Convince me
Someone convince me why I should continue with machinima. This isn't another "why machinima sucks" rant, but just a slight doubt I've been having lately.
As my skills improve in filmmaking, naturally my ambitions and goals increase with it. The kind of movies and the level of quality I wanted to achieve back in 2004 compared to today, is like night and day. Back then, simple machinima was good enough as long as it was captured well, edited cleverly, and was overall watchable.
From after the release of the first OTSS to right now, what I wanted for OTSS 2 kept growing day by day. At first I knew I had to step the quality up, which meant addressing the main issues with OTSS: emotionless and expressionless faces, looping stock animations, stiff "gamey" motions and graphics, and more importantly, the plot.
All that technical work would go to waste if the plot didn't live up to expectations. In my opinion a great film has all the elements in high quality, including (but not only) the story. Rewriting, rewriting, rewriting, etc., I won't get into it. I'm sure a lot of you have heard enough about OTSS 2 rewriting over the last few years.
With all that work and dedication into writing a short film, a film to be created in a game engine, not to be sold and not legally owned, seems somewhat ludicrous. Add to that, all of the custom assets that need to be created: character models, animations, textures, maps, props, weapons, and quality as good as or better than the original game the machinima film is made in.
It all seems way too daunting for the limits of machinima. Visually, I want to achieve a higher quality than the cinematics found in Gears of War. And the more I think about it, the more I realize that it won't be possible without a lot of money.
At that point you're taking away the main positive aspect of machinima: money. Time-saving is simply not a factor in custom content films. And when you're spending that much time and effort, why not spend a little more time to render out a full-blown CGI picture instead? If we were on a contract deadline, things would be different, but this is a personal project, one that could be potentially sold after it was completed.
But with machinima that isn't even possible, sometimes it is but still in limited form.
And the most disturbing thing that I've come to realize is this: video game engines become outdated. Fast. When one is obsessed with quality and technology such as myself, one must always strive to keep up with the times and choose the latest game engine on the market.
And everytime a new, stunning game engine is revealed that will "change the way machinima is made", this means you have to take the time to learn this new engine, the way its coded, its scripting language, the way its map editor works, the way lipsyncing works, the way the cameras work, what kind of workarounds or hacks there are, and so on.
In CG every new technology is an evolution of an existing tool, such as Maya or Max. Once you've learned how to use it, you don't need to re-learn it every few years!
So where exactly does time-saving come in, compared to a full CG render, if you're spending all that time learning different game engines?
I'm doubtful because over the course of many years, I would much rather spend that time mastering one tool, than trying to learn many.
17 Comments:
If cutting-edge visuals are, as you said, of obsessively high priority for you, then machinima is not your ideal medium, and never was.
I think you would find, however, that going to CGI will not amount to, as you stated, "just a little more time." Even factoring in the time taken to learn a new engine / toolkit.
Let's say OTSS2 ends up being exactly the same length as OTSS... 11 minutes, 28 seconds; and let's say you only went with 24 fps. That's 16,512 frames. And let's say you somehow manage to get an amazing average final rendering time of, oh I dunno, 2 hours per frame. That's 1,376 days of 24-hour-a-day non-stop rendering for just your final output. Check my math, am I computing that right? 11:28 = 688 seconds x 24 fps = 16,512 frames x 2 hours per frame = 33,024 hours / 24 hours = 1,376 days.
Can you imagine how many different engine environments you could master in 500 evenings? (Let's assume you take the other 876 evenings and spend them doing something else).
Let's say you got a server farm of HAL 9000s, and were able to render at an average of an unheard of fifteen minutes per frame. Rendering would still take just shy of 180 days, non-stop.
And that's JUST for the final render; no doubt there would be multiple trial renders along the way, a scene here and there, revision, render again...
Mind you, Pixar had a massive server farm for Cars, and it still took them days to render one second of final footage. Up to SEVENTEEN hours for a single frame.
I think you are sorely underestimating the advantage of real-time rendering. It's not just a small, convenient time improvement; it can literally be years of advantage.
And I would think, given your new circumstance which affords you less free time than you had previously, that time issue would be an enormous one for you now, maybe more than ever before.
I also think, if you're going to continue doing machinima, you need to give yourself a cutoff point after which you will no longer hop engines for an in-progress work, and just make the film. Lest you go the way of Duke4Ever, hopping eternally and never having anything to show for it. Perfectionism is admirable, but so is productivity. You know what I mean?
I know that feeling exactly. When I first saw Serious Sam engine, I was in love. Month of learning. Then MaxPayne, jesus... month of learning. And when I bought HalfLife, I was sure quest is over. Then, after month of ...you know... and two movies, and after I realized
that I don't own my movie I decided to go away from big games engines and never turn back.
