Help with shots and scene setup in blender for animated films

I have been searching around for information on this topic, and I can’t seem to find people discussing this.

What I’m having trouble with is how shots in animated films are made in relation to .blend files. Is it one shot per file? One shot per Blender scene, and one file per film scene? Is there a way to link “sets” and models over shots in order to save file space? Is there a way to utilize the Python API to render multiple .blend files? I am also interested in previz. Do you do previz the same way as final renders? How would rendering previz animations work?

I know that was a lot of questions, but I find it a bit hard to describe my thoughts, and frame it as a simple question. Does anyone with more experience with multi-shot animation want to share their workflow? I’m mainly trying to find the most efficient way to go about it, while still keeping everything manageable and keeping a relatively small file space.

Any tips or advice would be greatly appreciated!

1 Like

Usually this, as in one camera position or movement. As soon as it’s a different camera angle, etc, it’s a new shot and hence new file.

I did do a video that has an overview of how I did a short test animation shot, so that may help as a starting point.

My channel also has a couple of videos on folder/file management for animation. The sooner you get organised the better, the number of files add’s up fast.

1 Like

So I can’t answer ALL your questions but for my own short film I have a whole directory of Sequences, each sequence has a shot file, so if a sequence has 3-5 shots, then it’s 3-5 files at the very least, some of my directories have multiple versions of each shot as they evolve. Some are very simple and easy shots so usually only one version of that, but the shots that have a lot going on… then I version out my shot files.

In my short film I also rely a LOT on linked files, this does in fact save space and time but if you make modifications (before final render) then this is reflected downstream. So if I make a change to a character file that is linked in to a shot file, it will also reflect that change for any other shot that character is in. This was something I started from the start and was intentional because of technical reasons other than what I described above but mainly for that reason.

As for rendering multiple files? I am sure there is I am just not sure how using the API. What I do is I use the CGRU/Afanasy renderfarm project and just render locally on my small network, it handles how to render the files itself all you have to do is enter in some basic info in Blender and send it to the server. It’s open source and very active, very powerful tool.

I would invest in a cheap server and install the Kitsu project as this would GREATLY improve workflow as well, there are even addons that can link Blender with Kitsu, that’s just my opinion. It is an open source project management system built for animation and visual effect work, the developers do host however it is not exactly cheap.

1 Like

If planning to do that, it maybe worth a read on my adventures with it: Using Kitsu for solo production tracking

Not saying it isn’t an option, but I do think one needs to carefully consider if it’s worth the time (especially for a single person) vs say just a spreadsheet.

2 Likes

Just a hint for rendering multiple blend files: there is a free add-on called “loom” which has even a UI for that and you can easily setup that. https://blender-addons.org/loom/

1 Like

Just to add another 2-cents, to give you more ideas and perspectives…
(ultimately you will find what works for you, and even that will evolve over time based on your experiences.)


I use command-line rendering almost exclusively.

(I’m just one human with one computer, so the idea of rendering several .blend files simultaneously is not on my radar yet… but command-line rendering has extra options to make it easier to spread a work load across multiple computers in a render-farm.)

Command-line rendering runs blender without the full GUI (lower overhead), and blender prints out a lot of text to keep you abreast of progress (frame number, and render-times being the most important).
In practice, this means configuring the objects, animation, lights, cameras, output file-paths & formats through the normal blender GUI, then save & exit foreground blender and issue a command such as

$ blender -b fileName.blend -S sceneName -f 1..250

“-S sceneName” is optional if there is only one blender-scene in the file.
“-f 1…250” specifies the START and END frames to render…
so if a power outage happens after frame 41 completes,
then reissue the command as

$ blender -b fileName.blend -S sceneName -f 42..250

I keep an image-viewer program open in the output directory, so I can see how the frames are coming out. Stepping through the images (via an image-viewer program) helps you “scroll” forward & backward through the finished frames to spot problems.


I swear by the Workbench render-engine, but it’s quite unlike Eevee & Cycles, so it has to be learned separately. For my computer, every frame render takes about 1.5 seconds regardless, but the frame format seems to be fixed at 720p. Eevee is almost as fast, so you might prefer that, but I’ve come to enjoy the render options of Workbench. The “look” immediately tells me this is just a test clip (just testing animation motions).

