Ara's Tale - short film wip

loramel · 2011-11-11 09:00:10 UTC

Thanks guys for the heads up :).

next update:

                                                                                                                                                                                                                                     [[img]http://www.loramel.net/blender_minutes/wp-content/uploads/2011/11/shot21a_composite.jpg[/img]](http://www.loramel.net/blender_minutes/wp-content/uploads/2011/11/shot21a_composite.jpg)

Production of the remaining shots of Aras Tale is running under full steam now.

Now with the dragon finished its fun again to do the lighting and compositing for the shots.

I already had the tricks and workflow for all the shots where Ara and the environment is shown. Closeups have a completely different setup than wide shots e.g. For the dragon I had yet to find a good workflow and best practices.

This led to a revision of the linked dragon assets. I now have 3 versions of the dragon, each for a different level of detail. Depending on the shot, I use the most appropriate one. I do this to cut down render time and most importantly memory consumption. With complex shots and 1080p I currently reach a high of ~7GB. I am running 3 frames in parallel on my render machine which has 24 GB, so there is not much room left to play with.

The latest shot I have finished just today was by far the most complex one. And I guess this will hold true for the rest of the shots.

Its a shot with a long camera move covering almost all of the enviroment, partly quite close and also showing Ara and the dragon. That meant throwing in all the effects the evironment needs (clouds, mist, dynamics, sense of scale) and additionally balance this out to highlight the characters in the scene, which demand their own special treatments.

Up to now I was always good with the 20 layers blender provides for organizing all the objects in a scene. For that shot it soon showed, that this was not enough if I was to stick to my tested workflow.

The solution was to split the shot into 2 blender scenes ( one for the environment and one for the characters) and render a multilayer file for each scene. With this I have actually 40 layers at hand, which was plenty enough.

The total number of renderlayers in the combined multilayer files is 21, each one including color, z, alpha, shadow/AO and sometimes index and vector channels.

Putting this all together in the compositor is a tedious task and very iterative. Moreover, to gain the best visual experience, different parts of the shot require slightly different compositing values. This is where the everything-is-animatable paradigm of the new blender series come in very handy.

My workflow for this is to render out key frames ( 5 in this case out of 360) and tune the compositing node setup to the best result for these frames. This in turn triggers changes in the general lighting setup or maybe demands additional renderlayers ( special masks e.g). So its a constant back and forth.

When I am satisfied I turn on the render server and let it render out all of the frames. And after that I let it render out the composite. Thats the first time I actually see the shot in full color and lighting. This typically shows lighting and compositing problems very quickly and another round is due.

If I am lucky, no changes to the rendersetup has to done, which is true 90% of the time. Rendering out just the composite is very fast ( relatively spoken ) and the turn around times are very low. Just as example the 360 frames took ~6hrs to render, whereas a full compositing run is finished after 5 mins.

And this shows very dramatically why I love this type of workflow ( separating all the lighting information into layers written to multilayer files). The final lighting tweaks and tuning/polishing can all be done in the compositor at almost no cost. This invites to do experiments and try different things.

And again, after so much dry talk here are 3 frames directly taken from the movie to show you a bit of what the state of production is. ( The last two frames are from the above mentioned complex shot)

This latest shot took me almost a week to complete. The next ones are simpler again and progress should be faster.

BTW, total mount of time spent is now at 1406 hrs., and I start to feel the wish to finally bring this whole project to a completion

And finally I want to put an almost nostalgic bit at the end of this post:

Looking through the old and very rough concept sketches for Aras Tale, I found this one, which dates back to 30. July 2009. It was this image, planned as a still image project, which triggered the whole movie. And if you compare this with the second shot from ones shown above, you should see some resemblance

 [[img]http://www.loramel.net/blender_minutes/wp-content/uploads/2011/11/sc02.shot21a_concept.jpg[/img]](http://www.loramel.net/blender_minutes/wp-content/uploads/2011/11/sc02.shot21a_concept.jpg)

loramel · 2011-11-28 11:02:20 UTC

                         [[img]http://www.loramel.net/blender_minutes/wp-content/uploads/2011/11/blog_post_title.jpg[/img]](http://www.loramel.net/blender_minutes/wp-content/uploads/2011/11/blog_post_title.jpg)

Confirmed by comments on my recent post on my blog, I will try to refrain as much as possible from displaying potential spoiler images.

As this would mean, to cease showing any images from now on, I thought I could give you some look behind the scenes of the type of work I am mainly involved right now concerning Aras Tale.

This is especially for you, the small but constant number of followers of this project

The typical workflow right now is to setup a shot for proper lighting and rendering and then render out the frames as multilayer EXR files.

