Workflow for making an animated short

[nicmarxp]

My goal is to make an animated short lasting somewhere between 2-4 minutes.

I’ve made and rigged all the characters, made common actions and poses for them. I also have environments and have a detailed storyboard. I’m not sure about the timing and duration of the storyline though.

I’m trying to understand how to use the NLA editor to plan and edit my movie and switch camera angles and motions.

Most tutorials on the NLA editor ste very basic about making actions and blending them, but I haven’t found any on making a complete movie.

My idea is to make everything rough and ugly, to get the timing and approximate camera angles right, and then fine tune the animations.

What is a good workflow of actually putting it all together?

I’d love some tips on how to stay organized and effective.

[ania]

Is it with the shoulder shrugging alien?

I have a different workflow than you, not so NLA-centered. Doesn’t mean it’s better, just extract what you think will work for you - it can also depend on the kind of animation and story, if NLA is useful or not.
Good planning is key. If you spend enough time on planning, you will save a lot of time of working (you avoid making objects or animations which turn out unuseful and throwing them away later).

• First I write the story down. Some good ideas come while writing, and it helps not to forget the little small ideas / gags / extras, which often come to my mind one day and are forgotten the next week if I don’t write them down.

• Storyboard. I do it on paper, for me it’s more intuitive than digital and also helps the creative process with new ideas. Should be quite detailed, including camera, framing, composition. You say you have it, so very good

• Animated storyboard. Just scan your storyboard, throw it into the Blender Video Sequence Editor, with enough time for each drawing to see what’s happening. Here you will already see that some scenes don’t work well together, so you can further experiment and edit the storyboard. Work on it until you think everything will be understandable and will look good. If your storyboard is nicely drawn (mine aren’t), show it to someone at this stage and ask for opinion.

• Animatic. I spend much time with this, it’s the last stage of planning.
  At this stage I set up the blender file system. I have all characters rigged (but partly very low poly “dummies”, which will be refined later). I put each character into its own blender file. Then I make scene files, one file for each shot. That means, each new camera angle is a new file, basically. I link in the characters as proxies, so if I change their file, they change in all scene files.
  Then I animate very roughly. Just storytelling keys, with constant interpolation. That means that the character jumps from one pose to the next, with rough timing, with enough time for a viewer to read the pose, and it’s supposed to be understandable for the viewer. Instead of walk cycles, I mostly just move the character’s main pivot linearily, without moving feet. I set up the camera as final as possible. There always can be changes later, but I try to avoid - it’s always extra work, especially if I animate a character just for a particular camera angle and it looks bad in other angles.
  I render each scene file in OpenGL (viewport render) and cut it together in another blender file (Video Sequence Editor), render the video sequence. Here again, check if the shots work together as a cut. Refine, change the camera if necessary, insert new shots if necessary, delete shots which don’t bring anything into the story, check again. Sometimes you’ll have two great shots, but they don’t look good next to each other, and you have to change the camera angle or put a third shot inbetween.
  This is the point where you should show the entire Animatic to some trusted people and ask them for comments, ask if they understand what’s happening, ask them if they see redundand scenes which can be left out. They don’t have to be 3d artists, it’s even good to show it to some clueless people, they see different things, less technical, more story-centered.

• You could make a filmed animatic (act and film all scenes with your smartphone). I don’t do it, but I know some people do and certainly it’s a very good workflow element. If timing is important for you, a filmed animatic will certainly help a lot.

• Final animation. Scene by scene. I start with the most difficult scene for each character, to see if my rig really works in difficult situations, and if necessary, I refine or change the rig.
  I always film whatever I can for reference (with myself or one of my friends as actor). If that’s not possible (because I don’t have good sword fighters or dragons among my friends, for example), I search the internet for video references of real people and animals. Then I study that reference - for me that means drawing the key poses down on paper, with timing notes (for how much frames did he swing the sword, etc.). I don’t use that 1:1 in the final animation (cartoon animation requires timing exaggeration, and sometimes I just improvise without reference), but it always helps, especially for timing purposes, and also gives some new ideas about animation details and movement.
  And here again, show the animation (can be single scenes, or several scenes cut together) to someone you trust. You work on it all the time, you sometimes get blind for the obvious, or you understand what happens but noone else does.

