Jack Kilyan - Digital Matte Painting Projection Project

#1

Hi everyone,

my name’s Jack and I’m currently a final year student at Staffordshire University here in the UK, studying VFX and Concept Design. I’m just about to embark on my final project which I can choose to do on any topic I want, and after spending the last two years learning about different methods and paths within our industry I’ve decided to aim towards a career in matte painting and environment art.

My project is set to run for the next twelve week’s and I thought it would be a great chance for me to make a thread here so I have a condensed place to post, and also post (hopefully) on a weekly basis with updates on the project and get any advice I can from this great community!

My inital idea:
To create a matte painting projection CG shot to the highest standard possible.

Inspirations:
I want to create something along the lines of this, an amazing shot my Levi Peterffy.
http://vimeo.com/80501593

I’m probably going to tackle it a bit differently than he did, and I want to use world machine for my base, I’ve learn’t a lot from the matte department tutorial!

Also seeing as this is my first REAL attempt at this full workflow I know I’ll hit a ton of roadblocks along the way and I really hope I can gain some knowledge from this community and achieve something even near a quality standard.

Anyway that’s my little introduction, I hope this post is ok being in this thread and I’ve not accidentally posted where I shouldn’t! Thanks for reading everyone, I’m looking forward to the next twelve week’s :smiley:

#2

Welcome to the forum, Jack. You’re in the correct place. As for your final project, I would plan it a bit more before you jump right into it, since the main goal is to obtain a job in the end. You have the benefit to being able to prepare and cater to the companies where you wish to apply at. It won’t hurt asking them what type of work they wish to see in potential portfolios. For some fantasy scenes might be perfect, other studios might prefer real life shots with more complicated perspective challenges, matching a photo-realistic look etc.
Another thought would be that if you still decided to aim for something like the painting from Levi’s tutorial, I do recommend you stick with his techniques because he spent the time documenting an entire step by step around it. A number of supervisors also like to see a before and after comparison rather than you creating something from a blank canvas. Just a few thoughts…

#3

Hey thanks a bunch for your reply. :slight_smile: My issue with following EXACTLY the layout which Levi did was that I couldn’t seem to find much information on how he created the matte. Like from my novice understanding when watching that clip I posted, he UV’d the mountains and created the painting straight from Nuke, comping in the different projection elements? I do agree with your idea of planning myself more, I will definitely try and talk to some companies. My teachers have spoke to me today and tasked me to kinda do a “proof of concept” and just take an existing image, break it apart and animate the camera and then relight the image as a start.

I will try to find more information on Levi’s techniques and see if he has released the full tutorial? Thanks again for the help :slight_smile:

#4

Hio Jack,

Welcome to the forum, and as Milan said, this is a great place to have your own thread about this :slight_smile: Levi didn’t finish a full tutorial on this yet, just a matte painting break down of something he would do along those lines. He’s a pretty approachable guy too, so feel free to reach out to him on FB or email and he’ll gladly clarify what he’s done. Your teachers giving you a proof of concept is nice, but I would base it off something that’s already out there too so you don’t get ahead of yourself too much and get lost. Now having Levi’s tutorial here is a great base. We can enlighten on some of the steps he’s done.

He took the idea of an animation in his head through an environment, but as you can see, the camera move isn’t that big. Just enough to sell some of the illusion of feeling 3D. He mocked up the basic layout of the environment using real simple geo in (choose your own 3D program here), as it doesn’t matter. You would then do a wider field of view, master digital matte painting that would encompass your entire camera move, and use your basic models as a guide. Break up the layers with no atmosphere (you can add that later in comp on cards near the mountains), and project those layers from Photoshop onto those mountain geometry. Again, doesn’t have to be exactly the same layout, so you can use World Machine if you’d like, but you want to get that extra step of realism through matte painting techniques in Photoshop. Project, run the animation, see if it works, and then fix the texture stretches, or change around the model just a bit to fit your needs in the animated shot. No UV’s were done in this break down. He just added the checkerboard textures to show off the contours of the geo and how rough the model was. It was all projection based from one vantage point onto multiple geometry.