Those are then fed into the compositor where the real lighting work is done.
During the setup on my current shots, I recognized, that this would be a perfect opportunity to show you, what I am doing here. And whats best, without revealing essential parts of the core story.

Below you see a somewhat compressed look at how I typically setup the lighting and compositing and can follow the progress on an actual shot up t its final result.
Well, almost final, as the result in this case is only the base for further VFX work. It is also relatively simple in comparison to other shots I have/had to do, so the total amount of time is short enough to be demonstrated here. ( parts are accelerated though )

Well, here is the screencast

and here is the final shot

loramel · 2011-12-21 11:31:42 UTC

This time I deliver a rather dry text-only update:

With this post I want to summarize and share my experiences I had using the particle system in current blender 2.61.

The current production phase calls for a lot of vfx, mostly to be done using dynamic particles. I will not go into detail as to what exactly the shots are, but I think just describing the overall concepts and approaches will give you an enough complete picture.

It soon was clear, that using the particle system from 2.49 was out of the question, as I needed some effects rendered using the volume shading with point density input from various particle systems.

So the first task was to export all the camera animation as well as all the objects and animations for the other needed parts. It turned out, that blender 2.61 does a decent job in importing old 2.49 files. At first look, everything looked ok. On deeper inspection though, I noticed broken rig setups regarding constraints and drivers across multiple armatures.

The effect was, that the resulting animation wasnt exactly the same as in the original 2.49 version. Fortunately, the vfx work needed here did not call for pixel accuracy, so after a few manual tweaks I accepted the setup.

The next task was to come up with a solution and workflow to tackle the needed vfx work. The shots in question require several layers of moving particles, partly rendered as instanced objects and partly as volumes.My intention and need was to be totally in control on how those particles moved. I had a very precise picture in mind how the whole system should behave and move over time. With one word, I wanted to animate the overall behaviour of the particles in a very controlled and predictable manner. But the individual particles should expose a random and chaotic appearance.

An example would be e.g. a tornado, where I control the overall shape and movement but let the system fill in the chaos. Or another example are the vfx used for the death eaters in the harry potter movies. ( That does not hint at any tornado being used in the shots
)

My first idea was to use keyed particles, as those seemed very controllable in various aspects. What I did not realize then, was that they are too deterministic. The interpolation of the particles between the used keys is linear and gives a decisively unchaotic appearance. I was under he impression that random fluctuations and external force fields could be overlayed, but alas, that is not the case. Any physical properties are turned off as soon as keyed particles are used.

Well, that made it clear then to use the physics type particles and use force fields of all sorts to get the desired results. Well, there went the controllable and predictable attributes

The headaches using this approach are manifold. Let me just list some of the shortcomings using this approach:

[ul]
[li]The animatable properties of particle systems are rather limited.[/li][li]Controlling particle properties through textures work, but any animation of the textures themselves is completely ignored and noneffective.[/li][li]Force fields have to be animated by hand. The animated properties act indirectly on the movement of the particles and thus you have to guess the outcome.[/li][li]Multiple force fields with overlapping effects are even harder to control[/li][li]Trying to animate a moving particle stream along a path could be done using a curve force field, but you loose ( to a big degree) the interaction with other force fields and are left with something similar to the keyed approach.[/li][li]To see the effects, you have to rerun the simulation every time you change anything.[/li][/ul]The last point on the list wouldnt be such a problem, if the cache system worked reliable. It happened very often, that I didnt see any changes allthough I tweaked the force field animation a lot. Even freeing the cache didnt work. Only changing some basic parameters (like changing the numbers of particles) I was able to reset the cache.
And I had to do this for a lot of particle systems all at once, thus multiplying the amount of time needed to get a result which at least had some resemblance to my original concept.

The next surprises came when actually rendering out those particles. Each particle system was rendered out to a different render layer all with various lights and masking applied. It turned out that 2 of my 8 particle systems didnt render at all. They showed up very nicely when rendering the single layer alone, but would vanish as soon as all the layers were activated. Setting up those particle systems from scratch solved this problem. Another glitch was that the masking I set up in one layer was also active in another one, and thus ruining the render of that layer. Unfortunately I have not found any workaround for this one.

Another glitch is if you want to have a vector blur pass done. Allthough a lot of warning messages are produced in the console about missing objects in the speed table I got my vector passes. They look nice for 90% of the frames. But for some frames you get huge spikes in velocity for individual particles, which results in sudden long streaks in the compositor.

