Models, Workflows & Recipes
The three building blocks that turn an entity into media -- what each is, how they relate, and how you create them.
Ordinary Animator turns a description into a finished image or clip using three building blocks, from smallest to largest:
- A model is a single AI model -- the thing that actually generates a picture, a video, or a voice. Examples: Flux Dev 1.0, Qwen Image Edit 2511 or SDXL 1.0 for images, Wan I2V 2.2 for video. Each model has its own look and its own hardware appetite, and its name says which one it is.
- A workflow is one ComfyUI workflow -- a node graph that performs a single generation operation, such as "make a background image" or "animate a still". A workflow uses one or more models.
- A recipe is an ordered series of operations that reaches a goal -- for example "make a background, place the character on it, then fix the lighting". Running a ComfyUI workflow is the most common kind of operation; other steps are built-in operations (crop an image, resize, merge audio). A recipe also remembers how to feed one step's output into the next, so the whole thing can be reused. A recipe belongs to a step of the production plan: each ready-made recipe says which thing the plan produces it makes (a character's reference images, a shot's clip, the music) and, where it matters, when it applies -- for example a recipe for a shot with one speaker. The one you choose for each step is recorded under a visual style (Sound's, for the whole show) and in a Recipes chosen list on the Scene or Shot properties page (each showing only its own). A recipe names the models it uses -- each step is a specific model -- and carries words that describe it (its keywords, such as "anime" or "fast") and the kinds of recording it needs from a shot (its inputs: a first frame, your own acting, your voice). The Visual styles page searches and filters by those. The idea behind a recipe is called a technique ("film actors, then use the footage to guide generation"); techniques are background reading, not something you set. The plan step lists the recipe; the style can list others and the assistant remembers it. When nothing is listed and one ready-made recipe fits the shot, that one runs (the built-in fallback). When more than one fits, the assistant asks you once and writes nothing until you say yes. A yes is remembered at the top of the shot's visual style's list unless you say this scene or this shot; a shot's own choice wins over its scene's, the scene's over the style's, and the style's over the one made for the whole show before styles held lists. A recipe knows which shots it is for: the Talking shot recipe for two voices is for shots with several speakers, and not for a silent shot, which takes the next recipe on the list that fits (or the ready-made default). A scene's or shot's own choice can still be for one kind of shot. A shot's kind is worked out from its dialogue; on the Shot properties page you can set it yourself. Remove a recipe (on Visual styles, or a row in a scene's or shot's Recipes chosen list) and the next recipe on the list, or the ready-made recipe, is used again. Where a shot has no recipe yet (for example several characters speaking at once), the assistant says so instead of guessing.
Put simply:
A recipe is a series of operations. A workflow is the most common operation. A workflow uses models.
Recipes are visible and editable again. A gallery's Recipe panel shows its recipe as a list of steps you can open, fill in and run one at a time or all together with Run all. A ready-made recipe is read-only until you press Edit a copy, which makes it a recipe of your own that you can add steps to, remove steps from, reorder and rename. There is no separate Recipes page. See The Recipe panel.
A recipe is best run a step at a time. Each ready-made step says what a good result looks like: when its output arrives, Quality guidance appears under the step. Check the result against it before you go on, because the next step spends its time and money on whatever this one made. If the result is not right, run the step again for a new take. Some recipes, such as making a new pose for a character, walk you through this in the guide bar and only go on when you say the result looks right.
Three more words you will meet
- AI Models -- the model catalogue, in your account menu: every model there is, what to read about each, and whether its files are on this machine. It belongs to you, not to a project.
- Visual style -- what a scene means visually ("Normal", "Dream scene"): for each kind of media (a media purpose), the recipes it recommends in order -- the first that fits the shot is used -- each with an art style for every model it uses. A scene or shot can be given one, or none. That is the per-project half of model choice. See Visual styles. Every built-in recipe carries plain instructions -- what it does step by step, what it needs and what to prepare -- shown on the Visual styles page.
- Library -- a collection of workflows you can copy into a project. Ordinary Animator provides one; your local ComfyUI and Comfy Cloud are others. Copying a workflow in is what makes it available -- much like turning on a model.
Everything in a library already runs
There is no such thing as a half-finished workflow waiting to be filled in. Every workflow in every library is complete and ready to run, which is why you can copy one in and try it immediately. When a workflow is nearly what you want -- the right idea but the wrong style model, say -- it is adapted into another working workflow rather than left with blanks.