Before final rendering the next sequence, I render out all the frames in Workbench, then assemble the sequence using blender’s Video-Sequencer, and review the animation. This helps me catch & correct stupid mistakes or oversights before wasting a lot of time rendering in Cycles (my preferred final render-engine, but Eevee is catching up spectacularly).

To pre-check lighting and depth of field type issues, I do spot renders (using the final render-engine) of a few frames to see if I like it. Adjust, repeat.


Link & Append – Linking things from other .blend files into this .blend file… such that changes in the source file propagate to the calling file… versus Append which lets you copy things from .blend to .blend just once (later changes to source file are ignored).

Scenes – controls in the topbar, to the right, default scene is named “Scene”.

Background_Scene – in Properties > Scene > Scene > Background_Scene. This trick lets you divide the background and the foreground (animated characters) into different blender-scenes … if that helps you concentrate on the “action”.

These 3 topics can help you divide up the conceptual workload, and potentially save a lot of disk space.
Personally, I have grown leery of Link’ed assets. I understand the benefits, and “changes to the character a propagated to calling files”… but in practice, I’ve had this break working files.
Example: you have a working file. Then you modify the character rig… while doing so you notice a key bone-roll is off, so you change the bone-roll on the character’s rig. This can destroy the animations in every linked file.
Disk-space is relatively cheap, so I prefer to Append assets into each new .blend file that will use the asset. ==> Each .blend file becomes a historical record of what was working when I created that file.
If your hard-disk is really becoming full, you can always buy a thumb-drive and offload older projects en-masse. But it depends on your situation & resources. Linking is probably much more valuable for multi-person teams.


==> here differentiate between the word “Scene” as used by a script-writer (a set/location, and some actions by the characters), and a “blender-scene” as defined by the .blend file scenes controlled in the header top-right.

In general, I sort of end up with 1 .blend file per Storyboard-panel. If the storyboard is like a comic-book of the story, then the animated video is shooting each storyboard-panel with “a bit of motion” (character-animation/action, or a camera-pan across a room, or some trees swaying in the wind…).
Each storyboard-panel (shot sequence) will have an expected duration (seconds or frame-count).

In general, I develop major assets off in their own .blend files (with no animation). Major assets are characters, complex objects, and “movie-sets” (a room, a building, a landscape). Movie-sets might assembly a bunch of smaller objects.

When ready to animate the next shot, I Link|Append all the assets into a new .blend file. Keeping the overall frame-count (seconds duration) of this shot short, makes it a little easier to not trip up the actual animation keyframes.
Do the animation. Previz. Render frames to a folder for this shot/sequence.

If the next shot/sequence (storyboard-panel) is going to use all the same assets, I just copy this whole file for the next shot, and delete all the animation, then build the new animation.

Where I do use more “blender-scenes” in one file, is if the audio+action is continuous, but I want a change of camera… so for instance, characters are in a bar talking & drinking (planning their next heist?), and main camera is watching. But then I need a close-up reaction-shot of 1 character’s face. The motion is continuous, so there’s no reason to stop animation and move into a new file. Inside this blender file, copy all (including animation) to a new “blender-scene” now with a different camera right in one character’s face.
… but now for command-line rendering, you have to specify which -S sceneName, and remember that this other scene only needs to render -f 120…150. More of a headache to keep these micro-details straight… hence the preference for 1 .blend per sequence.
I think there’s a way to animate switching the camera, but the details escape me.


With all these .blend files, and folders full of rendered frames floating around, it becomes essential to have a logical system of describing what belongs where.

Back to script & movie jargon: Act : Scene : Shot(storyboard-panel) : Take .
So working on “projectName”
I might end up with .blend files with names like
projectName.1.3.17.003.blend (Act 1, Scene 3, Storyboard-panel 17, Take 3.

The “Take” part is unnecessary… except I’ve been burned by corrupted .blend files (it happens), so I like to have lots of checkpoint files.
I made file projectName.1.3.17.blend, and something didn’t work, so I’m about to make a major change, so I like to save that file to a new name (projectName.1.3.17.000.blend) before continuing.
Generally the file without 3-digits at the end is “the newest”, and the check point files have the extra .000 .001 .002 ascending sequence before the .blend… but that’s just me.

