Jade signature sky villa animations

[Macker]

Sounds fantastic. How as a team do you tackle these shots then? How many people work on something like this, and what's the overall timescale?

In my office there's one other 3D chap, but we never work on projects together. We each see through projects from conception to completion; but it's an architects office not a vis-house so we never ever get sign off on anything. Everything can change at the drop of a hat.

[Neilg]

It was 2 people to start with and a third joined on once we were getting the scenes detailed up. 4 weeks getting the storyboard/concept locked down (not really full time, some waiting in between meetings), 4 weeks of 3d production, then another 2 weeks of rendering and postwork/fine tuning the scenes we still hadn't started rendering.

[AlexP]

We started off with a storyboard of still images which got us the concept approved.
once that was approved i did an internal test of how the timing would work using the sheets from the pdf.
https://vimeo.com/135981534
pass: jadesig


After that we started animating cameras and did a few revisions of the grey storyboard
north - https://vimeo.com/138797130

central - https://vimeo.com/137072761
pass: jadesig

One they approved and signed off on that, we went back to still images - one frame from each camera. No camera was rendered until that single frame (and any others relating to areas we can see in it) was signed off in writing. Kept the whole process very smooth.
By and large i'm very happy with how close we stuck to the original idea. some cameras got moved but our timing and pace was the first thing to get locked down which made everything else super easy. Not one frame over the exact number we needed was rendered.

Thanks for posting, interesting to see the process followed.

Mistake that we've made in the past is rendering "low res" versions of the animation with full mats and lighting (due to client demand) Such a waste of time.

Do you always render the glass separately? I tried it on the last animation but must admit I couldn't decide whether the saving in render time was worth the extra comping time.

[rivoli]

beautiful work, and it's really fascinating to have some insight on the process. I'd love to see more of this, especially on quality work such as this one.

[jose-duenas]

Really stunning work. Congratulations.

So you say that each frame with all its glass layers were processed in 1 hour... in slow renderfarm machines?

Could you please share your render settings? are you using the current BF+LC method proposed with Vray3.3? At which resolution did you render the frames? Besides setting Reflection Maxdepth to a lower value for most materials... how do you get that great quality at that rendertime?

Congratulations again... beautiful videos.

[Neilg]

beautiful work, and it's really fascinating to have some insight on the process. I'd love to see more of this, especially on quality work such as this one.

I always like reading it when others do it so I figure it was only right to open up a bit

Jose - it was a few different methods, We did a base GI pass with all static lights using a saved irradiance map and light cache, then did a separate render which was just the moving sun (although it's hard to tell, it never stops moving from sunrise-sunset) using BF/LC.
We used this method because using BF/LC for everything and doing it in a single pass the ETA per frame was 6+ hours.
This cut the base IR down to about 30 mins and the sun pass with BF down to 15 or so. Some went higher, some stayed low - by completion they averaged an hour. We knew which ones were fast and which would take longer so we did the quick ones first to find out how much time we could push ones like the central kitchen shot up (went up to 1h 40 all in, the longest)
Resolution was 1920x810


Alex - dont always, but on this one we decided very early on that we would test this. We didn't have any other jobs on so we could spend the extra manpower, and in doing so decided to try and get the render times as low as possible with as little waste as possible.
It was an interesting test of our limits, but it was a lot of manpower. Cheaper than buying another 5-10 machines though, by a long way. Would've been nice to have for sure, but we didn't need them - I'd prefer to get better ones when we do need them.
Even with 3x the render farm though we still couldnt have delivered on time by rendering everything in a single pass. To use brute force and deliver on the same schedule we'd have needed an extra 25 machines. That's a phenomenal expense. The 2 weeks while it was rendering we worked optimising and breaking the other scenes into passes, we didn't need to have those 2 weeks free to work on something else. It's not a lot of gain.

[jose-duenas]

Thanks for sharing Neil.

Really appreciate it. Always hungry of learning.

Pretty interesting how you did it. I was curious to see how BF/LC would work at that level of production vrs deadline situation.

How would you handle moving trees , or other moving geometry with a similar method? would you still go with IM + LC and different BF layers with moving geometry?

[Neilg]

We would do 5 frame isolated tests early on before committing to anything

moving geometry and moving light at the same time would definitely need some figuring out.

[jose-duenas]

I understand
Thanks for your tips again.

[lukx]

Stunning stuff! Especially from images before and after grading. How did you do Clouds? Is it some stock vid or phoenix ?

[Neilg]

We shot that ourselves as a bracketed timelapse from the building next door. It was kind of funny - the first time we went, we had access to one building & went up on the roof for sunrise. your usual roof, walls around it but a tripod was high enough to see over. Safe.
We went back to get the sunset and found out that they didn't want to let us back up - so our client called the building next door and got us up on that roof for the next morning. We get up there and the entire thing is enclosed with a tall structure - presumably to guard the air con system from wind - except for one gap which could only be reached by laying on my stomach and pulling myself under some steel frame to get onto this 2ft deep ledge, and set up 2 tripods and cameras. We were doing this before sunrise too, in the absolute pitch black.
This is standing on the ledge, looking straight down:


Couldnt open the tripods properly but they were stable.

[lukx]

Neil, I tried Resolve but I have problems when using .exr files. Did you guys used linear workflow, because when I import .exr file into Resolve it's dark.

[Neilg]

We've not used it for any big job, but when we tested it we didnt open exr's straight in it - we were opening the 16bit output from fusion into it.
Always have to apply gamma 2.2 to exr's in fusion though.

[3LP]

Isn't that because your save output gamma in max is 2.2 in stead of 1?

Stan

[Neilg]

It's worth noting what opens fine in photoshop doesnt open right in fusion - and if you make it open right in fusion it doesnt open right in PS. if it doesnt look right one way or the other, do a .454 or 2.2 so that it does.