Missed Anniversary

Well, I was looking back on this log and discovered that I had missed the one year anniversary of the start of production, which would have been December 4th. I wish I could report more progress in that time, but I can't say that I've been working non-stop for the entire period either.

The timeline on the movie now stands at 9min 11sec of completed footage, and I'm still expecting it to clock in around 20-25 min in length when completed. I'm still working on the first chase sequence in the film, the one that introduces us to the Wolfman. There was a moment that called for a splash of water, and I am still not comfortable around post production particle effects (and I have no idea how to pull them off within Poser). Instead, I opted to use a bunch of minature 3D balls, made translucent and shiny, to replicate the effect. It's not great, but it works.

In 3-D movie news, there's a remake of Night of the Living Dead in 3-D on the horizon. You can visit the website here at nightofthelivingdead3d.com, where they've posted a 3-D photo gallery you can view with your red/blue 3-D glasses. As the project was filmed using a system developed by Dimension 3, the red lens of the glasses has to be over the right eye in order for the effect to work.

Universal Pictures Homage

Hope all of you U.S. visitors had a Happy Thanksgiving. I indulged in so many left-over Thanksgiving dinners over the past couple of days I fear I will never be able to eat again.
I've only recently been able to see the complete series of Universal Frankenstein movies. I had only really seen the original and the Bride of Frankenstein. Last year I picked up the Frankenstein Legacy Collection DVD set and finally sat down to watch all of them. They kind of go to seed from Ghost of Frankenstein on, but Son of Frankenstein, with it's great Expressionist sets and long shadows, is visually a neat experience.

One of the things I love about those movies (and this applies to the Wolf Man movies as well) is the use of gargantuan indoor stages to create outdoor environments. To see it produces a surreal effect, as if the movies take place in some kind of Gothic fantasy land and not necessarily our reality. I used to think this was done on purpose by the filmmakers, but I recently came across an interview with James Whale, in which he proclaimed a realistic interpretation of the story, and that it took place in the modern (1931) world!

Intentional or not, the effect is something that I was hoping to design into Frankenstein vs the Wolfman as a bit of homage to those classic films. Again, the movie I'm making is not a sequel or in any way related to the Universal films (the characters of the Frankenstein Monster and The Wolfman are quite different than those established by Boris Karloff and Lon Chaney, Jr), but I thought it would be nice to acknowledge the inspiration.

I've just completed a couple of quick shots that mirror a moment in Bride of Frankenstein - Leeta, being chased by the Wolfman, runs through an artificial forest of tree trunks, with a large cyclorama of a cloudy sky serving as the backdrop - just like Karloff all those years ago.

Returning to Work

Ok, I'm lying. I haven't returned to work... but I plan to return to work.

I haven't done anything on the movie for the past 3 months. I consider this time a vacation - some time away to get some perspective on how things are going. I've rewatched the existing footage multiple times in that period, and screened it for friends. I expected to be happier with the results than I have been, and I've been trying to get a lock on the problem.
My biggest complaint lies in the visual look of the film. It's very bland. Part of this is due to the constraints of working with 3D. You are restricted in how much you move the camera, and especially on how you angle the camera. Tilt shots are difficult, if not impossible, because they throw off the vertical registration of the stereo image.

The second hurdle is the lighting, again dictated by the 3D process. I'm trying to design the movie towards whichever format the most people are going to see it in. We know that the field sequential 3D version is probably going to be seen by the smallest group, so that leaves either the red/blue anaglyph 3D version or the 2D version as the one which will get the most exposure. If most people's experience with the 3D version will be anaglyph, then I find I have to over-light my scenes in order for everything to show up in detail all the way back into the frame.

By over-lighting everything, I'm losing the moody, shadowy look I had on RAVEN 2. Even the night scenes here are brightly lit.

There's also an issue with the speed of the action. When animating, I find it difficult to accurately judge how the speed of something will look when the clip is finished. I've been spending some time going through the assembly edit doing some fine cutting, and have changed the speed of some slow action in the fight scenes to make things a bit more natural and exciting. However, since we're eventually going to be dealing with an interlaced version of the film, I'm concerned that those clips that have been sped up will contain flickering artifacts in the field sequential version.

