How much does it cost to hire a 3D animator for CGI work?

[fishquail]

I need to hire a 3D Studio Max animator to animate and render a realistic looking model of a car into a flying car for a commercial. -It shoudl look Just like what you would see in back to the future.

I need this person to take the model (which was already pre-built) of a futurisitc looking concept car and make it fly in and land. It needs to look pretty realistic altho if its only 95% realistic is fine. The total on screen time would be about 10 seconds. Most of that its just parked not moving.

Now I dont know anything about 3D CGI so I have some questions;

1) How hard is this to do?
2) How long would it take to do?
3) How much would it cost to hire someone to do it?

Thanks!

[surf]

does the camera move, or turn?
it is not so hard I think 1-2 days maximum.

[Rich Lee]

1.not terribly hard
2.more than 1-2 days
3. like everything else, it totaly depends. you could get some kids just graduating from some CGI schools to do it for hardly anything/nothin....or you could get a pro cg guy to do it for like 2500-5k a week...(divide that up by 5 and you will see the day rate)...could be less....could be a touch more.

i suggest you go to cgchannel.com or maybe cgtalk.com and post the job and see what you get.

[surf]

well it also depends on motion tracking and animating the virtual camera.

[Greggl]

I find its just as easy to approach this from the other direction as well... based on
what was said above, you can figure out what your budget can support and go
from there... I'd be in the $5000-6000 range right from the get go.

Turnaround would be 1-2 weeks.

[Sony FX1 User]

Ask ilm www.ilm.com

[Greggl]

Ask ilm www.ilm.com

Hehe... ILM shut down their commercial division like 6 years ago... so that wont
be of much use. :dankk2:

[Matt Grunau]

Unless you have someone helping out with lighting, texturing, and rendering (and which 3D Max engine is the best for this??) you can count on a one man job either looking bad with fast results, or looking decent with SLOOOOOWW results. Best bet would be to hire an effects house. Prolly save you a little money too, and definately gonna save time.

You are definately looking at more than a week. Hell, the render time alone at 30fps X 10 with all the Antialiasing, Reflections, Motion Blur, Radiosity, Refraction, and various image effects is gonna be massive. A kickass (home) system streamlined to the gills MIGHT be able to kick out a frame every 4-8 hours. Maybe.

[Greggl]

I donno about that.. all the freelancers I know, myself included, have renderfarms
at home.

The base model is already built. Rendering automobiles is now pretty much top
notch for all the renderers and becomes even easier if its not in a hard cyc, limbo
room set.

Wildcards are going to be source footage (if any), moving camera and the quality
of the source model.

[Matt Grunau]

I donno about that.. all the freelancers I know, myself included, have renderfarms
at home.

The base model is already built. Rendering automobiles is now pretty much top
notch for all the renderers and becomes even easier if its not in a hard cyc, limbo
room set.

Wildcards are going to be source footage (if any), moving camera and the quality
of the source model.


Good point.

What kind of systems do you have in your farm, and which app are you running? How long would 1 frame take with your setup, with all the extras on it?

As a lightwave user, when I think of rendertimes for photorealism, I get physically sick. :Drogar-BigGrin(DBG)

[fishquail]

Im talking about Still camera, No motion.
Shot List:

1) Overhead angle Wide shot: Car Flies in and lands on pavement. 2-3seconds.
2) Front Wide shot: Car just sits there while actor speaks - 3 seconds.
3) CU: Driver seen inside of CGI car speaks - 3 seconds.

[dougspice]

And you have someone to do the compositing already? Sounds like you're talking about more compositing than 3d work, really.

[Greggl]

Good point.
What kind of systems do you have in your farm, and which app are you running? How long would 1 frame take with your setup, with all the extras on it?


I'm running a mix of machines. Dualcore G5, 4 older 1.3 Athlons, 3 SGI
Octanes and an O2. Maya for 3d, my compositing is Shake and Avid's
Media Illusion which was their wanna-be Flame compositing package.

In general, I'm shocked if renders get to more than 30 mins a frame... I
avoid using '3rd party' renders for my freelance work. I do much more of my
work in the compositing stage, rather than rely on the 'expensive' tricks that
we're getting in the new renderers.

[surf]

a frame every 4-8 hours

this sound shocking. when this happens, do not continue the rendering (10sec=250 frames 250*8/24=83,3 days :)) the quality is set too too high.
Mostly the final format is 720*576 and if the car do not fill completly the frame you will not need to render 720-576 (if the car is a lot smaller you will be able to render it at half resolution, and at compositing you can scale it)
and you have to render only a car in the 3d software...

