Skip to content
AI video & film

AI video series with recurring characters: step‑by‑step guide

Learn how to create an AI video series that keeps the same characters and locations across episodes using Carpathron Studio’s built‑in tools and workflow.

Team planning an AI video series with recurring characters in a studio
Illustration generated with AI.

Creators often struggle to keep the same people, locations and visual style when they turn a multi‑episode story into AI‑generated videos.

The direct answer is to use Carpathron Studio’s built‑in character and place libraries, combined with its “Next episode” feature, to carry forward every element from one film to the next.

How Carpathron Studio stores characters and places

Unified library for reusable assets

Carpathron Studio contains a built‑in library that holds every character and location you create during a project. When you add a new entry, you supply a textual description, optional reference pictures, and a unique name. The application saves this data in a local database on your PC, so the same definitions are instantly available in any subsequent episode you start.

Defining a character

To add a character, open the Characters tab and click “New”. Enter the name (e.g., “Alex the Engineer”), write a concise description (age, attire, personality traits), and attach one or more reference images. The images are stored alongside the description and are used by the AI video models to keep the visual appearance consistent across shots. Once saved, the character appears in the dropdown selector for every new shot, and the AI will automatically reuse the same face, clothing style, and voice settings.

Defining a place

Places follow the same workflow. In the Locations section, create an entry such as “Rooftop Laboratory”. Provide a short narrative (size, lighting, key props) and upload reference pictures of the setting. The library tags the location with these assets, allowing the AI to generate shots that start from the last frame of the previous scene or from a fresh viewpoint while preserving continuity.

Linking characters and places to shots

When you compose a shot list, you select the desired character and location from the library menus. Carpathron Studio inserts the selected assets into the prompt sent to the video model, guaranteeing that the generated image matches the stored reference. If a shot is later re‑rendered because the speech checker (Whisper) detected a mismatch, the same library entries are reused, so the corrected shot remains visually aligned with the rest of the episode.

Saving and reusing across episodes

The library is saved with the project file but can also be exported as a separate JSON file via the “Export Library” command. Importing that file into a new project instantly populates the character and location lists, making it trivial to start a second episode with the same cast and settings. This feature is especially useful for series production, where continuity is a core requirement.

Technical requirements

The library operates locally and requires no internet connection for storage. However, when you render using cloud models from fal.ai, the same character and place definitions are sent as part of the rendering request, ensuring the cloud‑based AI respects your saved assets. Minimum hardware includes a 64‑bit Windows 10/11 system with at least 8 GB of GPU memory and roughly 40 GB of free disk space for model files.

For a complete overview of Carpathron Studio’s capabilities, visit the product page.

Preparing the story outline for a multi‑episode series

Define the master narrative

Start by summarising the whole series in a single paragraph. Identify the central conflict, the ultimate resolution, and the thematic arc that will guide every episode. Write this synopsis on a plain text document or in a simple note‑taking app; the goal is a clear, reference‑able statement that you can revisit while drafting individual scripts.

Next, list the major plot milestones that must occur across the series. For a five‑episode arc, you might note: introduction of the protagonist, inciting incident, first major setback, climax, and final resolution. These milestones become the backbone for each episode’s outline.

Break the master narrative into episode outlines

Allocate the milestones to specific episodes. Create a one‑page outline for each episode that includes:

  • Episode title (optional but helpful for tracking)
  • Primary goal of the episode (what the protagonist tries to achieve)
  • Key turning points (e.g., a reveal, a chase, a decision)
  • End‑of‑episode hook that leads into the next part

Keep each episode outline to three to five bullet points. This brevity forces you to focus on essential actions and prevents unnecessary filler that could waste rendering time in Carpathron Studio.

Map characters and places to episodes

Open a table and list every recurring character and location you plan to use. For each entry, add columns for the episodes (1‑5) and mark with an “X” where the character or place appears. This visual matrix makes it easy to spot gaps, duplicate appearances, or accidental omissions.

ElementEpisode 1Episode 2Episode 3Episode 4Episode 5
Alex (protagonist)XXXXX
Dr. Lumen (mentor)XXX
Neon Market (city square)XXX
Underground LabXXX

When a character or place recurs, note any new visual details you want to keep consistent (e.g., a scar, a specific outfit, a landmark). Carpathron Studio will reuse these descriptions and reference pictures automatically if they are saved in its library.

