Three videos, from installing the panel to a finished short film. Each one has chapters, so you can jump straight to the part you need.
A quick tour of the panel: the four tabs, results and upscaling, models and their settings, the cost estimate, your license and fal.ai key, spend tracking and the privacy options.
Now that Summon is installed, let's take a quick tour before we get into the detailed tutorials. In After Effects, open Window, Extensions, then Summon. If History opens with the panel, close it for now. You can resize Summon to suit a wide or narrow layout, or dock it into your workspace.
There are four main tabs: Image creates and edits images; Animate turns images into video; Edit Video works with existing clips; and Lipsync brings together lipsync and translation tools.
The layout follows the same basic flow. Your generated output appears in Result. Below it, Upscale lets you work with either that result or a layer from your composition.
Next come the inputs and model selection. The model determines what you need to provide. A text-to-image model may only need a prompt. An image-editing model needs a source image as well. Add the required inputs before generating.
Generation settings also change with the model: resolution, aspect ratio, video length, and other options. Check the estimated cost, write your prompt, and click Generate. The output appears back in Result.
Now let's open Settings. Under Account, you can check or activate your Summon license. To run generations, you'll also need a fal.ai account and an API key. Paste that key into the field here.
Spend shows your estimated usage for the current project and your overall total. You can also add an optional fal admin API key. Then click Refresh to retrieve account information, including your remaining balance. We'll cover the models and upscaling settings in separate tutorials.
For now, open Options. The Activity Panel setting controls how you browse your generated media. With Drawer selected, History slides in when you need it and stays out of the way when you don't. Open it to browse previous images, videos, and upscaled results. To reuse an image, drag it into the input, close the drawer, and write your next prompt. Switch to Docked, and History stays visible inside Summon.
This works best with a wider panel, so you can keep your results in view and drag them into your inputs. If you prefer a narrow panel that takes up less space, Drawer is usually the better fit. Back in Options, you can sort the model menus by name, price, or recently added models.
Under Privacy, you can choose to send anonymous usage statistics to help improve Summon. This is optional and off by default. It reports things like which models are used and whether generations succeed. It doesn't send your prompts, file, or project names or API keys. That's the overview. In the next tutorials, we'll explore each workflow and start creating with Summon.
Install Summon, create and connect your fal.ai API key, and finish the first-run setup.
Let's install Summon and get it ready to generate inside After Effects. Download the Summon package from the website or your aescripts account. The download includes the ZXP extension file and a user guide. Keep the guide handy if you need a reference later.
To install the extension, download the free ZXP and UXP installer from aescripts. Choose the version for your operating system, then open it. Close After Effects before installing Summon. Drag the Summon ZXP file into the installer. When installation is complete, Summon appears in the After Effects list. You can then reopen After Effects.
In After Effects, go to Window, Extensions, and choose Summon. The first time you open it, a short setup walks you through getting started. First, you'll need a fal.ai account and an API key. Summon uses that key to connect to the generation service. Follow the link in the welcome screen to open fal's website.
Sign in to your fal account. From the dashboard, choose Get an API key, or open Settings and find API keys under Developer. Click Add key, give it a recognizable name, and use the API scope shown here. Then click Create key.
Copy the new key and keep it somewhere secure. This is the one-time view of the full key. Treat it like a password. The example here is hidden for privacy. Return to Summon, paste the key into the field, and click Verify key. Once verification succeeds, continue to the next step.
Summon keeps generated results alongside your saved After Effects project, organized by media type. If the project hasn't been saved yet, results stay in a temporary folder until you save it. Then Summon moves them automatically into the project folder.
Next, review how generation costs are shown. Anonymous usage statistics are optional, but enabling them helps improve Summon. You can change that choice later in Settings. Read the terms of service and the usage requirements. To generate, accept them and click Start creating.
You're ready to go. In Settings, the Account tab shows your license or trial status and lets you manage your fal.ai key. The license options may look a little different depending on where you purchased Summon. That's the setup. In the next tutorials, we'll use Summon to turn an idea into a finished shot right inside After Effects.
One project from start to finish: a character reference, a first cut, then shot-by-shot fixes with frame capture, image editing, lipsync and video editing, all without leaving After Effects. It ends with reusing past generations and adding a model through the Model Manager.
In the film: “I need a new background.”
In the film: “For my desktop?”