• Render, fine cut, bla bla. From now on it’s easy

Huh that was long. I hope you could extract something useful for yourself from this.

I only work with NLA and pose libraries if I think they are useful for a particular scene. In final animation stage, not sooner. Mostly I just animate directly, in a single action, but of course there are situations where NLA is the better way (especially for very repetitive or cycled actions, I think). And I rather animate and set poses for a certain camera angle, so the pose or action can look awful in other views - no use in making them before camera is finally set.

I get the timing mainly from filming or references, but actually I don’t care - in animatic stage it’s rough timing, it gets almost final in final animation, and completely final in the fine cut. It gets a minute longer or shorter - so what? (unless you’re animating for a music video or something like that).

I’ll follow your progress (as far as you’ll be showing it on the forum). I think you’re learning the same way I do - throwing yourself into deep water immediately, not practicing with bouncing balls first Much more challenging, but also more motivating this way I wish all the best for your animation.

I’ll end with some reading which helped me a lot. I’m not coming from a film school (if you are, it’s perhaps too basic for you).

• Roy Thompson - “Grammar of the shot”. It’s written a bit boringly, but systematic and basic knowledge about film making. Helped me a lot. There’s a similar book by that author, Grammar of the edit, about cutting.
• Steven Katz - “Film directing shot by shot”. Better to read than the Thompson book, and uses examples for good directing from different, well known movies.
• Marcos Mateu-Mestre - “Framed Ink”. Very inspirational book about composition and storytelling.
  https://vimeo.com/22552454 about storyboarding.

And an advice about storytelling I got from a more experienced collegue - each shot should end with a small or big “question”, which can (but must not) be answered in the next shot or somewhere later in the story. Also, each shot should show something new, which the audience didn’t know yet. Small example - character looks off camera: question is where? next shot shows the object. Or he grabs for something, question is what? nest shot shows it. Or, there’s a swamp and some suspicious waves moving, and the viewer asks himself what creature lurks there. Or an ordinary shoe is shown, the viewer asks himself why it’s so important to deserve its own shot, and it will be cleared up at a later stage of the story. Etc.
If there’s no question in the shot, not even a small one, or nothing new shown, it’s boring for the audience and you can throw it away as well. So for example showing the same thing from two different perspectives would be boring, unless the first perspective lets questions open (for example by not showing the entire object immediately, so the viewer first doesn’t recognize the object).

[nicmarxp]

Hi Ania, wow, that was an amazing and extensive post, thank you so much for taking the time to help me!

Yes, it’s about the alien, I was so surprised when I read that sentence, didn’t realize it was you.

I can agree that I tend to jump into deep water, but I actually did make a movie with bouncing candy balls :). It’s really bad (especially the 5 minute tractor) but I learned the basics of keyframing and physics, so it was part of my learning and in the mindset of making simple kids movies to make $$$ on YouTube. After I released it I realized there are 15.000 movies about ”Learn colors” released on YouTube every 24 hours…

But at this point I was already hooked on the art, and since then I have studied modeling, rigging, lighting a lot, and watched a LOT of tutorials, so even if I have very little time to work with this, I love it, and feel motivated to finish the movie. Now back to the questions…

Your answer is very similar to the workflow about animatic editing by Colin Levy from Sintel that I recently saw on the Blender cloud.

Scanning the storyboard and playing it back to set up timing is a great idea, will definitely try this next. I mostly draw on paper, and make small 4x3 cm draft images. I don’t draw very good, but it gets the idea through

Also putting up a file structure is very interesting, up until now I have a rather big file (which I think contains a lot of unused data), so I will make a group of my main character and link it into a new main movie blend, and keep continuos work in the separate files.