Being in school still, you can definitely use your teachers to further explain and help you in the process. Bug them lots! :slight_smile: Hope that helps!

#5

Hey David thanks for your reply, hope you’re well! :slight_smile:

Yes I did message Levi and actually he is releasing a tutorial within the next month or so which he demonstrated at Industry Workshops here in the UK to people who went to the event, as I was lucky enough to attend I’ll hopefully be able to get my hands in it soon :slight_smile:

Yeah my teachers kinda want me to prove I can break a simple image apart first and create some parallax and then relight that image as well before I dive into my own full piece.

Thanks for clearing up that UV issue, I’m just so used to seeing checkerboards when mapping my UV’s in Maya I instantly assumed it had something to do with that!

Little more background on my skills/softwares. I am confident in Maya and Zbrush and ofc Photoshop. I just thought using World Machine would be a good base for my matte painting, (I want to go for a Himalayan mountain type shot) and because I would have that geo already I could create a MUCH lower poly version for projecting? Maybe it would be easier just to model some quick low geo to project onto and go about the exact steps Levi used, creating the camera animation and base geo first.

My teachers aren’t ultra familiar with matte painting and projecting, (ones more of just a character painter and the other guy is a CryEngine master). So getting help from you guys is amazingly helpful. :slight_smile:

I’ll make sure to post my tests so you can see my progress, thanks again for all the help!

#6

Ok so I haven’t updated this for longer than I hoped, partly to do with being a bit ill recently and partly directed a group project at University which is taking up more of my time than expected!

So since I last posted I did 2 tests in an afternoon at uni, really quick things just get the hang of how projections really work in Maya. First thing I learnt was make sure you have enough image so when part of some geo gets revealed by camera movement it doesnt show white or grey! :slight_smile: Also learnt that my alpha masks have to be SPOT ON and that the geo has to be a little bigger than the image so that parts don’t get cut off by having nothing to project onto.

On the plus side my idea of creating the basic geo for my matte and then re-projecting back over a much lower res version of the same mesh seems to work a treat!

Going on from here and I feel like the only thing I can do is really get stuck in and carry on with my next steps!

So I created a moodboard of images to show the type of environment I’d like to create.

So I wanna create this like Himalayan mountain range with a buddist temple and animated flags. Most of the images I researched were just from google, looking at real buddist temples, but also the new FarCry game! as it’s set in a similar environment. I was also inspired by the Expedition Everest ride at Disney World :stuck_out_tongue:

Anyway from this research I created this rough concept to get an idea of what my shot will look like.

I figured I’ll make the pillar and temple in 3D, and create the flags and animate them. Then have all of the background as a matte projected back onto various pieces of geo.

So my next step is going to be playing more with World Machine and getting my basic geo created, putting it into Maya and then getting my camera move done. Hopefully I’ll be updating next week with the camera move done so I can get some feedback on that but until then I’d love to hear about what you guys think of my ideas and basic concept.

Oh and P.S. I got a hold of d’artiste matte painting 3 for my birthday last week so I’m ready to steal all your secrets David and Milan! :slight_smile:

#7

Damien also has some nice secrets to share in that book :slight_smile:
there’s also some good digital content you should take a look at :wink: the code is at the end of the book.

#8

Hey guys just another little update.

I’ve been working with World Machine and Maya and decided to follow Levi’s workflow of creating my camera move and basic geo first.

//youtu.be/MsjWnajkJNo

Here is my initial playblast, the idea is that the camera would be on a helicopter flying up and revealing the environment.

There are some issues that I wanna fix, I think the camera move needs to be tweaked but I would love to get some feedback on what people think of the general shot / composition. The red assets are going to be pillars/shrines and the blue lines the rope for the multi-coloured flags as common in the Himalayan area.

From here aswell as tweaking the camera and the environment slighty, I’ll be moving onto setting up my projection cameras and getting my projections to work nicely.

