Showing posts with label renderman. Show all posts
Showing posts with label renderman. Show all posts

Monday, 3 October 2011

Wayne's Reviews:: Render man pro server & Renderman for Maya

To most digital sculptors Renderman makes them think of one thing....fast rendering of displacement maps. So it may surprise you to learn that for me that's not the thing I like most about it. My favourite feature in the pro server / Renderman for Maya combo I'm using is the Renderman relighting controls. As I mentioned in my last review lighting a scene can be a pain sometimes if you're not getting the right feedback from your viewport. Renderman has a nice way around this by bypassing the viewport altogether if you wish and relight fairly fast. This mean you get to see EXACTLY what you're going to get at render time and not a viewport approximation.

So the ability to change your lights settings, colours etc is a great boon, although there are certain times when you will need to rebuild your shadow map. One thing would say is that for a scene such as the cathedral you really are going to have to wait a fair old bit. Because even though there is no displacement at all in the cathedral render in either Renderman or Mental ray version (its basically a game model on steroids), when your pushing that much information around even Renderman is going to take a while to catch its breath.

When Pixar were nice enough to let me have a free copy of the latest Renderman for a while I have to admit coming from non Renderman compliant render engines it was a culture shock. Renderman is at its best when arguably it's at its least 'intuitive' to a non coder. If you are prepared to write your own shaders, you can really make it do tricks as can be seen on virtually 80% of all major Hollywood films. If you don't then you are I feel almost gutting th heart out of the render engine in some ways.

While I was using Renderman for Maya, I was also using Renderman pro server alongside some apps of my own that aren't available publicly such as Exodus. This enabled me to take a look at pro server in a less modicodelled environment and use it 'in the raw (and by that I mean without a UI...not naked lol).

I did find both a bit of a pain to set up, although some of that was due to my extremely flaky development machine which is renowned for throwing errors no other machine on planet earth seems to. The one area I would like to see improved is the installer as messing around with environment variables if something goes wrong is no one's idea of fun! But again Pixar came to the rescue and managed to get me up and running in no time with some of the best customer support I've seen anywhere .

So how did I cope with having to use Renderman in Maya when I am mainly used to 3dmax and as a Mental ray user? I'm not going to lie and say it was all very easy , there was a lot of information I needed to take in, in what was a very short time. If you're looking to pick up Renderman and use it in a day or two to its best from nothing then you'd going to be mistaken.

Renderman needs time to be learned and does to some degree at least ask (rather than demand) you to do some coding to get the very best out of it. It's not the fastest ray tracer out there and physically accurate lighting and shading out of the box while not impossible is far harder to achieve than mental ray. But comparing Renderman and Mental ray, although both are render engines is like comparing a fish to an apple. (While both apples and fish are good at being apples and fish, but you wouldn't want a fish on your fruit salad.)

There is no 'best render engine' the best render engine is the one you have and the one you can get the results you require out of it the easiest. I wouldn't call Renderman a good engine for a beginner to learn unless they have a lot of time and are prepared to go quite deeply into rendering theory. Its only as good as the artist behind it and not a magic wand. Overall I like Renderman and can see m using it for jobs that fit it best, the same as I use mental ray for jobs that fit that best. (Plus a little secret...no one ever said you can't mix and match passes from both engines.)

Renderman's renowned fast displacement times are not ironically as big a deal for me as you would think, as that's a very small piece of a larger puzzle that is your rendered frame. But a nice plus is that Renderman can render a Mudbox vector displacement map without any special shaders or complex set up. It's a tried and production tested industry standard render engine an as you would expect it does what it says on the tin.

Sunday, 19 June 2011

OGAM (Of Gods & Men)...crunch time.

I've been a bit quiet of late for a number of reasons. The one that's brought me the most stress and been the hardest to deal with is the making of my short 'Of Gods & Men'. It become obvious a little while back that any hope I had of getting it done by my 2nd day of Lectures at the End User Event in Holland at the end of this month was impossible due to events outside my control. You may think that that was pretty self apparent anyway when you're trying to make a film quality short on a budget of £25, 90% of the time on your own. Had I been able to put the time in I needed , then there was a very good chance it would have been done.

