our maya license in a few weeks!
Thanks mate! It’s going really well, I’ve had a second surge of subscribers since the website has gone live so I’m really happy.
I definitely do region renders at full res throughout the job and only push out low res tests for clients.
The way the AA works is based on the pixel information so a larger render means more spread out pixels = less potential work for the AA engine.
lesson 3 covers advanced scene optimisation. Peter Guthrie used your version of settings but I effectively reduced his render time 25%!
My method seems to have substantial benefits across all scene types so far.
I vowed never to own a twitter but god have mercy on my soul I considered it today!
Yep… missed it. Cool thanks, sorted now.
You can also access the highres material previews in the subscribers sections.
I have massive plans for improving lighting and material previews at the moment. Going to struggle to get time but picture a dedicated page with previews from multiple HDR’s, downloads for the HDR’s themselves, generated color palettes and different geometry previews all in one tightly organized area.