[Beat Takeshi]

But you still have to match the lighting of the original plate and this can take a day or 2 in itself. If its an over cast day you have to make sure all your shadows match in softness and most likely you would be using a bunch or area lights which render slow. It wouldnt only be the car rendering either. You would have to do car render, shadow render, matt render, specular highlight render, and maybe atmosphere if its a hazy looking shot. HDRI would be the closest for matching the lighting but someone would have to create the image maps on set of the background plate shoot.

[oneinfiniteloop]

I'm running a mix of machines. Dualcore G5, 4 older 1.3 Athlons, 3 SGI
Octanes and an O2. Maya for 3d, my compositing is Shake and Avid's
Media Illusion which was their wanna-be Flame compositing package.

In general, I'm shocked if renders get to more than 30 mins a frame... I
avoid using '3rd party' renders for my freelance work. I do much more of my
work in the compositing stage, rather than rely on the 'expensive' tricks that
we're getting in the new renderers.

Greg, how do you handle passing renders off to the Athlons from your Mac? Are you running Linux on the Athlons and using mental ray and shakes command line render on them?

I'm trying to plan to do the same thing. I use a dual G5 for my workstation, but I would like to build a small render farm with some opterons for rendering. Any suggestions?

[Matt Grunau]

But you still have to match the lighting of the original plate and this can take a day or 2 in itself. If its an over cast day you have to make sure all your shadows match in softness and most likely you would be using a bunch or area lights which render slow. It wouldnt only be the car rendering either. You would have to do car render, shadow render, matt render, specular highlight render, and maybe atmosphere if its a hazy looking shot. HDRI would be the closest for matching the lighting but someone would have to create the image maps on set of the background plate shoot.



That's pretty much what I was trying to say, but I didn't put it as well as you did.

[surf]

But you still have to match the lighting of the original plate and this can take a day or 2 in itself. If its an over cast day you have to make sure all your shadows match in softness and most likely you would be using a bunch or area lights which render slow. It wouldnt only be the car rendering either. You would have to do car render, shadow render, matt render, specular highlight render, and maybe atmosphere if its a hazy looking shot. HDRI would be the closest for matching the lighting but someone would have to create the image maps on set of the background plate shoot.

ye HDRI/IBL seems to be a good and very fast way (not for rendering, but for you)
and when I told render only the car--> do not have to render the background. so that is transparent-->a bit faster. of course you will need shadows etc. bit if you set the properties well it should not take so long.
anyway I give a maximum if 5 mins/frame to render. but usually takes 1-3mins.

[Greggl]

Greg, how do you handle passing renders off to the Athlons from your Mac? Are you running Linux on the Athlons and using mental ray and shakes command line render on them?

I only use the old athlon farm for maya and I use the built in renderer. I'm
really oldschool in that department. The overhead for mental ray and prman
are unnecessary for the vfx work that I do. They get much more use dual
booted as windows multiplayer game boxes than as a renderfarm :)

I also tend to rely on the free shaders that mimic the 'popular' looks like
GI, radiosity, etc. Hehe... there is a drastic difference in what will pass
as 'cool' in a 1K online still over a white BG and what looks 'cool' at 24fps
with atmostphere and lens distortions, dust, motion blur and matching
film grain or digital artifacts. Live action is all I do and its MUCH more
forgiving.

I launch with http://www.drqueue.org/ and Maya 5. Disney ran LSF
to manage their renders and drqueue is pretty similar from a user
standpoint.

The irix boxes run illusion and the standalone renderer as well as maya.
They are also set up as a shake 3 farm. The mac is running shake 4,
but its not set up to network render. Shake 4/MAC is what I've been
using as a roto tool more than anything else. The bulk of the comping
is done on illusion with a homebuilt fibrechannel raid feeding frames.

I've been a LW/3DS user since the early 90s. At that point I was getting
30 min/frame renders. I kinda still aim for that, but what I can pack into
that 30mins/frame is VASTLY more complex than back then. As I said, I
get MUCH more tricky in the comp than I used to.

[oneinfiniteloop]

I only use the old athlon farm for maya and I use the built in renderer. I'm
really oldschool in that department. The overhead for mental ray and prman
are unnecessary for the vfx work that I do. They get much more use dual
booted as windows multiplayer game boxes than as a renderfarm :)

I also tend to rely on the free shaders that mimic the 'popular' looks like
GI, radiosity, etc. Hehe... there is a drastic difference in what will pass
as 'cool' in a 1K online still over a white BG and what looks 'cool' at 24fps
with atmostphere and lens distortions, dust, motion blur and matching
film grain or digital artifacts. Live action is all I do and its MUCH more
forgiving.

I launch with http://www.drqueue.org/ and Maya 5. Disney ran LSF
to manage their renders and drqueue is pretty similar from a user
standpoint.

The irix boxes run illusion and the standalone renderer as well as maya.
They are also set up as a shake 3 farm. The mac is running shake 4,
but its not set up to network render. Shake 4/MAC is what I've been
using as a roto tool more than anything else. The bulk of the comping
is done on illusion with a homebuilt fibrechannel raid feeding frames.