So, I'm warming myself up for a return to production (hopefully by this weekend) by designing a new main title sequence. The old one was kind of dull, so this new one takes the basic concept and spices it up a bit.

I also took the opportunity to back up a bunch of the raw video files that were sitting on my crowded hard drive, in order to free up more production space. I'm filling a small stack of DVD-Rs with the stuff.

More as it happens...

Egads!

For the past three weeks, I've been slaving over a 30 second sequence in which the Monster leads the little girl Leeta through his graveyard home. The full sequence consists of three shots; due to errors in the right eye (I miscalcuated the axis of seperation, making one image higher than the other, and not side by side as necessary), I've had to go back and re-render the whole thing.

But now - finally - it's in the can. You can see it here (the high poly gravestones are the culprits for the long render times).

Also, we recorded the voice of Milo, the third of our trio of children, this week.

En Fuego

Added two more finished frames to the Gallery section. Have spent this last month doing a lot of overtime, but still managed to get quite a bit done on the movie. There is now close to 7 minutes worth of footage completed.

The current scene I'm working on takes place in the Frankenstein Monster's lair - a cemetery. It took a couple of days to build the cemetery itself, painstakingly positioning each of the headstones individually. There's a special effect involved in this scene which requires some special planning in the rendering process. The Monster is carrying a lit torch, and in order to create this effect, I am using an After Effects plugin from Panopticum, called Fire. By supplying an alpha image of the object you want to burn (i.e. the head of the torch), the plugin creates a fairly good flame simulation.

However, since the torch passes behind a number of objects, each shot involving the torch has become a composite shot. The background elements, including the Monster and the other figures are rendered seperately from the foreground objects (gates, trees, etc). The flame plugin is applied to the top of the torch (also rendered seperately), and then sandwiched in between the foreground and background layers. The still I've uploaded to the gallery shows the lit torch, but in the full scene, the gates start off closed in the foreground.

I also wanted a nice, horror movie graveyard fog for the scene. Initially, I tried a fog object I found on the internet which used multiple foggy planes all ganged together to create a sense of depth; this proved too time consuming to be practical (I used it in the first scene of the movie, however - that shot of the Monster emerging from the mist demonstrates the effect).
For the graveyard, I opted to go the cheaper, faster route - simply applying a photo of Photoshop-generated clouds to a flat plane hovering above the ground, partially transparent. While this works ok (see the graveyard image for details), I'm trying something else in future shots. Currently, I'm using a Wave Plane (a water simulation that can morph and ripple) with the clouds photo applied. This way, the fog actually rolls and boils.

Lipsynching

On the last two RAVEN movies, synching the character's lips to the recorded words was a labor intensive, painstaking chore. For FRANKENSTEIN VS THE WOLFMAN, I'm using an impressive Poser add on, called Mimic. Mimic can be fed a recorded line of dialogue, and automatically interprets the sound file and generates lip movement. As an added bonus, it also generates facial expressions! This has created much more expressive characters than I've had in the previous two movies, and has greatly reduced the amount of time required to set up individual scenes.

I'm working on the first indoor sequence in the movie, which takes place in a sewer. The soon-to-be Wolfman is spilling his origin story to a captive, soon-to-be meal. This sequence will be done in two parts; sundown and moonlight, separated by a flashback scene. The moonlight part of the sequence will involve the central Wolfman transformation, so I'm amped to start working on that - possibly as soon as tomorrow.

Finally Added the Characters Page to the Site

Yep - by clicking on over to the brand new character's page here you can check out the who's who of characters from the movie, and the people who play them. It's not a complete list, as more characters have yet to be recorded and modeled, but we're underway.

I've been following my own advice - last weekend I animated about 30 seconds worth of stuff, and the computer has spent the bulk of the week rendering it out. It's not done yet, but I hope to get around to working on a new scene this Sunday. I have finally finished off the first sequence of the movie, which is also the first full beginning-to-end completed scene. Yaaay! Only 34 more to go!

