Posts Tagged ‘Proposal’
Recent Events
June 17th, 2009
Quite a bit has happened since I last updated this little soap box. First and foremost, The Proposal, the film that I had worked on for nearly 9 months, is being released in theaters this Friday. If you stay all the way until the end, you can see my name in the credits! Go see it to support the Massachusetts film industry!
After The Proposal wrapped up, I had about a week off before being called back to work on a commercial, and about 3 weeks ago, I was “reactivated” to work on a fairly complex shot for another movie that Brickyard has been working steadily on for quite a while.
In other professional news, Brickyard Filmworks is trying to launch a new training program designed to attract and train talent for entry-level positions when Brickyard wrangles in a new film. I think it’s a great move, not only for Brickyard, but also for the industry in MA. If enough serious talent is in MA, then more and more work will come out of LA and into my favorite state!
finishing school
March 26th, 2009
So recently at work, I’ve been assigned many tasks which have been dubbed “finishing school”. Finishing school is basically another word for anything it takes to put the final touches on a shot. For the most part, that involves painting. Painting is a great way to fix things like crunchy motion blur, popping keys and my favorite/most hated thing in the world, hair.
When restoring motion blur, I generally have to reveal the background plate first, then paint in as much of the original motion blur as I can. If the original plate shows through too much, I have to reconstruct blur by using a smudge tool or simply lightly painting in a solid color. Many frames later, I have to take my freshly painted frames, take a difference matte between the comp that I painted on top of and my painted frames so that I can restore the grain that I lost through blurring and smudging.
Hair is a little different story. In my experience so far, there are two types of hair. windy, fast moving hair and stiff furry hair. First of all, the reason hair needs to be painted in the first place is because pulling a key and doing edge refinement on hair is a) very difficult and time consuming and b) not worth the time in most cases. Windy hair is easy. For example, let’s say you replace the background of a shot with an actor in the foreground. Her hair is majestically blowing in the wind. I get the comp with most of the hair visible in the comp, but where small blurry strands of hair pass over the replaced background, they disappear. All I really have to do in this situation is find the right color and opacity to paint with, and carefully paint back in the hair frame by frame. Since those windy little strands move so much, painting on each frame works great since the hair jumps around so much in the plate anyway.
On the other hand… If no one bothered to comb the actor’s hair before shooting and he or she doesn’t move around too much, you’ve got a real problem on your hands. Small little stiff hairs that stay put are just as hard to key as those long fast moving ones (at least they are on the show I’m working on… we’ve had to work with some real crappy footage). In this case, I’ve had to start a new comp, using the frames from the compositor in order to track in still frames of hair. Some times, this isn’t so bad. If the actor doesn’t turn his/her head, it’s easy to get a good track and integrate the patch of hair into the rest of the head. If, however, the actor tilts and rotates (which of course happens way more) I have to do a million different little patches and tie it all together with paint strokes. This lead me to my current dilemma:
For the past 3 days, I have been working on the head of a single character in a single shot that is no larger than 400 square pixels when he is closest to the camera. The shot itself has been through 65 versions while I have taken it up to version 70. The pipeline of this shot is as follows… I first work in Nuke to track in patches of hair on top of the comp. This involves trackers, 2d transform nodes and gridwarp nodes in order to keep it looking like it should really be on his head. once that’s done, I bring those ‘prepaint’ frames into Silhouette so i can paint out color bleed from the original plate, blend the hair patches into the rest of his hair and fix some odds and ends (this process cannot be replicated by the way… if i change the patches, I have to repaint…). Next, since no one was able to stabilize the plate appropriately in Nuke, we had to have a Flame artist do it. So I have to take the frames onto a flame and run it through a Furnace Steadiness node. Finally, I take it back into Nuke so I can split screen a tree from an earlier version that somehow started to sizzle in the later version.
Now… some of this is bitching. I could re order these steps to reduce my work, but every step is so dependent and fragile that at this point, no one wants to mix things up and change the order of opperations. A rationale that I am ok with considering if something goes wrong with it at this point, I might go crazy.