All in all, I have to do 3 shots with this type of heavy vfx, and I am halfway through. This should give you an idea how time consuming this whole process is.

The verdict on the current state of particles in blender is quite clear: a nice toy if used in demonstrations or simple setups. As soon as you ramp up the complexity and requirements you will have to face serious limitations. I would not use the term production ready for the current implementation. And sadly the same has to be said about hair and cloth as well

Well, sorry for the long rant-like text. I just had to make room in my head to be able to go on My hope now is, that the 3rd shot requiring this sort of vfx will benefit from the experience gained so far, but one thing is for sure: compromises will have to be made

iceycooly · 2011-12-21 23:23:17 UTC

It’s been a while since you made an update, man. Keep it coming.

loramel · 2011-12-22 09:25:15 UTC

I’ll do my best

As I have mentioned in my previous post, several shortcomings in the particle system cost me quite some time.

And additionally it is getting harder to post progress reports without spoiling the story.

But I will keep you posted …

loramel · 2012-01-13 03:44:03 UTC

2012 - where it all ends.

I am talking about Aras Tale of course. For any other visions, apocalyptic and otherwise, I just refer to here.

And that does not imply that there is any apocalypse waiting regarding Aras Tale, quite on the contrary.

The teaser released about one year ago, stated a release date for the final film of 2011, but well, that was a bit optimistic ( again ).

What I can say now, is, that visually I am nearing completion. Shot production is down to 6 shots left, all of them relatively straight forward. Nothing in comparison to the latest vfx heavy shots I had to fight with.

After that, I have a small todo list of fixes for some shots. Parallel to that I will do the final edit on the movie, which will result in a time locked ( if this is a correct term) version of Aras Tale. A perfect base for Phil and Mikkel to start working on the score and the sound fx and final mix.

But to complicate things a bit, I have enrolled in an 8 week cgworkshop course on creature texturing with Mari. 2 weeks are now almost done and its a huge fun and learning experience for me. And as I want to make the best out of this course I will invest quite some time into it. That will call for some balancing of my available free time, but it works quite well so far.

With all that said, I expect the time locked version to be ready by end of february. This will certainly be worth an extra post.

I have no real clue how demanding the audio work will be, but from a CG point of view, I see the time after february a bit more relaxing, as then everything should be in place ( minus some fixes ). The rendering of the 1080p shots can be done easily while working with Phil and Mikkel on the sound. Current estimates predict a rendertime of 28 days. This does not include the managing overhead of tackling the 3TB of data

So from a visual standpoint the movie in full resolution will be finished by the end of april. That lets one hope, that it might be possible to have a finished movie at its 3rd anniversary at the end of July.

But I guess its safe to say it will be ready in 2012.

Otherwise it would be its own kind of apocalypse.

StephanRODRIGUEZ · 2012-01-14 03:07:15 UTC

I wish you best of luck for the end of your travel.

loramel · 2012-01-16 02:14:35 UTC

Thanks Stephan !

I see from your user description that yours is coming to an end too ?

StephanRODRIGUEZ · 2012-01-16 06:08:05 UTC

Yes Martin, 6 shots left, sound to be done, Phil will do my score, seems we are very close together
Unfortunately I’m working on something else for the moment (vfx for a short film) and I’m not gonna finish before my 3 years anniversary (2 of February) …

loramel · 2012-01-28 03:32:04 UTC

                         [[img]http://www.loramel.net/blender_minutes/wp-content/uploads/2012/01/final_cut.jpg[/img]](http://www.loramel.net/blender_minutes/wp-content/uploads/2012/01/final_cut.jpg)
 
 Finally the long awaited major milestone is here !
  
 All major visual development is done and I have edited the movie to a   final cut. This is not yet a picture lock, but what I call a timelock.   Now the path is clear for Mikkel and Phil to do their magic and add   sound and music to this movie.
  
  Now that the edit is done I have exact numbers on the running time of   the movie. Without the credits it runs 7:23 min or 11339 frames at   25fps, which is quite a good match to my estimates from 2010.
 
  Speaking of numbers, here is the actual table of hours spent so far on the project:


         ---------------------------------------------- 
     5:34    01-Script
  43:54    02-Storyboard
  73:54    03-Concept Art
  14:48    04-Animatics
  152:45    05-Modeling
  76:57    06-Rigging
  189:17    07-Animation
  261:00    08-Texturing
  228:11    09-Simulation
  51:48    10-Project Management
  131:54    11-Research
  16:29    Editing
  290:46    Lighting/Compositing  ----------------------------------------------
  1537:17 Total

Now a lot of tedious little bits of work still has to be done. Apart from a lot of fixing, I plan to spend some time on developing a proper logo, movie poster and finally give the official movie site a long needed overhaul. And not to forget, I am reallylooking forward to the song recording with Julia Schaller for Aras Song.

loramel · 2012-03-15 04:18:16 UTC

As this short film is nearing completion, I want to share the movie poster I have done for the film.