In this tutorial, we'll build a short film with Summon inside After Effects. A man asks for a new desktop background, but summons a whole new world instead. We'll create a character, generate a first version, then refine individual shots with image editing, frame capture, lipsync, and video editing. I'm skipping generation waits so we can focus on the workflow.
Let's start in Image. With no reference image attached, choose a text-to-image model. I'm using Nano Banana Pro to create a character sheet: one person shown from several angles so we have a useful reference for later shots. Set the aspect ratio and resolution, then check the estimated cost. "Why this price?" explains where that estimate comes from. It may use provider pricing or information from your previous generations, so treat it as an estimate rather than a guaranteed final charge.
Start with a straightforward description: a photorealistic man in his mid-30s, wearing glasses, with views of his face and body. You can write in your own language and use EN to translate the prompt into English. Enhance expands that description into a more detailed prompt for the selected model. Read the result before generating, and adjust anything that changes your intention.
A clear character description is more useful than adding detail without a purpose. Click Generate. The activity drawer shows the job and its estimated cost, while you can continue working in After Effects. Here's a completed character sheet prepared for this demonstration. The result appears in the panel. Open Info to check the prompt, model, settings, and recorded cost. This result is 2K, and its recorded cost is 15 cents.
Let's use that result to demonstrate upscaling.
Below Result, choose local or cloud upscaling. I already have Topaz installed locally, so I'll use local and select This Result. Comp Layer is the alternative when the source is a layer in your composition. The upscale appears as another job in History. Once it finishes, use the images and upscale filters to distinguish the original from the enlarged version.
We'll use the upscaled character sheet as our reference for the film.
Now switch to Animate. The model determines the available inputs. Some models use a start frame, some accept start and end frames, and others work with reference images or video. Choose the input method that matches what you want to make. Here, the character sheet is a reference, not the first frame of the movie. I'm choosing the reference-based Seedance option so the model can use the character's appearance while composing the shots.
Add the character sheet from History. You can also drag an item into the reference input or load a file from your computer. From Playhead lets you capture the composition, and we'll use that shortly. I'll add the Summon logo as a second reference. When writing the prompt, click a reference thumbnail to insert its image tag. That makes it clear which image describes the character and which one supplies the logo.
I've prepared a prompt describing the short story and its shots. Check the reference tags, choose the aspect ratio, duration, and available output settings, and enable sound for this example. Then generate the video. This gives us a first version to build on. We don't need every detail to be perfect yet. Let's watch it, then improve the shots that need attention.
In the film: “I need a new background.”
In the film: “For my desktop?”
Use Add to Project to import the video.
Summon organizes generated files in an AI Generated folder beside your saved project, with separate folders for each media type. To find a result on disk, choose Reveal in Explorer. If you started with an unsaved project, your results first go into a temporary folder. As soon as you save the project, Summon automatically moves those results into the saved project folder and updates their links in After Effects.
Your work stays together without moving files by hand. Create a composition from the video. For this walkthrough, I'm using a quick draft and then refining it. I want a front-facing shot for the opening dialogue, and later, we'll correct the character's hand position.
Move the playhead to the frame you want to use, switch to Image, and clear the previous references. Choose From Playhead. The image you see in the composition becomes a reference for the next generation. I'll add the character sheet as a second reference and ask for a front-facing camera angle. Use the captured frame for the scene and the character sheet for his appearance.
Enhance can help clean up the wording, but always check that the instruction still says what you mean. Here's the new image. The framing is wider than I want, so I'll add it to the timeline and adjust its scale and position in After Effects. There's no need to generate another image just to choose a tighter crop. This is where capture is especially useful.
The next step can use the framing I've actually built in the composition rather than the untouched source image.
Let's turn that image into the opening dialogue. In Lipsync, choose the workflow first. The menu includes options for video with an audio track, a talking photo, and speech translation. Each option requires different inputs. We have a still image, so choose a talking photo model. For the photo input, select From Playhead to capture our reframed close-up. Next, we need the audio the character will speak.
I've prepared the dialogue in ElevenLabs and imported it into After Effects. The file contains both lines, but we only need the first one here. Solo the dialogue track, then set the work area's beginning and end around that line. Under Choose Audio, use Render This Comp Audio to create an audio input from that range. You can also load an audio file directly.
Check the photo, audio, model, and estimated cost, then generate the lipsync.