Anyway, we’re almost done with the movie which is great since I’m getting very tired of working till midnight every day!
endurance and rotoscoping
February 2nd, 2009
I’ve been made aware that it’s been quite a while since I’ve written anything. The title of this entry is endurance and rotoscoping, so let me try to explain what that means. January has been a very busy month. Since January 1st, I have been at work every single day except a single Sunday. I’m fairly exhausted, but also quite happy with what’s been happening and how much I’ve gotten accomplished just in this month.
At work, we’ve had deliveries to make twice a week. Every Monday, we post low-resolution quicktime movies of our progress and on Friday, we ship a hard drive full of 2k .dpx sequences to LA. This means late nights during the week to get ready for Friday and weekends to get ready for Monday.
Another thing that has changed since the new year started has been our methods and practices regarding rotoscoping. There are 4 dedicated rotoscope artists and one of us is a lead roto artist. Her promotion carried new responsibilities that no one had before. Appart from rotoscoping herself, her responsibility is to schedule the 3 other’s time as well as her own and to approve roto as either “temp” or final”. This change happened at roughly the same time as our new VFX supervisor, Sean, arrived. Since his arrival, nearly every procedure has been adjusted and tweaked for efficiency and I have to say, I think we all feel a little silly at the way we were working before he arrived. At least, I do!
First off, my job as a roto artist has been defined clearly and concretely. I no longer use Nuke to roto, nor am I ever expected to pull keys or provide a blurred edge to a compositor. Before this rule, it was a long back and forth between roto artist and compositor about where an edge should be, how blurry and edge was or how best to pull a key. We now use a program designed for roto called Silhouette. There is no question as to what my final product should look like. Any lens or motion blur, any fine hairs, any kind of transparency is to be addressed by a compositor. At first, it was frustrating because I felt that my job had become too simple and I wasn’t being trusted enough as a compositor, since that’s really what I want to do. However, we’re absolutely churning through roto these days and the compositors seem to also be churning through shots at an amazing rate. Just last week, the drive with 2k comps on it left Brickyard with 78 shots on it. That’s way more than we ever sent in one delivery.
Anyway, I am super tired from working all the time, but I am so happy to have had (and keep having) this experience. The other benefit to working this way is that I feel like I’m getting really good at rotoscoping. Since I don’t even think about how blurry an edge is, or anything like that, I devote 100% of my time to animating shapes. Using trackers and corner pins help a lot to stabilize shapes, but there’s no substitute for layering in motion. It really reminds me a lot about what I know of character animation…
When I start a shot, I first look at what I have to do. If I have to roto houses, or some other kind of rigid object or if I have to roto people or animals. Is the camera panning, dollying or rotating? At this point, I decide if I should try to use trackers to extract motion. Things like houses and cars and planes are great to track since their shape doesn’t change very much. Even if there is a change in perspective, it’s worth it to track a point and animate the change in perspective after the motion has been applied to the shape. People are generally not good candidates for tracking points since clothes, hair and muscles change shape often, especially if the person is walking, or shaking their head or something. If this is the case, I go into character animation mode.
First, I draw separate shapes for parts of the body that I know move independently. Take, for instance, an actor shaking his or her head. I’d probably make separate shapes for the right and left side of the face, both sides and top of the hair, nose, chin and mayb eye brows and neck.
Next, I start to block out the motion based on key poses. If the motion is too eratic or it’s too difficult to tell when the actor has reached a pose, I just pick an arbitrary interval like 6 or 12 frames. Once that’s complete, it’s just a matter of subdividing each frame region until the motion matches. Sometimes, I can get away with changing the interpolation of keyframes to ease in or out, but I have found that it’s quicker to simply place a key fram in between and is also easier to diagnose problems later on.
The other important thing I’ve learned is to try not to ever manipulate individual points on a curve. Instead, try to transform the shape as a whole. This avoids that “wonky” drifting that happens all too often when you start pushing and pulling points. Obviously, sometimes you have to change the shape, like when an actor turns his head.