I've been a LW/3DS user since the early 90s. At that point I was getting
30 min/frame renders. I kinda still aim for that, but what I can pack into
that 30mins/frame is VASTLY more complex than back then. As I said, I
get MUCH more tricky in the comp than I used to.

Interesting, this gives me something to think about it. I haven't gotten into much heavy duty work, but I can see times where I'll need some extra CPU power. Thanks for the insight, I'll have to look into this some more.

:thumbsup:

[Greggl]

Interesting, this gives me something to think about it. I haven't gotten into much heavy duty work, but I can see times where I'll need some extra CPU power. Thanks for the insight, I'll have to look into this some more.


I wouldn't :)

I'd build the 'time to render' into your bids and use weekends and downtime
to render personal stuff.

As a 3D artist, you should train yourself to be a much higher producer of content
than your hardware can support. Your 'downtime' should be spent rendering the
backlog of projects that need to be output. If you think of 3D and VFX as a 'realtime'
process and train yourself to rely on the interactive nature of the modern tools,
you'll only really be able to grow with the increasing speed of hardware. The goal
is to learn the tricks and techniques to exploit the hardware, anticipate the output
and bascially work faster than any machine that anyone could put in front of you.

I've yet to meet a pro that has ever been satisfied with the speed of their
hardware :) We *always* push the limits, so you need to learn to work within
limits or you'll constantly be chasing that ghost of 'one more node in the renderfarm'
or that '400MHz' upgrade that'll quench that speed-lust :)

[oneinfiniteloop]

I wouldn't :)

I'd build the 'time to render' into your bids and use weekends and downtime
to render personal stuff.

As a 3D artist, you should train yourself to be a much higher producer of content
than your hardware can support. Your 'downtime' should be spent rendering the
backlog of projects that need to be output. If you think of 3D and VFX as a 'realtime'
process and train yourself to rely on the interactive nature of the modern tools,
you'll only really be able to grow with the increasing speed of hardware. The goal
is to learn the tricks and techniques to exploit the hardware, anticipate the output
and bascially work faster than any machine that anyone could put in front of you.

I've yet to meet a pro that has ever been satisfied with the speed of their
hardware :) We *always* push the limits, so you need to learn to work within
limits or you'll constantly be chasing that ghost of 'one more node in the renderfarm'
or that '400MHz' upgrade that'll quench that speed-lust :)

Good point, never really looked at it that way. I can see how this will help one stay grounded in the artistry/content vs. the endless pursuit of power.

[robo3687]

in response to saying it could pay 5k:

heck i'd do this job for that kind of money...lol.....doesn't sound that involved really.....the animation part wouldn't be too much of a problem, and if the shots are locked off...camera's shouldn't be an issue either....

you say the model is pre-built, what about textures? would they need to be made?

i've had about 5 years experience in max now and am currently heading into doing alot of cg integration in a feature i'm working on....

[im.thatoneguy]

I personally charge $45 an hour for freelance 3d Max work, I would expect a good professional artist to be somewhere between $30 and $80 an hour for freelance work. But if the project is interesting and looks like it is going somewhere, you could probaby find someone with some down time to do it for free.

I would estimate if the model is already textured (to your satisfaction) and there was no actual modelling to be done that this would take about 15-20 hours. Plus probably 5 minutes per frame for 300 frames which is about 25 hours of render time, so a 6 node render farm would take about 5-8 hours to render. I only charge 1.5 hours for a day of rendering if it doesn't interfere with anything else, so I would expect to pay whoever did it about that. Final cost for those 3 shots if you had the model with materials ready to animate and camera match about: $1000 all said and done. It would take 3 days of dedicated work.

It would have been best if you had taken an HDRI light probe while you were there, in the future it would help out whoever ends up having to do the lighting match. you can buy a Garden Gazing Globe for about $15 bucks online, they aren't a painful expense, and save you a lot more money than that.

If the model needed a lot of work, or there was no material yet, it would probably spiral up to about 60-100 hours of work depending on how much needed to be done.

If compositing also needed to be done, this shot would probably take about 4 hours to comp all 3.

Apply previous rates, and this thing could get kind of expensive pretty quick.


In response to Greggl:

Not only that but sometimes you find out you overestimate just what your render farm can do. ;)

I was on a project this year where I had to have the deliverable in their hands by a certain time the next day. By the time that shot came up in my render queue I did some quick math and figured out it was going to take 2 hours past my deadline to finish rendering... So I had to spend the next hour trying to save 3 hours (2 hours plus the hour optimizing). It's a real slippery slope spending time to save time... you never know if it's going to pay off. Moral of the story people: if you add anything to the scene which might slow down your render times, do a test render and make sure you can still meet your deadlines. :)