What I can’t get my head around is why to have each camera angle as an OpenGL render in separate files. It seems like a lot of work. If some adjustment needs to be made, and I recut the GL-files in the VSE, how do I get that back into the main timeline?

My movie will be like 2-4 minutes and could be split up into ”mini scenes”, with main parts of the storyline, but they are all in the same environment, so not really scenes. What’s the downside of just editing it right away?

The books seem interesting, but at this point I rather spend my very limited time on putting it all together. I’ll check the video though. I’ve read 4 blender books which I’m very happy about, but none of them was about composition and storytelling. I like the advice about ”end with a question”.

Actually it would be fun to share the progress, it was quite a while since I posted, and a lot has happened since. I really like the character now and will post an update in the thread.

Thanks a lot again, it was super useful!

[ania]

About the cameras - you don’t have to do it that way, of course. In my previous shorts, I didn’t separate the scenes like that, but had every camera in one blender file, and did all the animation, camera switching etc. there. It was of course much less work with setting up the file system, but it caused several annoying problems later, in final animation stage, rendering, and cutting:

• Main reason: If a shot (by shot I always mean one camera angle from cut to cut) becomes longer for any reason (mostly in the course of final animation), you have to move everything that’s after the shot. Every next camera, every action which follows, every object animation, grease pencil keys etc., has to be moved in time. That’s annoying and cumbersome. And if you forget to move some keys (happens), than one or all of the next shots becomes broken. Shots change duration all the time in final animation phase. If each shot is in a separate file, you don’t have to care about any future shots, you can just concentrate on the shot you’re currently working.
• If you need different light in a shot (because for a close up, for example, you might need some extra light or other sun angle, than for a total of the same scene), in separate files you just change the light. All in one file - you have to animate the light.
• Any render settings, which are different for different shots. In separate files, you just change, instead of animating. That’s less complicated.
• You can remove all objects or characters not visible in the shot. So you don’t carry (and have to animate switch on/off) all scene objects, but just the visible ones (which I usually keep in another file and link as group).
• Quicker render, quicker open, with many small files instead of one big file.
• I prefer animating a character between frame 0 and 500, than between frame 12000 and 12500. Shorter numbers But ok that’s stupid
• If the big file gets corrupted or broken for some reason, it’s a huge loss. Imagine you’re near the finish, and something like that happens. If one of the small scene files, or character files, or the cut file, gets broken, it’s a small loss. You can recreate it. Broken files do happen, not often but they do.
• Fine cut. Generally cut. I think it’s just more convenient to cut movies in VSE, than some NLA strip shifting. Also takes less time to render, once all shots are rendered.
  I first render the shots in OpenGL directly into some h.264 format (during final animation phase), and use them for cutting. OpenGL is just quickest possibility to render, and doesn’t require any setup, it just renders what you see in the viewport. Later, when all animation is final, I render the “clean” final render into image sequences (with composition etc.), and just replace the OpenGL movies in VSE by the final renders. That’s not much work.
  You ARE in your main timeline while cutting the OpenGL renders. The main timeline is in the cut blendfile. All other scene files are just single shots and start at time 0 and last the duration of that shot. No need to “match” them into the main timeline.

The environment - I would put it in a separate file, and link it in as group into each scene file (or several groups, if just part of it is visible in certain cameras).

[nicmarxp]

I see, so each separate shot, is also a separate blend file. I think that makes sense. It just seems like so many small files, but maybe it’s even better like you say, to not mess up future parts.

I’ve done a bit of regular filming, and I think it would make sense to shoot (animated/render) one camera angle, if it would be the exact same sequence, and then use parts of it in the final cut. For example:

0.00 - 0.05: A guy slowly draws a gun and aims at the camera. In the final edit, on 0.03 you see a close up of another guys eyes, looking scared.

I seems more practical to animate the entire “gun drawing” sequence of 5 seconds, and then finetune where the other shot gets in. Does that seem reasonable?