I remember early on, some one giving me a “standard” time for how long roto should take and it was something like 1 minute per frame or something like that. Now I know that that is an average of all possible circumstances, but it’s starting to make sense to me. Sometimes, you can really nail your key frames and the rest of the motion is automatic. sometimes, you have no choice but to touch each and every frame. I don’t think I roto that fast, nor does any one in reality, but I have to say that I understand it more now, and don’t think it’s a ridiculous number.
Anyway, We’ll be finishing in late March which seems to be approaching faster and faster. I am happy, however, that I’ll get to go to the Alumni Screening in New York this month to see “The Incident at Tower 37″ finally finished! I really am getting pretty excited about seeing it. I remember just last year feeling as if I couldn’t stand to see another fish person for the rest of my life. But now, I’m so happy that it’s been finished and from what I understand, it’s looking and sounding great!
Holiday Update
December 28th, 2008
I am still working for Brickyard VFX on the same movie. To be perfectly honest, I don’t even like the movie, but I count my blessings every day that I get to work on any movie at all! Our due date for the entire movie is March 23 and as of yet, we haven’t been told of another movie in the works. This is a little concerning since if there is no project for us to work on, we don’t work. I am, however, confident that something will come along soon because they’ve started to fill full-time staff positions which means they want the company to be a more permanent enterprise. So, in a nut shell, I hope to be here in Boston for a while.
The one big thing that is supposed to change quite soon is that I will be moving to Cambridge. I found an apartment about 2 blocks from Central Square, which is close to 2 miles from my office which should make my commute much much shorter. My future roommates are all my age and are either working or in grad school. I’m very much looking forward to moving to my new place and getting settled for the new year!
The Proposal Trailer
December 10th, 2008
Well, the trailer for the movie I’ve been working on is finally out. You can view it here on Trailer Addict. Now you get to see what I do every day. In all honesty, there’s not a lot of my work in there given the fact that rotoscoping is not exactly the most visible part of the post-production process.
Anyway, things are well. Very little has transpired that is new or exciting, however I remain ever hopeful.
Long weekends
October 9th, 2008
No, not the good kind of long weekends. Working on movies is hard. You have to work a lot. last week was 12 days straight of 9am to 8 or 9pm. At night, your brain gets fried, but you still have to manage to crank out that last little bit before you go home. It’s hard work too. It has to be pretty much perfect. you have to nail it pixel by pixel, and that takes a lot of time and brutal attention to detail. I’m trying to get better at it. It’s taken me a long time to get my shots done… in the past two weeks, I’ve been working on three or four shots consistently. when i hand stuff back to the lead compositor, it takes a while for him to get back to me with fixes, so i start something new or come back to something else i need to fix. It’s a big revolving door until a shot gets finaled. Luckily, the only shot that I was allowed to composite from start to finish is finaled!
It’s a relatively simple one in which a boat driven by ryan reynolds, sandra bullock along for the ride, sails romantically into a clear blue ocean. Except they were really just motor boating in some harbor. I had to take out the islands in the background, littered with houses and antennae and replace them with horizon.
The absurdity of this movie has become a big joke at the office now. We keep getting feedback about making the mountains and scenery look more like Alaska. We’ve been matching the concept photography pretty closely and using 2nd unit footage and in-house footage all shot in Alaska, but we still have a long way to go. We keep asking ourselves: “why not shoot the movie again, but this time, in Alaska?”
Of course I wouldn’t have a job if that happened, so for the time being, I’m content.
Planar Tracking
September 15th, 2008
Since I’ve been here at Brickyard, I’ve been learning a lot about rotoscoping and composoting. I thought I knew a lot, but it turns out I knew just enough to get this job! I thought I’d share a few of the most interesting things I’ve learned so far.
The first, and most revolutionary thing to happen to my rotoscoping philosophy, has been planar tracking. I’ll admit, I’m not too sure of the theory and math behind it, but I’ll do my best to explain it based on observation and backwards engineering.