Not matter how good engine comes out.
There is very decent gray area between horribly slow rendered CGI and big games engines Machinima. For example, MotionBuilder (path I took, and after month of learning and adapting my movies ideas to the new tool - very important step - I am learning every day that decision to left commercial engines was one of the best I ever did). Or, iClone - for couple of month this will be excellent tool. And far more important, all learning time invested in this tools are not wasted.
There is couple more free (in term of content and copyrights) engines and programs out there and there will certainly be more in future. After all, finding perfect tool for Machinima is not about choosing engine, it is about establishing exact and best work flow for your ideas.
You now that better then me.
Don't leave Machinima. Real time is future.
Been there, done that. Strange Company has vacillated backward and forward between technologies over the years. Whilst it's not well known, we've actually done a few projects, mostly corporates, using conventional pre-rendered 3D.
And I'm here to tell you - don't go there. Use whatever tools you like, use Softimage, use Filmbox, but stick with realtime, for pity's sake. As soon as you go non-realtime, it all gets much, much slower. I'm not going to throw figures at you - I'm just going to say that in my experience, making a film using conventional 3D animation is a lot, lot slower than even a high-end Machinima production.
The same applies to aiming to use hand-generated animation. Bear in mind that per character, a good 3D animator will turn out between 10 and 30 seconds of animation per week. This shit is SLOOOW.
I remember way back when, when Joe Goss was facing the same decision as you as to whether to go pre-rendered. He did. His project was never finished.
It's really a bit frightening how slow this stuff gets. It's not about the rendering, really - it's about all the other slowdowns that a conventional 3D pipeline introduces.
Machinima's a bastard. I use it because the alternatives are currently worse. Having said that, there are some creative ways to make Machinima - like I say, it's possible to turn Softimage into a Machinima package.
And Blender, too, for that matter, now bundled with a game engine.
You've been at Shiny for a bit now so your very much aware of both sides of the coin.
Your title is 'convince me' but you know the minute you see a release of a kickass machinima film, done by someone locked in a room somewhere night after night, fighting technical issues, battling import/export troubles, scouring the web for answers to obscure error messages, inching away until release, then wow. fuck. damn.
Your a machinima champ. Is you legacy behind you or still ahead?
Ken
Thanks for your comments guys, I really appreciate it.
You make a very valid point, Overman. I suppose I didn't think through the numbers as thoroughly as you laid out.
You are correct in that machinima was never originally my ideal medium. Back when I first started my goal was live action, and machinima was just a means to an end.
But after I finished OTSS and discovered machinima.com, that's when I began having hopes for and supporting machinima as we all do now.
As I dwell more and more with professionals in the game and film industries, I begin to feel like my sweat and tears go unnoticed. No matter how great a machinima film is, in the eyes of these people it is always seen as "lesser" work. I've never agreed with this mentality but over the years it has become quite discouraging. That and I grow tired of trying to explain what machinima is to others, only to be rewarded with a blank stare.
I have gotten the Duke Nukem analogy for OTSS 2 before, haha. And it really is starting to feel like it at this point.
I fully realize what you mean by the cutoff point. I struggle with that very often: when is enough, enough?
But it's somewhat ironic that you mention it, because the reason that I decided to switch the engine to Unreal 3 was precisely for the reasons I stated in the post - that I wanted one tool to master instead of learning a completely new one every project.
I'm set on learning the new Unreal engine as much as I can, but beyond that I don't think I will ever have the energy to learn another engine. From what I've seen so far, the Unreal engine should be fairly future-proof which is why I've decided to bet on it.
Tom, you are very right about the workflow. To be honest it's something that I haven't been able to get down. I can't blame it on machinima itself, as it's just my own inconsistency. I tend to either do everything at once or nothing at all. Still, by finding a consistent tool to use I hope I can change that more easily.
Hugh, on one hand I agree with everything you said, but sometimes it just doesn't fit. I've been watching a lot of CGI shorts lately and the more I watch, the more envious I become of them. A lot of these are being done by one-man-crews just like machinimators. I realize the amount of work is exponential, but it also feels like the results are exponential as well. (not always of course).
Ken, surprisingly working here has made me less enthusiastic about machinima overall. Maybe I am just getting sick and tired of having to deal with engine related technical issues.
I'm stuck in a weird place. Making in-game asset machinima (like Edge of Remorse) now seems too derivative and somewhat unoriginal. But making all custom content machinima seems too complicated and too much work. You might as well create a whole new game with all those assets...
I am prepared to see OTSS 2 to completion with all custom assets, but beyond that I just don't know anymore.