Write episode‑level scripts

With the outlines and matrix ready, expand each episode into a short script. Use a simple format: scene heading, action description, dialogue line, and any specific camera direction you want the AI to consider. Keep sentences concise; the AI video model performs best with clear, unambiguous instructions.

For recurring scenes, copy the previous script block and adjust only the necessary details. This practice reduces the risk of inconsistencies and speeds up the “Guided” mode in Carpathron Studio, where you approve each step before rendering.

Tip: Export your character‑place matrix as a CSV file and import it into Carpathron Studio’s library feature. The app can then auto‑populate character descriptions, ensuring visual continuity across episodes.

By following these steps—master synopsis, episode outlines, a character/place matrix, and concise scripts—you create a solid blueprint that translates directly into a multi‑episode AI video series. The structured approach minimizes re‑work and lets Carpathron Studio focus on rendering high‑quality shots rather than fixing narrative gaps.

Step‑by‑step workflow to create the first episode

Carpathron Studio lets you turn a written story into a full‑length film in a handful of clicks. The following procedure shows how to launch a new project, attach the characters you have already saved, write the script, select the rendering mode, generate the shots and finally export the episode. The free tier permits up to three complete films per day, so the first episode fits comfortably within that limit.

  1. Create a new project. Open Carpathron Studio, click New Project, and give the episode a descriptive name such as “Episode 1 – The Arrival”. The dialog automatically creates a folder on your disk where all assets and renders will be stored.
  2. Select a film preset. In the preset dropdown choose a genre that matches your story, for example Science Fiction or Comedy. The preset loads a default shot structure, music style and color palette, which speeds up the later steps.
  3. Add saved characters and places. Switch to the Library pane, click Add Character and pick the avatars you previously defined in the character library. Do the same for locations by selecting the saved places. Because the app stores descriptions and reference pictures, each subsequent shot will keep the visual consistency of the cast and settings.
  4. Write the episode script. In the Script editor type the dialogue and action for the episode. Carpathron Studio parses the text, extracts spoken lines and identifies where each character appears. You can also paste a pre‑written outline; the software will automatically generate the corresponding shot list.
  5. Choose rendering mode. Press the Mode button and select either Auto to let the AI produce the whole film in one pass, or Guided to approve each shot before it is rendered. For a first episode the Auto mode is usually fastest, while Guided gives you fine‑grained control over tricky scenes.
  6. Render the shots. Click Render. The app selects the appropriate local model (e.g., Wan 2.2) if your PC meets the NVIDIA 8 GB requirement, otherwise it offers cloud models from fal.ai. Before each paid cloud render the price is shown; the free tier will use local models as long as the disk space (≈40 GB) is available.
  7. Review and re‑render if needed. After the initial pass, the built‑in Whisper speech checker listens to the generated audio. If a line does not match the script, the shot is flagged and automatically re‑rendered. In Guided mode you can also manually trigger a re‑render for any shot.
  8. Export the episode. When all shots are approved, press Export and choose the desired format (MP4, WebM, MOV or GIF). The export dialog also lets you add subtitles. Save the file to your project folder and you now have a ready‑to‑publish first episode.

All steps are performed inside Carpathron Studio; no external tools are required. For more details on the free tier limits, see the product page where the three‑films‑per‑day restriction is described.

Reusing characters and locations in the second episode

Open the first‑episode project

Launch Carpathron Studio and use the File → Open command to load the project file that contains the completed first episode. The interface shows the saved library of characters and places on the right‑hand pane. Because the library is stored inside the project, all entries appear automatically when the file is opened.

Start a new episode with “Next episode”

Click the Next episode button located above the storyboard area. The application creates a fresh timeline while copying every character and location entry from the previous episode into the new one. You will see the same names, reference images, and descriptive text in the character list, confirming that the continuity data has been transferred.

Verify and adjust the reused assets

Scroll through the character list and check that each entry shows the expected portrait and description. The same applies to places: the location cards retain their background pictures and textual tags. If a minor visual tweak is needed—such as updating a costume color—edit the description field; the AI director will incorporate the change when generating the next shot.