Thanks again, your advice is really super-helpful!

[VANDERHORST]

ania

Good read…have you ever tried to work in scenes for one shot instead of a file per shot?

[ania]

nicmarxp - yes, it absolutely makes sense. I always animate and render each shot slightly longer (both start and end), so I have some room for cutting. And if a shot is interrupted by another as you wrote, but then continues with the same action, I also would put it into the same file, yes.

VANDERHORST - yes I did… It was my very first animation project, everything was stuffed into one file with several scenes, it was one big mess. I don’t remember any more why, but I didn’t like scenes, maybe it was just because the entire project was chaotic, and I was a Blender noob at that time.
What is the advantage of scenes? I might try it in my next project.
I did 3 shorts so far and now working on the 4th, always learning from the mistakes I made in the previous ones, the workflow changed and evolved. The current one (which I described) works well for me and feels organized, but it’s not perfect, and I’m always open for ideas how to improve it.

[nicmarxp]

Thanks, I loved your demo reel by the way, especially the “worm”-thingy, which I remember I’ve seen in some science magazine, in huge magnification from en electron microscope. Is that correct?

[VANDERHORST]

ania i have no idea what the advantage of scenes is, its only a question about your experience and they are far more that i have.
At the blender conference, mad animation studio (https://www.youtube.com/watch?v=f6yHZDzoT1c start at min 8) show there workflow for the very first software test. They have different blend files for each shot with a specific scene name (shot_01 and so on) and link in the end all scenes in the VSE.
Its very smilar to your wotkflow…i didnt know about the scene linking in the VSE.

[nicmarxp]

That’s funny, since I also just found that I can add scene when I messed around with the VSE. To add a scene from another file, you first need to link it in, and make sure to name the scene first, since you can’t change the name of it afterwards.

I thought this would be an option instead of using GL renders, and show the live action as it actually is in the scene. That should make a faster workflow if it worked.

However I noticed playback problems with the VSE, which you can also see in the video VANDERHORST linked at 9:39, it plays at 7 fps.

I just linked a very simple scene of a box spinning at 500 kb, the scene is available in Blender, however when inserted as a Scene in VSE it plays at 6-7 fps.

When I rendered it to an AVI JPG at 960x540, and imported it into the VSE as a movie, it plays at 24 fps, even at 50 fps if I change the frame rate.

So am I doing something wrong, or is linking scenes not a feasible option?

[ania]

I haven’t thought of linking scenes into VSE like that, thanks for the video link. Certainly this workflow has some advantages, and less “intermediate” files crowding the hard disk.

OpenGL renders slower than movie files though. The difference becomes larger, the more complex the scene is, since it has to calculate all the geometry. Movie or image input in VSE will always render the same speed, independent of scene complexity. It only depends on the input and output formats, some are faster than others.

Well OpenGL renders have to be done anyway at some point, at least if you don’t have instant viewport playback for watching what you animated, so it’s just the decision if you do it immediately from the scene file like I do, or while cutting like MAD studio does. I suppose they have more powerful graphics cards at the studio than me at home, so perhaps they don’t feel the slowdown so much?

I like to render directly from the scene files while working on animation. I render into quicktime h.264 format, and view the files externally with a movie player, to see what has to be improved in the animation. Mostly I don’t even open the cut file at that point (except sometimes for the end and start of the clip, to match movements with the previous and next shot).
Just for the final render I render image sequences (.jpg, .exr or .tif depending on the pass), then put them through compositing, render a .jpg sequence from there, and use this as replacemet for the OpenGL movies in VSE.

About the “worm”-thingy , that was commercial work, I made only parts of it (my main contribution was animating the “worm” = tardigrade = water bear) https://blenderartists.org/forum/showthread.php?301205-Wild-Little-World here’s more details.

[nicmarxp]

Thanks for everything, I’m ready to start animating now. Feel free to comment on my updated post on my alien modelling.