Last week, I was told that one of the shots I had roto’ed still had a jittering matte line. The line was clearly jumping up and down on the rock jetty that the camera pans by. I had tried pulling this matte a hundred different ways, so when I was told that it was still unsatisfactory, I couldn’t believe it. I had tried locking at rotoshape from between 1 and 4 tracking points, applying translation, rotation and scale, but my matte would not lock to the image. I also tried a series of overlapping chroma and luma keys, but the lighting and focus on the rocks produced an edge that buzzed like crazy. The only way I could think of was to rotoscope the jetty by hand, frame by frame.
I was not looking forward to this task, but I attacked it head on. Half way in to one frame, I was stopped by our director, Geoff, who asked me what the hell I was doing. He sat down with me while I went over what my issues were, and he suggested i do a series of planar tracks in Mocha. I looked at him blankly as I had never heard of planar tracking or Mocha. When I opened up the program and started using it, however, my head almost exploded.
Basically, planar tracking is basically tracking many points at once instead of tracking one point in particular. Many points means a polygon. A polygon means a rotoshape! I started by tracing small sections of the jetty. Just 5 to 10 rocks per shape, and after each one, I tracked it to the end. What happens is that each point automatically tracks based on its neighborhood of pixels. so as a rock moved, or a corner shifted, the point defining that edge would move along with it!
Of course I had to help it a long from time to time, but the whole process took me about two hours and produced the best matte I had yet. Mocha was the program I used to track my shapes, and I rendered out a single alpha channel for use in Nuke.
What puzzled me was how quickly and accurately it tracked all those points. When I was tracking some of those rocks in Nuke, it took me 3 or 4 tries to find a good rock that would track the whole way through. Even then, it would take 3 or minutes for that single point. Mocha took about 1 second per frame with the shape shown here. I can only guess that there is some corner cutting going on in the tracking algorithm. It must use the relationship between the points in the shape to guess where the next point is going to be, because sometimes I could see points slowly drift when the color of the point converged with the sky.
All in all, this was a pretty cool experience, and I hope to learn more cool things like this as I go along.
rotoscoping
August 19th, 2008
My first week at Brickyard went pretty well. My job is complicated in its simplicity. My primary role is rotoscoping. For those of you that aren’t CG savy, rotoscoping is the process by which objects in a moving image are isolated in order to apply any number of visual effects to that image. As you can imagine, the process is not terribly exciting. It involves a lot of tracing and while technology has gone a long way in making the process more efficient, it is still largely a manual process.
Having said that, I’m enjoying myself very much. I’m working beside a lot of really cool people and in an exciting and productive environment. There are about 13 of us. 7 are leads, 5 are juniors (including myself) and there are several other people that work in capacities I have yet to determine. They all keep me quite busy with shots to roto and have been great about helping me adapt to using Nuke, our compositing software.
I hope that later on down the road, I’ll be able to take a more creative role on a shot in order to add to my portfolio of work. It’s one thing to say that I simply worked on this project. It’s another to be able to show a full shot that I created from start to finish.
Brickyard
August 12th, 2008
I started working at Brickyard VFX yesterday. It’s a pretty cool company. It’s located at 353 Newbury St. in Boston. The building is three stories, retail on the ground floor, commercial production on the first floor and film work on the second. I’m on on the top floor working on a Disney feature film.
At the moment, I’m rotoscoping a shot for use by one of the compositing artists. I have to remove the blue screen and part of the background behind two actors having a conversation. One of the actors is in perfect focus which makes pulling a matte quite simple. The other actor, however is in the foreground and thus quite blurry. With a little manual rotoscoping and blue screen matte pulling, I was able to isolate both actors in about 10 hours.
I really like working here so far. I’m paid well, and the company pays for lunch as well as pretty much all the snacks I can eat. The computer they supplied me with has no expense spared with eight processors, a pretty nice video card and two matching LCD monitors.
I wish I could post some screen shots of the stuff I’m working on, but that would definitely be a breach of the non-disclosure agreement!