How to Make 3-D Movies With Poser

I'm really getting derelict in my duties of updating this webspace. The job situation mentioned in my last post has been rectified (I survived the cuts, and am keeping my employment), but my hours have changed. This has cut down on the amount of time a day I can spend working on the movie. My new strategy is to animate as much stuff as I can on weekends when I have free time, and then let the computer do the work during the week. I am one shot away from having the first complete sequence finished for the movie.

Now, the question has become - how do you make a 3-D movie using Poser? The whole concept here is to photograph your subject from two perspectives, separated by the distance between your eyes (approx 2 1/2 inches). To do this, I put a reference object in the scene (namely a simple ball) and tell the camera to always Point At it. Where the ball is positioned will be the screen plane of the shot - anything coming between the ball and the camera will appear to come out of the screen, anything that falls behind it will appear to receed into the screen. The ball is hidden in the final render by simply turning it off. The final render of this shot becomes the Left Eye view. Then, I rotate the camera on the Y axis 2 degrees to the right, and re-render the scene. This is the Right Eye view.

This technique works best when the camera is facing the object straight ahead. When the camera is tilted to look up or down at an object, the Y axis becomes useless, and I have to manually track the camera to approximate the perspective shift. This has resulted in a couple of errors so far, which has resulted in too much parallax (the seperation between the final images when put together in 3-D). Too much parallax, or worse, a misregistered image (when the Right Eye image is slightly higher or lower than the Left Eye) can cause some serious eyestrain in the viewer. So, some of these shots had to be fixed by adjusting the Right Eye image later in post processing. There are two shots so far in this first sequence that will have slight black bars on the side of the frame, because I had to slide the images together to correct a problem with depth (objects that were supposed to be at the screen plane were too far in front of it).

The two final images, if they were rendered without problems, are combined in a program called Stereo Movie Maker to make both an anaglyph and field sequential version of the shot.

Voice Recording

It's been busy around here. There's a bit of a transition going on at my job (which could leave me out of one), so that's been on my mind lately. We rounded up a new group of voices and recorded them last week; we're now left with one major role to cast, and one or two supporting roles. Here's the cast list to date: Grace Carreno, Tom Zack, Charles Reeder, Meghan Campbell, Andrew Carlson, and Elaine Clarke. One of these days I'll actually get around to posting the cast photos!

I went and designed much of the main title sequence after having an inspiration one night; I actually started trying to ape some of the typeface and graphics of the old 1930's movie trailers, but settled on something that looks like a 3-D version of something you'd see on late night Chiller Theater. The depth works great, and I'm happy.

I'm currently working on that opening sequence - another shot that has to be split into multiple layers to render without overloading the computer. This is taking longer than anticipated, but at least we're moving forward.

Steady As She Goes

Hit a bit of a setback, as the 20 second shot I was talking about in my last post had to be scrapped after a week's worth of work. It turns out I didn't calculate the axis of depth correctly for the 3D and both right and left eye views had to be trashed. This after rendering out six layers of elements per eye! The shot is now complete, and correctly registered, but talk about trying a man's faith.

I have moved on to other shots in the sequence, but progress is slow. I've abandoned the idea of animating everything and then coming back to render it later, in favor of setting up the animation and then rendering out the dual eye versions. This takes between 2-3 days per shot, and maybe more time will be required later, as the scenes become more complex. I'm doing this to check the final 3D version to make sure everything is correctly registered.

Going Forward...

I have been able to achieve an acceptable level of production from Poser, so I guess that means we're back in business. I've also had to re-evaluate a couple of positions within the last week.

First, since Poser absolutely refuses to produce the opening shot of the movie, I've had to chop it into layers to get it done. Basically I'm doing it in passes - on this pass, we render only the ground; on the next, the kids walking; on the following, the trees, etc. The shot in question is an overhead tracking shot that follows the three children from our story to a grisly scene in the woods. It will consist of at least five seperate layers. Once each layer is rendered, I'll composite them together in After Effects. I did this sort of thing a lot with the final sequences of RAVEN 2.