Edit only the new dialogue

With the reused assets locked, switch to the Dialogue tab. Here you type the lines for the new scene while the existing cast and setting remain unchanged. The AI director, either Claude or Meta Muse, reads the updated script and produces a fresh shot list that references the previously stored characters and places. If you have your own Claude or Meta Muse API key, paste it in the AI Director Settings panel to use your quota and custom rules; otherwise the built‑in rules are applied automatically.

Render the second episode

Press Render in Auto mode to let the AI generate the full sequence, or choose Guided mode to approve each shot before it is finalized. Because the character and location data are already cached, the rendering step focuses on the new dialogue and any altered actions, reducing overall processing time.

Tip: After confirming the reused characters, click the small refresh icon next to the library pane. This forces Carpathron Studio to re‑validate the reference images, preventing mismatches when the AI director selects visual details for the new shots.

When the render finishes, you can export the episode as MP4, WebM, or MOV using the Export dialog. The exported file can be appended to the first episode for a continuous series, or kept separate for individual publishing.

For more details on how the AI director integrates with your own API key, see the Carpathron Studio product page. If you need assistance with multi‑episode workflow, the contact page offers direct support from the Qubitpage team.

Comparison of Carpathron Studio plans for series production

When producing an AI video series with recurring characters, the choice of subscription determines how many episodes you can render each day and on how many machines you can work simultaneously. All three plans give access to the full feature set of Carpathron Studio, including the character library, shot‑by‑shot quality checking and cloud rendering via fal.ai. The main differences lie in daily rendering limits, the number of PCs that can be licensed, and the pricing structure.

The Free tier allows unlimited use of every function and model, but limits you to three complete films per day. This restriction applies to episodes as well, so a three‑episode series can be finished in a single day, but a longer season will need to be spread over multiple days. The Pro plan removes the daily cap, lets you run the software on up to three PCs, and costs €3 per month or €19.99 per year. The Max plan adds unlimited PCs, which is useful for a studio environment where several editors work on different episodes in parallel; it is priced at €5 per month or €49.99 per year.

Below is a concise comparison that highlights the limits most relevant to series production. For a deeper look at each plan, see the Carpathron Studio product page.

Plan Daily rendering limit Maximum PCs Cost (monthly / yearly)
Free 3 complete films (episodes) per day 1 Free
Pro Unlimited Up to 3 PCs €3 / month or €19.99 / year
Max Unlimited Unlimited PCs €5 / month or €49.99 / year
Cloud rendering cost Shown before each paid render (e.g., $0.72 for a 3‑shot film with LTX‑2.3 Fast) Same across plans Pay‑as‑you‑go to fal.ai provider
Support channel Email support (contact@qubitpage.com) for all tiers Same Same

For a ten‑episode series, the Free plan would require at least four days of rendering, assuming you stay within the three‑episode daily ceiling. The Pro plan lets you finish the entire season in a single day if you have sufficient hardware, while the Max plan enables a team of editors to work on multiple episodes concurrently, reducing total production time dramatically.

Choosing the right tier depends on your workflow size. Solo creators often find the Free plan sufficient for short series, whereas small studios benefit from the Pro plan’s multi‑PC allowance. Large production teams should consider Max to avoid bottlenecks and to keep all editors synchronized across episodes.

Exporting and publishing the whole series

Carpathron Studio can finish a series with a single click by exporting every episode in the same format. The export dialog lets you choose MP4, WebM, MOV, GIF or a subtitle file (SRT). Select the desired container, set the resolution (the app uses the resolution of the rendered shots), and start the batch job. The software writes each file to the folder you specify, adding a numeric suffix so that Episode01.mp4, Episode02.mp4 and so on appear in chronological order.

Batch export workflow

1. Open the Project Manager view and mark all episodes you want to publish. 2. Click the Export button, then choose Batch export. 3. In the format selector, pick MP4 for universal compatibility, or WebM if you target browsers that prefer open codecs. 4. Tick the Include subtitles checkbox to generate an SRT file alongside each video; the subtitles are taken from the script that Carpathron Studio created for each shot. 5. Press Start. The progress bar shows the rendering of any missing shots and the subsequent encoding of each episode.