StephanRODRIGUEZ · 2012-03-15 11:35:30 UTC

Very nice, really looks pro. Can’t wait to see it finished.

loramel · 2012-04-07 02:13:00 UTC

Finally all render/compositing jobs are done and I have now the full 1080p footage ready.

I still do have the intermediate multilayer renderpasses for each shot, in case there is the need to revisit some compositing tasks. The storage demand is quite huge for these (a single frame typically has ~140MB) , but it already saved me a lot of time and trouble. I guess I will only delete them once the movie is eventually released

The compositing pass creates the final frames which are delivered as OpenEXR files. Here we typically have 23 MB per frame.

I did some intense testing and exploring on how to best tackle the color grading workflow. As I now did a very close look to the whole color behaviour, I noticed that initially I had wrongly calibrated my monitor and had done the primary color grading in the compositing on a display with a gamma setting of 2.4. That means, that on a correctly calibrated display, the movie displays too bright and flat.

My initial idea was to apply the appropriate gamma correction in blender, but as already mentioned in my blog , I decided to try out DaVinci Resolve Lite. The new problem that this did create, was that I had to switch to Windows to work with this software. That in itself wouldnt be a big problem, but I soon found out that the display driver under Windows acts quite differently to the one under Linux. In fact I didnt manage, to get the same color dynamic under both OSes. I got the gamma setting ok, but for the white and black point I had absolutely no chance to get it to work correctly under Windows. The solution would be of course do buy a real hardware calibration tool, but that is not an option right now and I have to find an other way.

This different behaviour made me a bit uneasy and I did numerous tests on both platforms to learn how these differences show in the final color grading. That was when I learned that my monitor has a gamma drift over time. When freshly turned on, I need a correction of 1.20 and after 4-5 hrs I am down to 1.05.

What I do now, is to constantly check the actual gamma value ( using the cool test images at http://www.lagom.nl/lcd-test/) and apply a correction if necessary. Its a bit tedious, but with this I get relatively reproducible results.

What got me a bit frustrated were the tests I did on my 40″ FulHD TV. It provides a load of settings, none of which seems to have anything to do with a gamma 2.2 setting
Most settings have blown out highlights or oversharpening. The 100Hz feature destroys completely the movie feeling. I really wonder how this movie will look like on the various displays when eventually released. This is something that is completely out of my control.

Anyway, for all those interested, I have a 2 shot sequence in 1080p for you to download and look at. I would really be interested how this sequence looks like on your individual display. With a correct gamma 2.2 display, you should see a relatively dark movie with stark contrast, but there is no completely black area. You should still be able to see features even in the darkest spots. But these dark spots should really be dark and not misty white washed. Please have a look and tell me how your viewing experience is and what display settings you have.

Test Sequence

Narntson · 2012-04-10 12:17:07 UTC

Hi Martin,

I’ve had some time to think about how my project will show up on the retinas of my viewing audience… I am not the most into tweaking tweaking at the end, I would rather get it right the first time on a complete render but the main issue is matching shots for continuity! I don’t mean to hijack your WIP thread so I posted a longer version on the main page to ask members about their methodology:
http://forums.cgsociety.org/showthread.php?p=7293324#post7293324

Anyways I had original hopes (on my own material) to render out a single pass without the fuss of color correction, then I soon learned I had to reconsider what was fine on my computer monitor was not on my HDTV. In my case, my plasma tv shadows crushed detail that I never adjusted in my computer monitor. Therefore after bumping up my pc monitor’s gamma the renders looked better on the plasma. I do plan on watching your scene at later in the day (will let u know) and when you finish your project will you give us a lesson on what you’ve learned and how to troubleshoot for optimal results? Thanks!!

loramel · 2012-04-10 14:42:03 UTC

Hi Noah

Thanks for stopping by

I am really interested on any posts in the thread you created. Looking at all these different displays, I began to suspect that there must be some way of ensuring that the material will be seen how it is supposed to be.

Now I am not so sure anymore that there is a single simple way. You have absolutely no control how your movie is going to be shown. Starting from the type of projector, color settings, screen type etc.