I assume for this shot I’ll need more than 1 projection camera, does anyone have any good tips on placing projection cameras/testing how the projections will look before committing to matte painting?

Thanks again to everyone whose replied and helped out so far, it’s very much appreciated. :slight_smile:

#9

Hi Jack,

Milan alerted me to this thread and I thought I would chip in and offer some thoughts to try and help you.

Your general workflow seems OK. Bear in mind that creating all that geo in world machine will only really serve you if you stick to the details in the geo, otherwise detailing it out using photos and just having the proxy geo for the projection will serve you just as well and save you an extra step.
The camera move is a good first pass, try and study some real helicopter moves to get all the little detail regarding how they move for the final.
If you want to get this looking totally photoreal I strongly suggest you find one or two very similar photos and lean heavily on them to inform the lighting, colour palette, design and general mood.
Projection cameras are best figured out from the camera move. Once you have the camera move nailed, figure out which frames have the best coverage and start using them to create your projection cameras. Sometimes it’s worth widening the focal length of the projection cameras or moving them back so that they have a slightly better coverage than the final shot camera.

You have set yourself a very ambitious project, good luck with it! For the record, your teachers suggestion of stating with just splitting up a photo and doing projections from that with a bit of relighting is a good one to get you familiar with the process.

Keep us updated on progress, good luck!

N

#10

Hi there Nick, thanks a lot for your input! Yeah my next step is going to be tweaking my camera move and trying to nail that. Thanks for the help with the projection cameras, would you literally mean stop my animation camera in certain positions and then create a new camera with a slightly wider focal length in the exact spot for the various projection cameras?

The plan with the world machine mesh was to light/texture it and use that as the base of my projection mattes, adding photo details on top then re-projecting over a lower proxy geo. Would you think this is a decent method or would you just say that skipping straight to the lower geo and creating my projections 100% purely from photos/matte painting?

I did find the tests very handy actually just learning how projections can work within Maya and making sure I have enough matte to project that any parallax doesn’t show any holes!

ALSO on a technical note I had an issue with my alpha channel not lining up 100% with my diffuse, creating like a very small white line or edge, curiously do you use separate jpegs/tifs for the diffuse and transparency within the projections or do you use a format which has the alpha built in like PNG/TGA?

Thanks again for all the help it’s really great of you! :slight_smile:

#11

Hi Jack,

Yes, as far as projection cameras go, that is what I’m suggesting. If you can do the projections from copies of the shot render camera at specific frames then you minimise doing work that won’t be seen by the shot camera or will be from an angle that is so different from the shot camera that you start experiencing stretching or splits in the projection.

As far as your workflow, that sounds fine - if you are planning on actually texturing and lighting the geometry and just using photos as a bit of a touch up then you aren’t wasting your time by detailing that out and making it as good as possible. Just be aware of course that the more you end up replacing with photos then the less use the detailed geo becomes, and you may find that it’s easier to achieve photorealism by just replacing all of the render as it’s a tough job getting a natural environment looking 100% real in CG. But you won’t know how good you can get it until you have a stab at it, so by all means go for it! If it doesn’t work out then at least you can use it as a scale and lighting reference.

Look forward to seeing updates!

#12

Hi Nick,

That’s great thanks so much for your help!

I’ll attempt to texture and light my geo and touch up with photos and see how that works out for me, if I find it not working particularly well for getting it to a photo-real finish, I’ll go back and replace all the detail with photo’s.

Thank’s again for all your help, really appreciated, I’ll keep the thread updated. :slight_smile:

#13

Hey guys, time for another update.

So since I last posted I’ve really been trying to fix up my camera animation and boy it’s really one of those things which I feel like I could tweak forever. I did look at some YouTube footage of like helicopters flying to try and see how I could fix mine and make it feel more realistic.

I also came across this AMAZING bit of reference. It’s a 3D video of a helicopter flying over the Himalaya’s, which you can look around in and choose the camera angles yourself! How cool, kinda like a video HDR image!
http://video.airpano.com/Video-Himalaya-Nepal/video-high-full-en.html

After trying to study where I was going wrong here is my new camera move:’

//youtu.be/GAnlLAILi1w

Now it’s still not perfect and moves slower than a real helicopter but I’m trying to go for that more slow-mo cinematic feel.