At this point, the one minute sequence which I animated back in January - and have been rendering out ever since - is done. I'm using this as my test sequence, and have prepared a DVD with the three versions of it for study. Although my wife says that my character animation has improved since the RAVEN days, the overall look of the footage is taking some getting used to. The desaturated color scheme is more severe than it appeared on my PC monitor. In order to accomodate the 3D, I overlit everything - and on TV it looks almost too bright (especially since the scene in question takes place by moonlight).

Due to the nature of interlaced TV fields, the field sequential 3D version displays more noticable jagged lines on objects than I was anticipating. The anaglyph version requires significant tweaking of the TV's color settings to achieve a tolerable effect (it does work, but the depth effects are not as pronounced as the field sequential version, and ghosting is a fact of life). Maybe this could be remedied by including a couple of setup screens on the eventual DVD.

And, the streaking title sequence will probably not make it to the final film. The style is just not in keeping with the tone of the movie. Maybe I can use it in the trailer...

I keep looking at the thin script sitting on my desk and wondering why I wanted to make this movie so complicated. It's scope, in both the number of principal characters and its locations, are much more ambitious than the two RAVEN films. At this point, nearly four months after starting the project, it feels like I'm standing at the base of a mountain, looking up at the summit - and it's beginning to dawn on me just how much work is going to be required to pull this off. I'm toying with the idea of taking another pass at the script and paring it down, striking most of the peripheral characters and making it more of a three character piece: the little girl Leeta, the Frankenstein Monster, and the Wolfman.

We'll see what happens. The computer's time for this week, and maybe some of next week, will be taken up with rendering out right and left pairs of each layer of this first shot. More as it happens...

Production in Jeopardy

It has now been three months since any real animation work has been done on "Frankenstein vs the Wolfman". And now we've hit a snag that may cause the whole project to collapse.
I upgraded my computer to an AMD 64-bit 3000+ processor and brought the system RAM up to 1GB. You would think with the added power, things would work smoothly.
You would be wrong.

Poser is a tempermental program. Although it supports animation, it's really designed for making still images. I've come to this realization because for the past 2 weeks, I've been trying to get it to render out more than 40 frames at a time without crashing with "Out of Memory" errors. This is despite the fact that it's only using a fraction of the total memory on my machine (both the physical and the virtual memory combined). Forty frames is just a hair under 2 seconds worth of animation, and it would take a lifetime to complete this movie at that speed.

Funny, but with 512MB on my previous machine, I could get Poser to output between 100-140 frames before giving up the ghost. Was it too much to expect that additional RAM would allow me to render more, and not less? The upswing is that it renders faster, but it hardly seems an even trade.

I'm not defeated yet. There are a couple of ideas I have for tweaking some of the system settings, hopefully to coax the program into better behavior. But, if I can't get results better than 40 some frames before a crash, then I'll be shutting down the movie.

Stay tuned...

Casting Underway

Last week was a pretty productive week. Andy arranged to have about half of the voice talent come in to the studio and perform their lines; once casting & recording are complete, I'll post cast names and character photos to the Characters page. We've only have 2 leads and a few secondary characters yet to fill. This gives me enough material to begin animating characters.

Unfortunately, this week has been slowed to a stand-still. I've been trying to work on some animation, but my PC resisted. The current file that I'm working on takes up 726MB of system memory, draining the system resources and slowing the computer to a crawl.

So, as seems to happen each time I make a movie, we had to upgrade the computer. The new beast chugs along quite nicely, and is rendering a test frame from the troublesome scene as I write this.

Main Title Animation

What makes a memorable 3D title sequence? I've been looking at some of the title sequences created for 3D movies of the past in search of inspiration. In the 1950's, the titles didn't actually shoot at you - instead, they appeared as static three dimensional block lettering, hanging just infront of the screen.