I just suggested using a string of digits .1.3.17.002. to keep things straight… but these are not very informative. (They do keep things in order in a directory.) So sometimes it’s easier to have filenames like:
projectName.library.17.002.blend (for scene in the library, storyboard-panel 17, Take 2).
or
projectName.a1.s2.p17.blend
(find something that works for you, and at least be consistent in THIS project. You can change it up in the next project.)

When rendering output frames, append a 4-digit frame counter to these file names for each image file.
projectName.1.3.17.0248.png (.exr) …rendered-frame #0248
or
projectName.library.17.0248.png (.exr)

I know I should be using .exr, but there is so little support for viewing and manipulating these files… :frowning:


By the time you have 10,000 frames rendered… it behooves you to have a different .blend file for the Video-Sequencer. In this file you can add the audio, and all the image-sequences, and clip-transitions (fades, holds, etc). As the total frame-count grows, it’s easier to grow this file as you render each shot/sequence, tacking each next sequence on to the end.


directory layout is important to keep all this organized.
I generally start out with

projectName/
    audio/       # music & voice tracks
    blend/       # all .blend
    frames/      # rendered frame sequences
    stills/      # random still images & test-shots I want to keep
    story/       # script & storyboard
    video/       # moving pictures

and end up with a lot of sub-folders beneath this basic layout.


Cheers! :slight_smile:

3 Likes

Wow, thank you so much for all of this useful information! I will definitely be using some of your tips.

With your naming convention, do you keep all of the .blend files in a single folder just to keep everything in one place?

Another thing I’m not quite sure on is previz. I am planning on using Workbench. Do you work out your simple previz shot in the same .blend file that you’re going to animate the final scene? Or is there a different process for it?

1 Like

I have a lot of the same questions you have, but I keep all my project files in the same place so…

1 Like

Nope, that would quickly become a nightmare, hence a folder structure mostly based around the general stages of film production.

Imagine a single folder with .blend files for each asset model (including various versions), along with layout, animation, simulation, general library files, shot files, render files, etc. Even for a short animation, you would have over a thousand blender files. Now find just one file with the final layout for Sequence 2, Shot 5 that needs to be linked into the Render file…

Then consider all of the image texture files and any other support files that relate or are directly used by one or more of those blender files… where you going to put those.

Seriously, you need to think about folder/file organsiation first before anything else or before you know it, you will be spending half your time trying to find the right file. Or worse, spend all night rendering a shot, only to then discover it’s using an old version of the character model or a previous version of a image texture that had an error or something.

Nope, totally different files. Check out my Animation Production video, but basically any previz (likely also called layout) is in it’s own file. For the most part the only animation that is sorted out there (and it sort of has to be) is movement of the camera.
It’s then all linked in to the animation file, since you animate to the camera, so you need to know what the camera sees in each shot first. And then the process continues from there.

Now sure you can make some adjustments based on the type of project/animation and what works for you (assuming it is all just you working on it), but the overall general workflow was pretty much established by Disney almost 100 years ago, especially by the time Snow White was made. While computers and 3D have changed the tools somewhat, the basic process isn’t all that different.

Disney even have a site that shows a fairly complete overall of how a feature is now made: https://disneyanimation.com/process/

2 Likes

@Maxdoucette

Maybe I’m misunderstanding “previz”?
If by previz you mean checking the motion/animation & that the camera is keeping everything in frame, I definitely do this in the same file that will provide the final render of this shot/sequence. … just before the final render of this file, I change the render-engine to Workbench, and render the sequence, assemble the video, and review for errors. For lighting, I use the same file, with the final render-engine (Cycles for me) to check lighting, materials, transparency, fog effects (stuff that Workbench can’t handle). I just sample 1 or 10 random frames and render them as stills to preview the final “look”. When these 2 steps are complete, I final render this complete sequence.


projectName/
    audio/       # music & voice tracks
    blend/       # all .blend
    frames/      # rendered frame sequences
    stills/      # random still images & test-shots I want to keep
    story/       # script & storyboard
    video/       # moving pictures

This layout is just the topmost view of just one project. This is to keep the file-types and software used somewhat organized.