Maybe it's more than machinima itself, too. I may just have been burned by a certain false hope in the past. Blah.
If Nomad had feared the future, there would be no Bloodspell. If he had waited for a better engine, he'd be in D4Ever mode. SC hunkered down and committed and made a good story come to life using a crappy engine; they made the best of what they had, and the result is far from shameful. Right?
Mark calls these kinds of decision points in life the "save game" syndrome...
You have great stories to tell. That's what you should focus on. Sure, you have to choose the tool best suited to your story, but there are all those other factors (time, money, effort, passion) as well.
I can assure you, having been there, live-action is also not an easy thing, and it *does not* get the respect it deserves either, except in *rare* circumstances. If you want to make one I'm here for you. Crap in live action is GLARING. Crap in machinima is less so. Live action by amateurs with a volunteer crew of 12 and little to no budget, free equipment from a school, etc., gets you about 10 shots a day. About 0.5 - 1 minutes of usable footage. For an action film like OTSS, you'd be lucky to get 1 minute of footage a day. And I'm talking 18 hour days.
Let us focus on making machinima a respected art form. It will take some time, but we're already climbing that mountain, and you are a major contributor to that ascent in many ways.
If you stand before the mountain you may fear the climb; if you fear the climb you will not start; if you do not start, you will not climb the mountain.
Case in point,
Ingrid
I guess I have already started to climb the mountain.
But I'm halfway up, and see a bigger, more majestic mountain next to me, and I wonder why I didn't start climbing that one.
Hm.
Don't make me slap you.
:D
This post has been removed by the author.
I disagree with Overmans Figures on the render times, sure rendering a complex scene with all the extras (HDRI, GI, DOF, Billion + polygons) all stacks up. BUT in real world productions nobody actuallys renders full length films in this manner. You have to understand the difference between rendering for a STIL shot, and rendering for animation. For a still shot that is going to be over 12 MegaPixel and in 4K resolution, sure calculting Radiosity with HDRI and DOF and million plus polygon models make sense, after all you want photoreal scene. But for animations is as you said is not feasable, (note I know some still renders that have taken over 48Hrs).
Firstly for a full blown CGI film you will typically model ONLY what the camers sees, secondly you will model only as acurately as needed (eg things far away can be simple box models with textures, sometimes just plain textures), thirdly ALWAYS things are rendered in passes, this reduces render times massively, and many things can be done in post MUCH faster than in render, EG DOF and motion blur (for example, a Z-DEPTH map is imported from into an Comp application like AE or Combustion, and DOF is applied there). Also lights shadows, all can be changed when you have seperate passes.
Also there are many other tricks such as Fake Radiosity, texture baking etc. There are many other tricks like these as well that help to speed render times massively. Also go watch toy story 1 (I was shocked at how low modelling was done on this, recently when I watched I was mentally thinking how it was constructed, some textures was really simple, a lot of the scene was fairly simple, no complex render effects used.) lets not forget how much computer power has come along in that time.
Nowadays a Boxx worksation with 16 Cores is common.
And home "Mini" render farms are easy to set up, most people now have many old PC's and laptops around, even if you had access to three PC's this mini "Render-Farm" will reduce your render time by a factor of three, the more PC's you add the more you divide your time.
Or you can just send the file to companies that have ready made render farms, that for a fee will render your animation for you.
In the end its all about control, I myself am struggling with Full CGI vs Machinima (I'm from a Full CGI background) the real-time aspect of Machinima appeals to me, but the lack of full CGI CONTROL and ultimatley control and Quality makes it less appealing.
So the choice is: Ultimate Quality (long render, slower time) Vs Real-time Production (low quality, limited control)
Anhar Hussain Miah
Thanks for your input, anhar.
Everything you say is very true as well. No one would do a full render EVERY time they make a small change - it would be at a lesser resolution, with other effects turned off or hidden, etc.
The skill of CG rendering is to decide WHAT to render and WHEN, I think, instead of simply having the fastest computer(s).
I have experienced some of this while creating Edge of Remorse. It was largely created in After Effects with many, many layered effects and compositions in every single shot. In order to do all that in 720p, in one month, I needed to figure out a way to render out the shots in an efficient manner.
I can only imagine that CG work is a lot more complex than 2D compositing, but I think the thought process of efficient rendering would be exactly the same.
99% of machinimators don't deal with rendering at all, so I can see how this would be hard to understand to some.
"hard to understand to some"?
Wow, throw me under the bus, why don't ya Jas?
"The average time to render a single frame of film for 'CARS' was 17 hours. Even with a sophisticated network of 3000 computers, and state-of-the-art lightning fast processors that operate up to four times faster than they did on 'The Incredibles,' it still took many days to render a single second of finished film."