Choosing the right container

  • MP4 (H.264) – works on YouTube, Vimeo, Facebook and most mobile devices. Use this when you need a balance of quality and file size.
  • WebM (VP9) – ideal for embedding on websites that favour open formats; the files are slightly larger but avoid licensing fees.
  • MOV – preserves the original codec settings from the AI render; keep this for archival copies or further editing in professional tools.
  • GIF – useful for short teaser loops (up to 10 seconds) that can be shared on social media without a video player.
  • Subtitles (SRT) – add them to any container to improve accessibility and SEO; platforms like YouTube can auto‑sync them during upload.

Tips for uploading to video platforms

Before you upload, rename each file to include the episode number and a short title (e.g., 01_The_Mysterious_Artifact.mp4). Most platforms accept batch uploads, so you can drag the whole series folder into the upload window. Set the video description to contain a brief series overview and a list of episode titles; this helps viewers navigate the playlist.

Enable closed captions by attaching the SRT file during the upload step. On YouTube, select “Add subtitles” and choose the language that matches the Piper voice you used (English, Romanian, Italian, Spanish, French or German). For platforms that do not support external subtitle files, embed the subtitles directly into the MP4 using a tool such as ffmpeg before upload.

Finally, check the thumbnail for each episode. Carpathron Studio can export the first frame of any shot as an image; use that frame as a custom thumbnail to maintain visual consistency across the series. If you need assistance with batch thumbnail creation or platform‑specific settings, contact our support team at /contact.

For a complete reference on export settings and batch processing, see the Carpathron Studio product page.

Frequently asked questions

How do I keep a character’s appearance identical across episodes?

Carpathron Studio stores each character’s description and reference pictures in a persistent library. When you start a new episode, select the same character entry and the app automatically re‑uses the saved visual attributes for every shot. The system also carries the last rendered picture forward, so the next shot can begin from the exact visual state of the previous episode.

Can I change a character’s outfit in later episodes without losing continuity?

Yes. In the character editor you can add alternate outfits as separate reference images. When you want a wardrobe change, choose the new outfit for the specific shot while keeping the original character ID unchanged. The software will still treat the figure as the same character, preserving facial features and voice consistency throughout the series.

What is the difference between Auto and Guided mode for a series?

Auto mode generates the entire episode in one pass: the story director creates the cast, locations, shot list and renders each segment without user interruption. Guided mode pauses after every major step—character placement, location selection, shot rendering—allowing you to approve or edit the output before the next stage proceeds. This gives fine‑grained control over continuity while still leveraging the same AI models.

How many episodes can I render with the free version of Carpathron Studio?

The free tier permits three complete films per day, and each film can consist of multiple episodes if you use the “Next episode” feature. Because the limit applies to full films rather than individual episodes, you can chain several short episodes together as long as the total count does not exceed three finished films in a 24‑hour period.

Is it possible to use my own AI key for the story director?

Carpathron Studio supports custom keys for the story director and quality checker agents, such as Claude or Meta Muse. By entering your own API key in the settings, the app will route the narrative generation and quality checks to your account instead of the built‑in rules. If no key is provided, the built‑in models are used automatically.

Wrap‑up: consistent AI video series made easy

By defining characters and locations once, then re‑using those definitions across every episode, you keep visual continuity without manual copy‑pasting. Store each persona in Carpathron Studio’s library, attach reference images, and let the AI fill in details for each new shot. When you reach the end of an episode, click “Next episode” to inherit the same cast and settings, then continue the story in Guided mode to approve each step and correct any mismatches instantly.

Leverage the built‑in Whisper speech checker after recording dialogue; the tool will flag mis‑read lines and automatically re‑render the affected shots, saving you time on manual dubbing. For higher‑quality renders, switch to a cloud model such as LTX‑2.3 Fast via your own fal.ai key, watching the cost preview before each paid render. Finally, export the completed series in MP4 or WebM, and keep subtitles synced for easy distribution.

Ready to turn your script into a polished, recurring‑character series? Explore Carpathron Studio and start building today.

Written by

Mircea S. Rusu

Founder and engineer of Qubitpage. Support engineer since 2005, CMS developer, former technical consultant at Niguarda Ca' Granda Hospital in Milan, and author of the book “The Architect”.

More about the author →