In the film: “I need a new background.”
Here's the talking photo result. It's a useful starting point for this demo, even though the performance could be refined further. Add it to the timeline and place it over the part of the original film where we want the character to speak directly to camera. Check the timing against the surrounding shots and mute any overlapping dialogue.
We want one clear voice, with the new close-up doing a specific job in the edit.
Next, there's a visual problem. The computer reacts as if the mouse was clicked, but the man's hand isn't on it. We can repair that shot instead of regenerating the entire film. Find the relevant shot in the timeline and mark its range. Before editing the moving footage, we'll create a reference image that shows the hand in the position we want.
Return to Image, clear the previous references, and capture this frame. Ask the image model to change the man's hand position so it rests on the computer mouse, while keeping the rest of the scene consistent. Here's the corrected image. This gives the video editing model a visual target rather than relying only on a written description of the change.
For this example, I render the selected shot and import that short file back into After Effects. This gives us a video input containing exactly the part we want to edit instead of the entire original movie. In Edit Video, select the imported clip and choose From Comp Layer. You can also choose a file or a result from History. Check that the video input is the short shot, then attach the corrected hand position image as a reference.
Choose a video editing model that supports the inputs you've provided. Here, I'm leaving duration on auto, so it follows the supplied shot. The prompt asks the model to match the man's hand position to the reference image. Use EN if you need to clean up or translate the wording. Review the settings and cost, then run Edit Video.
We're changing one action within a shot rather than rebuilding the whole sequence.
The edited shot is back. The hand begins on the mouse, then moves as the character looks around. Add the result to the timeline, fit it to the composition, and align it with the part it's replacing. Keep an eye on audio as well as picture. For this edit, I bring back the original shot's sound under the replacement image.
Watch the transition in context and make sure you haven't doubled the dialogue or lost the sound from the surrounding scene.
The final line still has the original generated voice. To keep the character consistent, we'll replace it with the second line from our prepared dialogue. Choose the frame you want and capture it as the photo input in Lipsync. Select only the closing line in the audio track. Before rendering the composition audio, solo that track or mute every other sound source.
Otherwise, the lipsync input may include unwanted audio from the original movie. Render the marked audio range, choose the talking photo model, and generate the closing performance. This time, I'm using the OmniHuman option.
In the film: “For my desktop?”
Here's the corrected closing line. If you've moved the playhead since making the capture, History can help you find your place again. Open the Captured filter and inspect the frame number. In this example, it's frame 324. Return to that frame in the timeline, select the generated result, and use Add to Timeline. Fit the new shot to the composition and remove the overlapping original dialogue.
The new performance is a little longer, so adjust the ending to let the complete line play. Keep the final consonants and a small natural pause. Then, watch the film from the beginning to check the flow between the new shots. You may still want to match color and framing across the generated shots. Upscaling is also available for images and video, locally with Topaz or through the cloud options.
The finished example shown in this tutorial has had a separate upscale pass.
To revisit a generation, right-click its item in History and choose Reuse. Summon restores the relevant workflow, prompt, and inputs, so you can change one part of the setup instead of rebuilding it from scratch. The same idea applies to the other results. We can return to an image edit or a lipsync setup, check the references, and make another version while keeping everything in the After Effects project.
New AI models arrive every day. Summon's Model Manager makes it easy to keep up. Add the supported models you want and remove the ones you don't without updating or reinstalling Summon. Click Manage beside the model menu or open Settings and choose Models. You can see your installed models and search the catalog from fal. Let's search for Qwen and add an image editing model.
Once it's added, it appears in the model menu and is ready to use. I'll give it a reference image and ask it to change the man into a woman. Then generate, just as with the other models. You can also remove models you don't use. That keeps the menus focused on your workflow instead of filling them with every available option. Here's the edited result.
The key idea is simple. Your model list can evolve as new models appear, while you keep working in the same Summon panel.
That's the complete workflow. Create a character, generate a film, capture frames from the timeline, refine images, replace dialogue, and repair individual shots. Summon brings those steps together inside After Effects. If you need help, contact support@summonfx.com. For a technical issue, open Settings, go to Options, and save a problem report to include with your message. Thanks for watching. Here's the finished film.
In the film: “I need a new background.”
In the film: “For my desktop?”
Questions? Write to the Summon team at support@summonfx.com.