Many happenings
July 22nd, 2009
Due to the infrequency with which I’ve been updating my blog, I’m going to have to do this in somewhat of a list format. **EDIT: some of this post is a repeat of my last one, but with more detail. I should read my previous post before writing a new one**
-The Proposal:
The movie that I had spent close to 9 months working on was released to the public almost two months ago. While this was and is very exciting news, being overcome with both joy and frantic life-purpose finding, I neglected to announce this fact and offer any reflections. (I was also unsure of the terms of any non-disclosure agreement I was under, but I believe it’s safe to talk about now).
Having completed such an epic undertaking, I have a hard time remembering how we all got through it! The sheer volume of data that went across all of our screens seems inconceivable to me now. Along with my fellow junior artists at Brickyard, I can safely say I’ve learned, un-learned and re-learned everything I thought I knew about the visual effects and computer graphics industry. From theory to process, I’ve gained a solid foundation and even more questions. I can only express what I have now as an unquenchable thirst for more!
-The (movie I probably can’t mention the name of yet):
More is what I got. Almost as soon as The Proposal wrapped, I was called back to work on a shot for another movie. Considerably more complex and of a wholly different nature than anything we’d done before, the shot consumed 100% of my time and is the biggest culprit responsible for my blogging hiatus. Instead of rotoscoping, I was tasked with producing 3D assets for a 25 second aerial shot involving the placement of photo-real roof-top objects that, at times, are very close to camera. Without going into too much detail, this shot has tested every single aspect of my knowledge and talent and still continues to do so.
Working for a new and fairly small company, one must be utterly flexible and be prepared to wear more than one hat. Often times, wearing multiple hats at once. Over the past month and a half, I’ve played the role of pipeline developer, modeler, shading and lighting artist, matte painter, render wrangler, rotoscope artist and compositor. With the addition of another similar shot, I’ve had the chance to refine my methods and I’m pleased to say that the second time around is going quite a bit smoother than the first. The fact that I got home at 9pm instead of 5am the day before a delivery is proof of that!
-Brickyard University:
In related news, Brickyard Filmworks is going to be putting on a training program for young and aspiring artists in the Greater Boston Area. This program is designed to train local talent for eventual work on feature films that Brickyard will work on. The reason this is important to me, is that I will be the instructor! Working with our VFX Supervisor, I’ll be teaching people how to rotoscope and paint-fix for feature films. I’m excited about this for two reasons. First, I love teaching. Second, with more fresh talent at a small company, that means I get to move up and possibly become a compositor, which is also something I love!
The program is supposed to start soon, although the actual start date depends on our current production schedule.
-SIGGRAPH, 2009:
The Incident at Tower 37 will be in the SIGGRAPH 2009 Computer Animation Festival! We always talked and dreamed about it, and it finally happened! I bought my plane ticket to New Orleans today, and I can’t wait to go see it on that big, gigantic screen alongside friends and hundreds of similarly enthusiastic people. I feel very proud and fortunate to have been part of such an amazing team of artists and filmmakers that have (see above paragraphs for proof) profoundly shaped my life.
-Miscellaneous:
Other more sundry things in my life… I bought a bike. I spent 300$ on a road bike to take me to and from work every day. Ironically, it was more expensive than my recently plane ticket to San Diego.
I bought a new camera too! Since the untimely demise of my old Canon Rebel, I had been reluctant to buy a new one due to monetary concerns, but the upcoming Ultimate tournament in Wildwood, NJ prompted me to get a new Rebel. This one is smaller, lighter, has a larger image sensor, and can even shoot live action HD footage! It uses H.264 compression, but I couldn’t ask for more from a still camera.
In anticipation of assisting my boss with an FX Phd (www.fxphd.com) class he teaches on advanced Nuke compositing, I’ve been given a semester of classes at FX Phd. I’m taking a class on DSLR cinemetography, crowd simulation with Massive and 3D tracking with SynthEyes. The classes have only just begun, but I’m already having fun with Massive and learning how to better use my new camera.
-The End:
I hope to write more often so I can go more into more depth about individual topics that cross my mind. With the volatile nature of the VFX and post-production industry, who knows… I may have a lot of free time to write!
Tags: Brickyard, FXPhd, Proposal, SIGGRAPH, Surrogates
Posted in Comments | No Comments »