Depending on the size of the project, there may be many sub-folders in these, or none. I just have this (empty) folder layout pre-existing. When I get “the next great idea ™”, I copy this empty folder structure and rename it “projectName”. /story/ will contain just text files (scripts), and for me, .jpg photos of my pencil-and-paper storyboard. (I still haven’t found a software solution that I like for storyboard development… so pencil-and-paper → photo .jpg it remains).

/frames/ will have a sub-folder for each shot/sequence. Hypothetically, these are disposable after assembly to video, so this localizes the frame image files. If I were really pressed for disk space, this is one-stop-shopping for things to delete. When assembling in the video-sequencer, this gives me an obvious place & structure to go find the next image-sequence to import.

/audio/ can have sub-folders for dialog, music, sound-effects, mixes. If using voice-actors, I tend to number each “line” to be spoken/recorded. (projectName.library.17.67this_book_is_overdue.wav (.mp3/.ogg) – for the library-scene, storyboard-panel 17, line 67 of the whole script + short reminder of what’s being said).

/blend/ can have sub-folders for actors/characters, assets/objects, sets/locations etc… and depending on the project size, mine tend to have sub-folders for each movie-scene. I try to use the numbering to keep things in order when viewing folder contents (so scene “library” might better be 03library in the naming – for the 3rd movie-scene)… but I don’t really want to look at more than 10-20 filenames when I list a directory. However it helps you to keep things organized.

Re-usable characters actually get developed far away from this project structure in a folder named model/actors/ . But 1-off extra-characters needed only for this project might get developed here.

The video-sequencer assembly tends to get a name like projectName.assemble.blend. If I had to break this up by scene, I might add a sub-folder /assemble/ containing projectName.03library.assemble.blend and so on for the other movie-scenes … whatever you need to do to keep everything straight.

:slight_smile:

disclaimer: I have no “industry experience”. This is just my attempt as a lone human, at reining in the chaos of ever larger projects. :slight_smile:

1 Like

sorry, @thetony20, I wasn’t replying directly to you. My previous post was intended for @Maxdoucette… but the quoting + replying logic got the best of me.

Thank you for sharing your workflow ideas! :slight_smile:

I don’t think that’s what I would call previz. I consider per-visualization at most rough layout and borderline storyboard stage.

In many ways I think it is more storyboard, but that has always been considered static drawings/sketches or even small thumbnails of what the camera’s PoV is in any given line of text from the script. In other words, is it a close up or medium shot, can you see just one character or two or more. Are you viewing them both side on or looking over the shoulder of one, etc.

That’s all well and good for general films, but then they got more complex, fast action a mass of live action and CGI, etc. As such, and with directors wanting to explore and really see/get a better idea of how the shot will look (before all the expense of green screen filming and then all the CG work, etc), the basic storyboard wasn’t cutting it.

So they added previz, where usually within a 3D environment, they would drop in a bunch of ‘generic’ pre-made assets for characters and simple environments and then move the camera around, cut from one shot to another and do basic location movement of any main character or stuff blowing up, etc.

Basically a moving storyboard that the director can then sign off on and all other departments can dial in what will be filmed, what is physical effects during filming, what will be CGI, etc. Then based on all that, pretty much everything is then made and done fully as part of the whole production process.

About the only thing carried over then from the previz, is likely to be the camera settings and movement, since nailing down what the actual shot needs to be and see is the whole point of it, before going to the massive expense of then actually making everything that it sees.

At least that’s my take on previz.

2 Likes

I will add my 0.5 cents to the discussion.
You can have separate .blend files for separate shots, however the way I usually do it is by either creating different camera markers in the timeline or having a new scene within the same .blend file. I don’t have a linear workflow, many times I find myself adjusting materials / lights / meshes on the go and having different .blend files for shots containing the same elements would become a headache to sync together content-wise every time.
So most times I try to do what I can in the same file or scene.

1 Like

re: definition of “previz”

Thank you for the clarification. From your description, it sounds like this is an enhanced “blocking-out” stage. Aside from pre-visualizing for the Directors, is this the version they give to the voice actors?

