Update 35: Investment


Greetings!

TLDR: The Kiri episode is rendering as I type, by the numbers it will be doing so for the next 4 days, allowing me a couple days to do my post processing and final touches, this should drop sometime next week (I'm aiming for Thursday) 

With that said, I wanted to talk today about where all your money ends up!

Believe it or not, I'm not actually buying a yacht with it!

This project represents the first true 'all hands on deck' situation for me, It's the first time that an entire production has started with, and been worked through by multiple members.

So along with paying the growing team of passionate individuals (all of which are equally as excited to see this vision rendered) a lot of it gets spent on hardware.

New computers, graphics cards, hard drives, that kind of thing, I'm going to explain why this is so helpful, by talking about cum!


Simulation for the nation


I spent about 2 weeks of this month running fluid simulations for this scene, When I said before that you should all expect a scene like Mercy X Aimee but messier, I meant it!

That sounds like a lot of time, and it is! But when you consider I was using a whole new system which I had to learn the ins and outs of... well.. it's still a lot of time! 

But honestly, that's just simulation in general, it is a tedious, yet necessary part of productions.

And when I say tedious, I mean tedious, below is an example of a FLIP simulation baking in real time:


Just so you can understand the numbers here:
  • The simulation is running on a 13900K, the fastest possible processor for this job
  • That gif represents the time it takes to see the result of a single frame of calculation
  • The simulation -will- slow down the longer it goes on, due to the amount of particles in the simulation increasing.
  • This is for 234 frames of simulation with an extremely small amount of fluid.

This is a system I've been using for many years now, a lot of the fluid dynamics in a lot of my productions were made using it, so I'm pretty good at it at this point, I've had a lot of practice. 

But even with that level of experience, a simulation will almost never turn out well on the first try, you almost always have something to tweak to get it to look right.

So suffice to say, considering some of the simulations in this scene go on for thousands of frames, this was not going to work.

I like to think I have an extremely patient supporter base, but I think even you guys would get upset if I told you all I'd be waiting another 2 months on this project! (I did not pull that number out of the air, it's literally how long it would have taken)

So, things had to change. Enter fluid particles. By comparison this is a much larger scale simulation running, with 10 times the particles, on 20x the scale:
Awesome right!? But this is simulation, Simulations still go wrong.

They go wrong a lot

So, you're always going to have to iterate, and at the length of time this sim goes on for, it was taking about an hour to an hour and a half to see the results of each sim.

The solution? run 3 iterations of the simulation with different values at the same time, across a networked set of computers:

You'll notice those move a lot slower, it got a bit more complex as time went on!

In the end, in a single week, I was able to complete something in the region of 150-200 iterations of the simulation until I found the settings that produced the result I wanted

Well, actually the result I wanted was Version 47, but I ran about another 20 after that just to see if it could be made any better.

Overall, the fluid particle system was a wonderful addition to my arsenal, and the time I've managed to turn this project around in is a reflection of the level of investment that's brought us this far

Without the support of you all, I would never have been able to afford all the devices that make this kind of stuff possible!

It is my hope that, if this continues, we can only continue to make better and greater things. As competent as this system is, I have made contacts that are going to enable me to start using Houdini for my simulation needs, and then we're going to see stuff on another level entirely!

With that said though, this production really is on another level from my previous works, we are finally starting to see the return of the investment that I've been building on for the past few years!

Light her up!


Another decision which, may not have been as popular is, This production could have actually been done and released by now, the timing was such that it did have an, albeit slim, possibility of coming in just under the wire.

What ended up costing us was the addition of volumetric lighting, It is a very good looking, but also very costly addition to the rendering process.

Just to put it in perspective, these are frames from different stages of the animation, all of them are to be considered WIP (even the one at the top of this post)

This was the first lighting setup I was going to go with for this production, at this stage this would render in about a minute per frame (17,175 frames, 3 computers = about 4 days)

I got some feedback that the light direction was a little boring, so I twisted it a little, messed with the range and added some cut-out sections of it, at this stage it's still taking about a minute per frame.
Then volumetrics came about, speak to any 3d artist and they will know that volume is an extremely costly, but also extremely beneficial addition to how real a scene feels. Just this level added 30 seconds of render time -minimum- to each frame (that makes the render take 6 days instead)
Nice right? But we can do better here at MJP.

This final adjustment, along with some additional tweaks pushed the final average frame render time to 2 minutes per frame (8 days rendering total)

Soooo yeah, that's why I missed my deadline at the end of last month, I think the results speak for themselves though.

But I said 4 days at the top? how come it's going to be done that soon if it was going to be rendering twice as long?

Well, I bought another GPU, another thing I was able to do, because I have so many of you supporting us. 

If I'm being honest though, I made a lot of similar decisions during the production of this piece. 

The whole, get content out fast VS realising your own vision of the scene debate is one that will run through the minds of any independent artist living upon the whim of thousands of generous strangers.

It is taking a measured approach, and spending the extra time that resulted in the production looking the way it does now, and I sincerely believe that, that's why you're all here, to see the finest quality pornography that exists.

Until I release my next piece of course!

I leave you with an image of some frames in motion from the animation.

Please bear in mind this is a screen recording/crop of a small section of the frame before any final post processing, it is also a GIF which are low frame rates, the final production moves as smoothly as any of my others do:

Bye for now!