My approach now is to concentrate on a proper gamma 2.2 calibrated display and grade in such a way that a) my artistic intention is still intact and b) I am in a safe region, especially in the bright areas. On two of HDTVs I experienced a tendency to increase the highs and got easily washed out results. Staying a bit away from the possible maximum helped in this regard without compromising the look on the computer display.

This is purely empirical knowledge and even in this only tested on a few selected displays, so I might be totally wrong.

Narntson · 2012-04-12 18:27:14 UTC

Okay I looked at the clip from my wife’s macbook pro (Gamma 2.2) and it looked great. The PC looked great with windows media player! How did you export your movie BTW?

loramel · 2012-04-13 00:00:02 UTC

Good to know it worked on your systems.

For exporting I used ffmpeg under linux to create the h264 movie clip you have seen.
This was just a first test to see what can be done using this tool.

When released I will have to provide several versions for different playback targets. Currently I am investigating very high quality settings, which are compatible with the bluray standard.

loramel · 2012-04-20 01:20:21 UTC

I declare the visual development for Aras Tale as finished .

This is not 100% true, but I declare it nonetheless. Actually I still have to do the lipsyncing for two shots. This will be done after Aras Song is recorded, which will happen this very weekend.

The lipsyncing part should not interfere with any visual developments done so far. I just have to adjust the lip movements to Julias voice, feed the shots into my render pipeline and be done with it. You can be sure to hear if this turns out to be different

For the sound recording, I purchased a very simple but quite capable recorder device, a Zoom H1. It records uncompressed WAVs at a sample rate of up to 96kHz and 24 bits.

I got some very helpful tips from Mikkel on how to approach this recording. One of his tips was how to build yourself a cheap popfilter, which is an essential tool for the song recording. You may know the filters from images, where a big round net is positioned in front of the microphone.

I built mine using an old badminton racket and a pantyhose !

This filter works amazingly well. With this type of professional gear the recording session with Julia should go like a breeze

And finally, I put some stills directly taken from the picture locked directory for you here on the thread. Looking at them as stills reveals quite some problems an shortcomings, but used in the actual movie they work pretty well.
( The images are actually links to the full 1920x1080 images, but somehow they do not open when you click on them, so just follow the link and you will get the full resolution)

loramel · 2012-05-24 09:50:13 UTC

I guess doing the lipsyncing as the last step in the production of a movie is not what you would call common practice. It is not even good practice. But then, thats just another first, of the many others encountered during production

After the recording session with Julia I managed to clean up the takes and provide Phil and Mikkel with a bunch of solid raw takes. Julia you did a fabulous job. Thank you !

While being swamped with work, Phil was able to do a very first draft of the song, mixing Julias voice (properly postprocessed) with orchestration. This first version is already very beautiful. Thinking back at the moment when I actually wrote the passage in the script, I never imagined to have such a touching song in my movie.

Again, this was one of those magic moments I encountered throughout the production, when individual pieces come together to form something new and this new is more then the raw sum of those parts.

Well, anyway, this finished song provided me now with the last missing part to finalize the visual part of the movie, which is the lipsyncing.

And it came to no big surprise that some lessons were learned,
again.

First, the rigging of the mouth sucked horribly in regard to create plausible mouth shapes for the song. But changing the rig at this point was out of the question, so I had to tweak a lot. Having a song with long held notes helped definitely here

Next, some decisions I made for speed improvements came now back to me to bite me very hard. For directional reasons and to fit the images better to the music, I edited the song sequence in such a way, that I used parts of some shots twice with different points of the song. Regarding rendertime this was a good decision, but, as you can imagine, trying to get this to work with lipsyncing is a big fail.

For one shot I was lucky, as Ara is only seen from relatively far away, and the overlap is at long held notes of the same vocal. Phew !

For the other shot, which is a closeup of Aras face, I had no chance to cheat in this way. I cursed the editor and tried to cheat anyway, until I noticed that in the edit I still had 14 unused frames at the end of the shot, which allowed me the shift the overlapping part by exactly this amount. And since the overlap was only 10 frames I was saved. Double Phew !

Another interesting thing to do, was not only to match the lip and mouth movements, but to also use the facial controls to match the emotional state of the song. I did some self observations how my lids, brows and forehead moved when trying to sing the song myself (it is a good thing nobody was in hearing range
). It is always an enlightening experience to observe what small movements of the brows can do the a faces expression.

The last two nights I managed to finish this final task as well and currently my render machine is humming along and creates the final renders. These will be finished on Sunday and then the visual aspect is truly finished and I am eagerly waiting for the sound and music department to fill in the missing pieces.

Its soooo close now

magilla · 2012-05-24 18:59:58 UTC

excellent stuff - so much hard work