- Source:
http://www.rottentomatoes.com/m/cars/about.php
I'm operating under the assumption that Pixar knows something techniques to optimize their render. I'm also assuming that they are part of the "REAL WORLD" way of doing things.
I also assume Jason is reaching for top quality, but I gave him the benefit of the doubt that he wouldn't reach too high: I assumed he would seek a less-than-Pixar level of quality which would be achievable in a little over one TENTH of the time: Two hours per frame. I would think this would be quite a feat considering Jason is likely going to be performing his render with approximately 2,995 less computers than Pixar used. If this is unreasonable, demonstrate for me how it is unreasonable with specifics.
Then I went even lower, to the 15-minute scenario... which somewhere in the neighborhood of 1% of render time consumed by Pixar with their 3000 computers. Again, no small feat. (If he had 1% of the computers, to make resources drop proportionally to duration, he'd have 30 PCs engaged in the task.) If this is unreasonable, please demonstrate with specifics.
I'm not claiming to be a rendering genius, but please don't make me out to be an idiot either. I know how to use a calculator, and made what I thought were some reasonable assumptions about how much quality Jason would shoot for. If I did either of these things incorrectly, then please show me how to calculate it correctly, with specifics.
THEN condescend me, and I won't grumble one bit. Promise.
Sorry if I sounded like I was "throwing you under the bus", Phil. I think that last sentence I wrote did not really come out in a positive way. The only thing I really wanted to say was that all this talk about rendering doesn't really apply to the average machinimator. I know you did some form of rendering with those DoF effects (at least) in 'Only', so it wasn't directed at you if that was the impression you had.
The problem is that I am just totally stuck in the middle and don't know which is right or wrong anymore.
Your points and calculations make perfect sense. You've pretty much convinced me big time, but then I watch a CG short done by one man that looks better than any machinima I've ever seen, that didn't take years and years to make.
Then I start to think, if he can do that, then there must be some effective tricks and efficient methods to do all those crazy renders without spending a year on each frame and without having a 5000 computer render farm.
Then again, maybe he did spend a ridiculous sum of money to achieve that. I keep going back and forth like that until I give up and go do something else. Man, I am so lost!
I overreacted, I am sorry.
Maybe the speed at which processors are moving forward is so great that soon render time won't be a big issue any more. I do hope so. I look at something like "Elephant's Dream" and salivate, big time. In spite of my arguments, I crave that ultra-quality too.
Maybe someone will come up with some kind of cross-internet workload sharing technology. Kind of a cross between P2P and shared CPU cycles... where a few hundred people could be connected to a virtual network which could collectively donate CPU power to a big render task.
I dunno, I'm talking way above my pay grade now. :) But it's interesting to think about. If that kind of sharing is possible over a LAN, could it be over the Net as well?
10-15 minutes per frame is pretty achievable with simple scenes (a tad above game style complexity).
There are also things like Nvidia's Gelato, where you can use GPU units in parallel to achieve very good looking, fast, non real time renders, all without the need to set up an actual farm.
With traditional 3d the rendering engines aren't really optimized to give you the 30fps+ of games, but you do have a fairly decent amount of control with the trade off between photorealism and speed. As was said before there are many tricks you can use to make things look ok to the casual observer, but not actually be accurate. I think pixar might have such large render times because they can't get away with such things on the big screen - and they have the budget to throw at it so why not?
CGI done with maya/max/c4d/etc. vs. machinima is really just a trade off in control. It doesn't have to be all one or the other either. With match moving and compositing you can bring in elements from both at will. There are so many options and no real reason to say "this is what I'm only going to do." Continue to add skills to your palette and don't worry too much about it I'd say.
- Elf
Honestly, I love machinima as much as anyone who plays about 3 hours of video games a day, but I agree with you on some parts. Although it may be easier (in a roundabout way) to make a film in a game engine, there are so many limitations and changes that pop up out of nowhere it's almost impossible to keep the viewers happy. I mean, how many time have you changed the game engine for OTSS 2? First Max Payne 2, then Half-Life 2, then Unreal or something like that? It's almost impossible to keep up.
Now, Jason, I can't say I blame you, it's kind of your style to change things up and show people the unexpected, but you're eventually going to have to stop changing your mind and get to work!
If you do decide to change your mind again and make it a CGI, then I suggest using Max. It's much simpler in my opinion than Maya, and if you need animators or texturers I'm up for it. Just let me know, dmstelevision@gmail.com or dmsN64hawk on AIM. You've already got my contact stuff, but just refreshing.
Post a Comment
<< Home