I’d love to hear some opinions on it and see if there’s anything you’d change. After this week I’m not going to play with the camera move anymore and move onto the rest of the project.

I’ve also been working on figuring out my projection cameras! Thanks to the advice Nick gave me I’ve created my basic cameras in Maya based off my animation camera. I tested how well these would project using checkerboards and I’ve noticed that I’m gonna have quite a few areas which have stretching and other issues due to the camera move.

However I luckily got to speak to Levi Peterffy and he said to just project in Nuke and then basically create extra cameras where I’ll have issues and “patch” over them with an extra piece of matte.

I’ve tested this to get my head around how it’ll work and checked out Garrett Fry’s super helpful tutorial on using multiple projection cameras on the same piece of geo.

https://www.youtube.com/watch?v=H-a4cIJUl64

So I’ve spent a lot of time I guess working out some potential issues and how I’ll overcome them this week. I want to not touch the camera move by the end of the week and then focus purely on starting up the actual matte painting and then projecting the base and then patching up any stretching and problems I have.

Also I’ve picked the type of lighting I’m going for and as Nick suggested I’ve picked two similar photo’s to really help inform me of the colours and lighting!


Sorry for such a long post, next update I hope to show you guys some more physical progress! Thanks for reading and thanks again to everyone whose helped me so far. :slight_smile:

#14

Just wanted to add this little update to what I posted last night.

So knowing that after this week I wasn’t going to change the camera move I looked at the composition of my shot and changed the placement of my assets slightly, to try and create a more aesthetically pleasing shot.

//youtu.be/z_oNM9J8DSY

Ignore the checkerboard in the background, they’re just a test from my projection cameras which I’ll talk about in a moment. In general though I think the camera tweak and the new placement of the assets creates a much better overall shot and creates this “reveal” at the end of the large temple/monastery.

So I’ve now got high and low poly versions of my mesh. The high poly I’ve popped a very basic tile texture over and I’m lighting so to create base for my matte work. I also wanted to plan my lighting in Maya so that when it comes to creating my assets and comping them in I can match the lighting perfectly.

I used the mental ray physical sun and sky to create my basic lighting, and as I posted in my previous comment the lighting is kinda a 1-2 in the afternoon just after the sun is at it’s peak, casting some shadows on certain mountain faces but generally being globally lit.

Pretty happy with how this has turned out for the base, the material is a little washed out but I really like the cast shadows and their shape. It’s literally just going to be my base so that I don’t start with a blank canvas when doing my matte.

The low poly mesh is going to be my proxy mesh to project onto, I’ve created a few basic projection cameras from my animation camera and tested them with different checkerboards to see where my matte will be stretched and have to be covered up with a patch projection.

//youtu.be/YKXl2WPmDWU

There’s definitely a lot of areas which will have to be fixed, mostly geo laying flat which will need to be patched up to avoid a stretched out projection but I think for the base these seem pretty good?

Definitely would love to see what you guys think of these as a start for my projections, if I should try and tweak these before moving on or just use these to start with and then patch over as needed?

Thanks again to all whose looked through my thread and helped me out :slight_smile:

#15

Hi Jack,

Another nice update. Quick question, do you have s specific requirement for the shot to be the length that it currently is? It feels very long to me, so long that you probably wouldn’t even put the whole shot on your reel, but the extras time near the tail of the shot doesn’t really reveal anything new or sufficiently different to justify all that additional render time.

I’m looking forward to seeing where you will go with the architecture, the design of that will be important and could make our break this one.

Keep it up, enjoying watching this one develop.
N

#16

Hey Nick,

thanks for the reply,

no I don’t have a specific time frame for the shot, I originally just had in mind around a 15 second shot, would you suggest shortening it?

In terms of the final few seconds, the idea was the final piece of architecture as going to be a large “main” monastery which was revealed from behind the mountain edge?