During the 80's 3D wave, a company called Celestial Mechanix dreamed up the memorable 3D title sequences that streak toward the camera in JAWS 3-D, FRIDAY THE 13th PART 3, METALSTORM and THE MAN WHO WASN'T THERE. I was reading an interview with them in an old Cinefantastique magazine that revealed their method of composition was to start the title at the screen instead of (dimensionally) behind it, and then to have it zoom towards the audience, with the streaks going back to the screen plane as a reference. Apparently, without that reference, your eye has trouble determining just how far off the screen the titles appear.

I have also examined the titles to SPACEHUNTER, (done by R/Greenberg associates, most famous for the SUPERMAN titles and the inviso-Predator effect), AMITYVILLE 3-D, and TREASURE OF THE FOUR CROWNS, most of which start behind the screen before zooming toward the audience. Only SPACEHUNTER's title graphic makes use of the streaks, and in direct comparison, Celestial Mechanix's philosophy seems to be the simplest and most direct, producing the best results. (As a comparison with modern 3D titles - Robert Rodriguez's SPY KIDS 3-D title graphic starts behind the screen and streaks up to but not past it.)

After searching around (with no luck) for an After Effects plug-in that would create the streaking title sequence, I decided the best way to go about it would be to just create 3D object text in Bryce and animate the title graphic extending toward the camera on the Z axis, producing a long streak. The background of the sequence is a CGI cloud shot that I generated in Photoshop and After Effects, and composited in position of depth a few inches (or feet, depending of the size of your display) behind the screen. The inital credits - "Daredevil Films presents", etc - actually appear on the screen plane, but seem to hang in 3D space due to the fact that the clouds appear so far behind them. Then, the composited title logo materializes and shoots out toward you, and the dimensional effect is pretty dramatic.
The sequence runs about 20 seconds in total. It may not be the most original thing in the world, but hopefully it serves as a homage to those great 3D title sequences of the 1980's.

New Entry and Minor Site Update

Posted field-sequential 3D versions of the two gallery photos for those of you with LCD shutter glasses for your PC (again, if you're into 3D TV, field-sequential systems are the way to go- Razor3D has a single pair (wired) for $39.95).

I just discovered the first legitimate, feature length field-sequential 3D DVD is now available - it's Full Moon's THE CREEPS, which I caught theatrically in a one-night only 3D showing a couple of years back. The movie itself is the same kind of B-movie stuff Full Moon has been cranking out for years (they're the compnay that brought you the Puppetmaster & Demonic Toys movies), but the picture quality is better than the bootleg features available on eBay. Hopefully, if this disc sells well, it may convince other studios to release field sequential versions of their library titles.

We are currently recording voices for Frankenstein vs the Wolfman; as of this writing we've got one part in the can, with others scheduled to be recorded in the following weeks. I haven't really been doing any new animation while I wait for the voice work, but our home PCs are hard at work cranking out final renders for the completed scenes.

Also, I've done a test with Lipsinc's Mimic software, which is a program that analyzes sound files & automatically generates Poser lip and expression data: I fed the Frankenstein Monster a line and have seen him talk for the first time. The surprise discovery was how much life the program gave to the facial features while the character talks - much more than just simply matching lip movements to words. Very cool.

Ok, Here's the Plan

While working on some new shots, I've been rendering out the right eye version of the first couple of shots that were produced; I now have my first seconds of final 3D video. The 3D in motion works as well as my test frames suggested. I did make a sizeable goof in one 75 frame shot (about 3 seconds worth), which now needs to be re-rendered for both left and right eyes to correct the perspective problem.