Workflows are also kept deliberately small and single-purpose, one stage of the job each. That is what lets you see every intermediate result, keep the good ones, and re-run just the stage you are unhappy with instead of starting over.
Choosing your models is mostly a hardware decision
The main reason to choose models per project is what your hardware can run. When you generate locally, picking your models -- and their quantization (a compressed build of a model that uses less graphics memory in exchange for a little quality) -- is one of the biggest decisions, driven by your GPU's memory. A model that wants 24 GB will not load on a 12 GB card, but a quantized build of it might. Comfy Cloud keeps a large library of models available, so this hardware question matters most for local generation.
The second reason is a consistent look: reusing the same models across a project keeps characters and style recognisable from shot to shot.
Nothing is chosen up front, and nothing is filtered away. Every model the catalogue has an operation for is offered, wherever you are; what a project has already settled on simply comes first and is marked Ready. See Choosing the right model.
How each building block gets made
You rarely start from scratch, and whatever you make becomes reusable. From most common to rarest:
- Pick the operation you need, and its model, together. A gallery's Recipe panel lists every operation -- ComfyUI workflows and built-in operations alike -- and under each one the models that can do it. Choosing a model the first time asks which build; the project remembers, and the pair is Ready from then on. The assistant can help: tell it what you are making and whether you generate in the cloud or on your own computer, and it suggests models that fit.
- Browse and install from the AI Models catalogue in your account menu when you want to read about a model or get its files onto this machine.
- If it takes more than one operation, ask the assistant. Describe the result you want and it works through them on the gallery you are looking at, one at a time -- running each, then setting the next one up with what the last one produced.
- If an operation has no workflow, bring one in or adapt one. Copy a workflow from a library into your project, or start from one that is close and have it adapted to what you need.
Each step falls back to the next only when nothing suitable exists, so your project grows richer the more you work.
You can also start from the other end. Browsing workflows first is perfectly reasonable: find one that does what you want, and install whatever files it needs to run on your machine. Nothing has to be turned on first -- every model the catalogue has an operation for is already on offer.
Every one of those routes works by hand, and the assistant can do each of them for you. It will find what fits -- the Suggest button on a media gallery's guide bar asks it to look at what the gallery already has and your chosen models, then recommend the best next step, with the real settings each model wants. It will compose a recipe: describe the result you want and it builds the steps on the gallery you are looking at, wires each step to the one before, and fills in what it can (see Running operations for you). And it will adapt a workflow to run where you need it -- swapping a model for a smaller build, and the loader nodes that build needs, so a workflow that was too big for your graphics card can run locally, or one built on your machine can run on Comfy Cloud. It proposes the adapted workflow; the platform checks it before it is allowed to run.
Where the prompt comes from
When a workflow step needs a text prompt, you do not have to hand-write one for every model. From what you already captured about the shot -- who is in it, the location, the camera, the mood -- a prompt can be generated for you, phrased the way that step's model prefers (some models want short comma-separated tags, others want a flowing description).
You also never pick the model twice. In a recipe, the model belongs to the generating step: that step's workflow -- chosen from your AI Models -- determines which model runs. A prompt step that feeds it reads the model back through the recipe's wiring and defaults to the same one automatically, so the prompt is phrased for the model that will actually generate. You can still override the prompt step's model if you want a different phrasing.
The tailoring is not cosmetic. Where a step asks for its prompt to be rewritten, the text is put through the prompt conventions recorded for the model that step will actually run, so the same shot yields a short tag list for one model and a flowing description for another. When no model can be worked out from what the step is bound to, the prompt is left exactly as you wrote it rather than phrased for a guess.
The model knowledge behind the advice
The assistant's model suggestions and prompts are only as good as what it knows about each model: what look it produces, how much graphics memory it needs, and how it likes to be prompted. That knowledge is gathered per model, so a new model arrives as a new entry rather than a rebuild. See From Storyboard to Video for how this feeds per-shot advice.
Those notes ship with the platform and the assistant reads them when it recommends -- along with a verdict on whether each model fits the graphics card you told it about, so it can say "this one needs the cloud" instead of recommending something that will not load.
Where to go next
- The Media Gallery -- where operations are chosen, filled in and run
- Choosing the right model -- how the choice is made, and where it lands
- Workflows -- the project's library, and what it settled on per operation
- AI Models -- the model catalogue: what exists, what to read, what to install
- ComfyUI Integration -- how ComfyUI workflows actually run
- From Storyboard to Video -- how all this carries a shot to a clip