Maybe it would be worth me creating a shorter camera animation?

I will be moving into Nuke this week so any changes I do to my camera will be final after the next day or so! :slight_smile:

Thanks again for all the help :slight_smile:

#17

Hi Jack,

The intention to have the reveal on the main piece of architecture is a good one, it gives the shot purpose and gives you a chance to show off your range of ability. My worry was more about putting a reel together. You only have realistically 2 minutes for a reel, any more than that will be to long IMO (some may disagree). So by the time you are 10 seconds into a reel, usually you will be into shot 2 or 3 including a title plate. If you have shots that are 15 second long, that will eat into a big part of the time you have to show yourself off. So I would see if you can get the shot closer to half that length and see if you can do that without losing any of the impact. If you can it will save you a huge amount of render time as well as not taking up too much reel time. 8 seconds would still be considered a reasonably long shot. In production most sit around the 2 - 5 seconds range. I’ve worked on one big long shot that was 1 minute 30 before now, but when it came to cutting a reel, I only used about 3 or 4 seconds of it…!

Just some things to bear in mind. Your thought process again is solid, just make sure you won’t put a lot of extra work into something that then takes up so much time in your reel that you’re tempted to cut it down or redo as a shorter piece :slight_smile:

N

#18

Hi again everyone, time for another update!

Firstly, thanks again for your help Nick I can’t over emphasize how awesome of you it is for putting time and effort into helping me out! :slight_smile:

In terms of my shot and my reel, I’ll be graduating from University in the summer and this shot is kinda my whole reel at the moment, I decided to try and create something big and as impressive as possible to show to various companies and aim to get a starter position maybe as a runner and work my way up. :slight_smile: I’ll definitely keep in my the length of shots in the future though, especially as you mention that most shots are as short as 2-5 seconds! and as I create more of this type of work create shorter length shots.

My teachers wanted me to create a matte still as a “proof of concept” of how my whole shot will look like. (apart from my final temple)

This took up some of my time and I created the pillar’s for the shot in Zbrush and the basic flags in Maya to create kind of a “super concept”. Would love to get some feedback on what you think works well/not well with the matte work.

I discovered that it’s damn hard finding the right photos to work from! I spent so much time finding images which were would fit, especially struggling to get them at a super nice HQ. (Seems like 70% of Himalayan images on the web are taken from an old film camera from ages ago!).

Nevertheless it was a good exercise to get myself more used to matte techniques!

The image is from the final frame and although the camera move reveals a lot more terrain than what is shown in that image I didn’t wanna waste that matte work so I’ve gone ahead and started to project in Nuke.

//youtu.be/9ExFn23MN5Y

I created this quick test render to show where I am at with it currently.

Here are some shots of my Node Tree and the 3D view in Nuke. :slight_smile:


So basically I feel like I’m at the stage where now I just need to keep on creating matte pieces to project onto my geo (where all the checkerboard is). + (small area on the right which doesn’t have checkerboard yet but there is geo so I need to either widen my projection camera or create another one.

Once all the “base” matte work is in, I’ll then patch areas which are broken by the camera move!

One small issue I’ve noticed and not sure why it is occuring is that for some reason some of my matte work isnt being cut out correctly by it’s alpha? The geo underneath still shows and shows as pure white?

This image shows what I mean, its also happening right at the very back bit of the projection. I’m using PNG’s with an alpha channel which should cut out the exact shape of my RBG images so not sure why this is happening? Any insight would be great!

Hope to ramp up my speed a bit this week and complete as much of the base matte work as possible and build my CGI assets (the main temple).

Hope to update you guys soon again and once again thanks for all input! :slight_smile:

#19

Just a small text update, I realised those white areas which weren’t cut out correctly from my Alpha were just where I hadn’t premultiplied my Read’s! The silliest of things to overlook, hope to update with all my base matte work on the geo soon! :slight_smile:

#20

Very inspiring project Jack :slight_smile: I’m starting on the same journey but with Fusion. Best of luck and I’ll be following your thread with interest!