Anyway, the plan... I'm trying to figure out how this thing is going to look in its final DVD incarnation, so planning has to be undertaken at this stage to ensure the best quality result. I'm producing all the images at 852x480, then importing the resultant right and left eye versions into a freeware program called Stereo Movie Maker. The program creates 3 AVI files for each shot - a 2D version (made from the left eye video file), an anaglyph 3D version, and an interlaced 3D version. This will probably cause me storage headaches down the road, as about 15 seconds worth of footage equals about a 400MB uncompressed AVI, per version.
Then, I take the 2D version into Adobe Premiere and edit it. The goal is to go straight from the Premiere timeline to a DVD-ready MPEG2 file without any intermediary steps, which would cause compression loss. To acheive this (since I'm using Premiere 6), I'm evaluating a program from Videotools called Premiere Video Server, which allows you to use an encoder, such as TMPGEnc, to produce a final video stream directly from Premiere. I did a test using Cinema Craft's MPEG Encoder, but was unsatisfied with the results, and have opted for the freeware TMPGEnc for the quality of its video output.

The 852x480 16x9 image is smashed into a 720x480 anamorphic frame, which is then decoded back to a 16x9 image by your DVD player. I haven't burned a test file for playback yet, but looking at it on a computer monitor displays no edge problems associated with most video scaling operations.

Once I have the 2D version finished, I can then have Premiere replace all the 2D clips with their 3D counterparts, while retaining all my edit decisions. The encoding process can then be repeated for each subsequent version of the movie.

So, there you have it. The initial tests of this process have gone well, so we're off to the races. For those of you interested in current stats, there is 1:14 of finished animation; 39 seconds of that is final, rendered 2D animation; and 17 seconds of that exists in 3D.

Website Update

Back to work on movie. Taking a breather here to do some website reprogramming. Upgraded the Synopsis & Trailer pages, and added a working Photo Gallery. Posted two finished movie stilsl in both 2D and 3D.

Reworked the poster, which uses virtually the same artwork as seen on the main page of this site. Since the robe the Monster is pictured wearing broke repeatedly when I was trying to pose it, its been changed to Daz's Morphing Trenchcoat for Michael, which looks and works a lot better. Hopefully, you'll be seeing it soon...

First Slowdown

I've been procrastinating about getting back to work on this sucker. For the past week, I keep circling the computer. It dares me to sit down at it and do some work. I had been working on a sequence that's come to be known as Scene #29, which is the opening of the final battle sequence. After completion a chain-of-events, I'm stuck now dreaming up how to handle the next section of the scene. I have in mind what needs to happen, but working on the same set, with the same characters lends it a feeling of deja vu.

I've come to realize that to keep my interest high on a project, there needs to be something new interjected into it at regular intervals. That can be as simple as the introduction of a new character, or something more complex, such as a change of scenery. In all the movies I've done so far, the fight scenes are the most challenging because they have to be done in short cuts and multiple setups, all featuring wild motions. For those of you who saw the terrible climactic fights in the two RAVEN movies, rest assured that the footage here is looking much improved. There's a better sense of fluid movement and physics, which hopefully will serve to make it a little more realistic and exciting.

Andy tells me that the basement recording studio is coming along. Prior to this, he, and his cohort Rob, had a mobil setup that allowed them to bring facilities to the talent. Now, with dedicated space available, they hope to make something more permanent. Once the equipment is all wired in, we can begin recording the actors.

I've been generating some headshots of the movie's characters, both in 2D and 3D, for eventual posting on the Characters page. I'm still holding out to see if we can get photos of the actors playing the parts to post along side them.

Happy New Year

The holidays broke my forward momentum on the project, and I'm still trying to get back on track. I'm using my PC to do the animation, and my wife's computer to render it out. Unfortunately, the scene I'm rendering is so memory intensive that it keeps clogging up the system around frame 195 or so (about 3 days worth of work), and has to be restarted. I have set it to render out only to frame 190, after which I will render out the remainder of the 278 frames as a seperate file.

At some point here I have to do an encoding test to see what this will look like on DVD. I'm rendering the frames out at 852x480 widescreen, but DVD only accepts 720x480. The idea being that the widescreen aspect can be squeezed anamorphically into the DVD compliant frame size. We'll see if that holds water after I have a minute or so of completed footage to test it.

We still haven't done any recording. Hopefully, that will come together soon. I will post more artwork and the script to the website once the bulk of the voice recording is in the can.