So I made a decision to finish half of it and in the spirit of the last Harry Potter book split it into 2 shorts (that I'll release eventually as a single short.) Hey if it's good enough for holly wood its good enough for me lol. This should have made things far easier, but that's not exactly the case, as I still have 3/4 absolute nightmare shots to finish and render out. It's the rendering that's taking the time as these shots are :

A - created using my own app 'Exodus' (don't worry I'm not going to release it as it'd be a legal minefield lol)

B - as such they can only be rendered on my machine

C - The render farm I've had use of doesn't have a Renderman compliant render engine ,on it which is what HAS to be used for those shots

D - They have raw polygon counts that are between 100million and 750 million (not counting displacement maps...and all of those polys are animated. I'm pretty sure most people don't want me blowing up their render farm

As exodus is only for me, and is a mish mash of all sorts of crazy code from all sorts of places, it's about as stable as an Italian taxi driver behind 2 nuns in a tractor. This doesn't make for an easy workflow and many man hours have disappeared either on development or solving crazy issues. (such as it suddenly deciding to render bright purple using a constant shader and dump all animation ...Just for shits and giggles it seemed lol. Or the time a forgotten floating point interferred with animation on some characters making them all look like they were having spaz attacks.) That doesn't do good things for your stress levels as come what may I have to have this ready for a complete breakdown on my 2nd En User Event lecture.

I've also had in development for myself only a mental ray PTEX enabling shader, and my own propriety 1 click auto rig and skin solution....and a few other things I'll keep to myself. But alas these items mentioned simply were nowhere near ready for use in even a personal short as stability needs to be worked on. My idea behind developing them is mainly so I can have apps that do what very expensive ones do without having to resort to piracy. Which I cannot and will not condone.

This way I have no license fees to pay, I don't have to wait ages for support emails as I can fix it myself, and if it doesn't do exactly what I need it to I can simply research and add the feature myself. Another reason I won't be putting them out for sale or for free is twofold. Firstly, as not only are there other apps that do these jobs nearly as well (or in some cases better), but also its a mash up of various source code and legally it'd mean I'd make buttons anyway. Secondly I want a set of tools to do certain jobs easily that no one else has to give me an edge in production. I don't want to give my best tools to everyone. Harsh but honest.

Thankfully although I am convinced that this short is cursed due to the sheer amount of things that have went wrong, sometimes nice things pop up that help in some way. So I was surprised today to have a message on twitter from one of the guys working for Pixar on the Renderman side offering a free 1 month license so I could get my short finished (or the 1st part anyway)..Hopefully once that's up and running I won't be reduced to the single core frame at a time crap that I am having to put up with from 3delights free version lol. As I can then use 2 machines and all my cores to render it should...in theory mean I can get the 1st part of my short done.

Of course there are still only 24 hours in the day so let's have a little list of what's left to do before EUE on the 30th June / 1st July.


The List of Doom:

  1. · all the complex high poly animated scenes need rendering out , inc passes
  2. · final soundtrack edit
  3. · final short edit
  4. · compositing
  5. · final colour grading
  6. · write lecture script for both lectures
  7. · do breakdown videos for EUE
  8. · Sort the PowerPoint presentations out.
  9. · Make beginning and end final titles for short
  10. · final encodes of short in various resolutions and colour levels (so I can keep it looking perfect no matter what the projectors colour balance).
  11. · Round up and arrange all files used
  12. · Do sculpt tests for 1st lecture
  13. · organise private course for a teaching gig just after EUE.
  14. · Have nervous breakdown

That's a list of the top of my head and I KNOW I have left stuff out. Once EUE 2011 is over with I will then have to try and finish of rendering the complex scenes that will be left for the 2nd part of my short before the Renderman license runs out. A big thanks to the guys at Pixar for the helping hand with the free Renderman for a month.

There can't be many times a blokes CGI short is bailed out by Pixar lol. If by some miracle I have time free I'll put an unreleased shot or two up from my short nearer the time.