The catch-22: the dialog timing sometimes sets the time-extents for the physical motion animation. But what do you show the voice-actors, so that they can understand and “feel” the story?

You’re correct then, that this would be a completely different .blend file. Thanks! :smiley:

Thanks for your insights and justification for using a single .blend file. More perspectives is great!

Another gotcha, with using a single .blend, even in the scenario you describe, are certain corners of blender where frame number constants are set (in Physics? In the legacy Particle System? Some baking actions.).

If you have a really long animation all in one file (say 10k frames, ~7 minutes), and then discover that you need a few more seconds at the 1 minute mark… you should be able to grab all the keyframes (in the Dopesheet) after the 1 minute mark and just slide them to the right, to make room for more animation… but if I’m not mistaken, the Dopesheet doesn’t reflect these frame number constants. :frowning:

For a motion-graphic music-video animation, I would probably use your approach. But too, I find blender becomes more sluggish as I increase the complexity of any single file. This isn’t blender’s fault, it’s just more to process. Smaller single purpose files keep the response crisper, IMHO.

What you say is true. In fact I only use this approach for smaller things, like a couple shots that take place in the same environment. Full fledged animations are best to split into parts for all good and different reasons. For example, I am working on a short movie. I have all major scenes of my short work independently from one another. But scenes that take place in one single environment / set (say the house one I have) will probably have some shots that can be done within the same file, for convenience. I have another scene, however, where the same environment / set is used but it’s in a different state (it’s a burnt down house now) so there’s very few assets I will reuse from the first scene, therefore no need to make this one in the same file as the previous one.

1 Like

If it’s a usual/pure 3D or even 2D animation, then I really don’t think there would be any previz at all. It’s much more for live action films, where what they are trying to do is pre-visualize the special effects (usually CGI stuff).

Blocking out is usually for the character animation, which would still be done as a first stage of said animation, after voice recording has been done.

What the actors would get is the script, concept art, storyboard and maybe a look at the animatic to get a feeling for the timing and how the lines should then likely be delivered. Once all recorded, it would be dropped into the animatic edit and the director would decide which take to use, what order for each line, etc and likely have some updated storyboards made, before looking each shot/sequence down.

Then, it goes to layout, animation, simulation, lighting, render (where very likely each of those stages would be a different .blend file).

2 Likes

I am kind of trying to skip storyboarding, because even though I know that it is very rough, I suck at drawing and don’t particularly want to go through the process. So it leaves me with either writing out all of my shots on paper, or doing previz/blocking. I think I’m leaning towards writing out shots, because the creative control part of shot design can stay loose.

Thus why I think previz on a short animation might not be necessary.

Interesting idea, for some reason doing this slipped my mind. When you do this, do you render all of your shots and then cut them together, or do you do it on a shot-by-shot basis?

It certainly is a very good attempt!

My advice, don’t skip it. How you do it is less important and as long as it works for you, then anything is just fine. But it’s a very important step to quickly (ideally its quick) workout the camera angle, FoV, point of focus, number of characters in the shot, important background/environment elements, critical character expressions, etc of any given shot. Writing it out just doesn’t work, it needs to be visual.
It is the first actual visual version of your story, then placed in an animatic (I use Davinci Resolve) with any scratch audio (usually just dialogue) and rendered to video and played back at 24fps.
It’s then based on that, which a whole lot of initial choices are made. Does it get the story point across, is the staging clear, does the continuity make sense, is the pacing right (are some shots too long, some too short), etc, etc.
I too suck at drawing, but I made myself do it even for my short animation test, just to work through the whole process, to see that it works. And while things did change a little through layout, animation and final edit, it still provided a valuable base framework to build the shot.

I’m thinking of solving my drawing issues later this year when I build a new PC and use my current one as a local AI box.
My plan is to then see if I can setup a local text to image AI that can spit out basic storyboards based on the script and maybe trained with concept art, etc of my characters.
I have no interesting in using AI for all the actual creative stuff, but if I already pretty much know what I want and just need it ‘drawn’ fast, I see no reason not to use AI for that.
At least that’s my current theory, won’t know how we it turns out till I start playing around with things.

1 Like

I was actually brainstorming how this could be done.