Japanese anime studios are quietly testing a new AI tool that can generate 200 in-between animation frames in under 20 minutes—without the legal landmines that have plagued publicly trained models. CrestLab's ANICRA platform, trained exclusively on artwork entrusted by participating studios, is being positioned not as a replacement for artists but as a production assistant to alleviate chronic staff shortages.

The demonstration shown to MBS Television puts a number on the efficiency gap: five prepared drawings produced 80 in-between frames in about 10 minutes. CrestLab says the system can generate roughly 200 such images in 10 to 20 minutes, compared with an average manual workload of around 30 minutes per frame. That's a reduction from about 100 hours of manual work to just minutes for a batch of 200 frames. For an industry where a single episode can require 4,000 to more than 10,000 drawings, the math is compelling.

But speed alone isn't new. What makes this story different is the copyright architecture. ANICRA is trained only on materials entrusted by anime studios—not on scraped public data, not on fan art, not on anything a studio didn't explicitly provide. That distinction is the fulcrum on which commercial adoption pivots. Studios cited by MBS said they remain cautious where the origin of model-training data is unclear. A generated image that resembles protected material can create a legal and distribution risk late in a project, when the cost of correcting or withdrawing a scene is highest. ANICRA's approach aims to make rights management workable by design.

What You Need to Know About ANICRA's Technical Claim

CrestLab's platform covers in-betweening, coloring, and finishing—the repetitive, labor-intensive steps that connect key poses into smooth motion. In a typical manual workflow, after a key animator draws the critical poses, second key animators clean up the drawings, and in-betweeners fill the gaps. An experienced in-betweener can take 30 minutes on average for a single frame; complex frames can take an hour. ANICRA's output, when prepped with five key drawings, yields around 20 frames per minute.

But the generated frames are not final. CrestLab says images that are ultimately released should still pass through human hands. The tool acts as a first draft or a filler for repetitive motion, leaving artists to handle emotional expression, correction, and creative polish. That's the practical compromise being tested: AI produces the bulk, humans retain control and quality responsibility.

Other AI tools are also entering the pipeline. MBS reported on systems that generate lip and eye movement from a character image and audio in about one minute, and tools that apply CG reference motion to a character's hair, clothing, and expressions with AI-drawn movement. Background styling tools can convert sample backgrounds into a preferred style in around a minute. One company claimed an 80% reduction in conventional process time for motion transfer.

The Staffing Paradox: Efficiency vs. Training Ground

The anime market has more than doubled in size over the past decade, with around 300 titles produced annually. Production companies can't hire enough animators to match the output. So, AI tools that reduce the manual burden on in-betweening and cleanup look like a necessary pressure valve.

But in-between and cleanup work have traditionally been entry points where junior animators learn timing, anatomy, consistency, and production discipline. Veteran animator Akiko Nakano, with 46 years of experience, told MBS that if fully AI-generated anime becomes possible, "the drawing department could disappear." She also acknowledged that humans may still add finishing touches and convey emotion. The fear is that removing too much low-level work may ease a short-term labor shortage while weakening the pipeline that develops future key animators and directors.

This isn't a theoretical worry. Some animators and production companies refused to be interviewed once they heard the topic was AI and anime, MBS reported. Views among fans and creators are sharply divided, and some show strong rejection of the technology. Merely being associated with AI can carry a risk of public backlash. As a result, some studios are believed to be using AI quietly even when following rules, because the current mood makes disclosure difficult.

What This Means for Creative Workflows on Windows

For Windows users and IT teams supporting animation studios, the ANICRA story is not about a consumer AI feature you can download off the Microsoft Store. It's about workflow governance in a specialized production environment. If your shop is considering AI-assisted anime production, here are the practical implications:

  • Dataset control is non-negotiable. You cannot use a model trained on unknown data. You need documented permissions and a chain of custody for every image in the training set. IT must isolate project storage and enforce strict access controls to prevent unauthorized data from leaking into training pipelines.
  • Review checkpoints are critical. Even with studio-trained models, output must pass through human review. Your workflow should include secondary artists reviewing AI-generated frames for consistency, copyright violations (however unlikely), and artistic quality before final compositing.
  • Cloud services are a red line. Do not send unreleased artwork or voice assets to a public cloud AI service. Even if the tool promises data deletion, the risk of exposure or model contamination is too high. On-premises processing or private cloud instances are essential.
  • Documentation and traceability. For commercial distribution, you may need to prove that no infringing material was used. Maintain logs of what data went into the model, what the model produced, and who reviewed it. This is not just a legal hedge; it's a requirement for distribution contracts and insurance.
  • Junior staff development plan. If AI takes over in-betweening, how will you train new key animators? Studios will need to create deliberate training tracks that compensate for the loss of entry-level drawing hours—perhaps through formal apprenticeship programs or simulation-based training.

These aren't hypotheticals. Studios that adopt AI without these controls risk not just legal liability but also the kind of public criticism that can sink a project. Recall the backlash when an anime-style work used OpenAI's video-generation app, which was then shut down. That incident galvanized the Japanese industry's caution.

How We Got Here: The Long Emergency of Anime Production

The current AI push isn't born from techno-optimism. It's a response to structural pain. A single cut (a continuous shot) of 1.5 seconds of finished footage can pass through key animators, animation directors, episode directors, second key animators, and in-betweeners before color and backgrounds are added. A single episode might require 4,000 drawings; some recent productions exceed 10,000. From planning to broadcast, a title can take several years, with more than 100 people involved.

Meanwhile, the talent pool isn't growing. The number of animators can't keep up with the 300-plus titles a year, so studios are "working aggressively to secure available artists," as one production supervisor put it. The math is simple: if each in-betweener can do about 16 frames in an 8-hour day (at 30 min each), a 4,000-frame episode would need 250 person-days just for in-betweening. For a series, the numbers explode.

Copyright fears are not academic either. The OpenAI anime incident, where a video-generation tool produced clips of popular characters, led to the service's termination. For a studio, using a tool with unclear training data means that a completed work might be unable to air or stream. The cost of re-doing scenes or shelving a project is existential.

So, the industry is exploring AI not out of a desire to cut jobs, but out of a need to keep production lines moving. The hope is that AI can handle the repetitive grunt work, freeing human artists for higher-value creative tasks. But that hope is tangled with fears of deskilling and job loss.

What to Do Now: A Practical Guide for Studios and IT

If you're a studio considering AI tools like ANICRA, here's a step-by-step assessment:

  1. Start with a legal audit. Before evaluating any AI tool, have your legal team or external counsel review the tool's training data provenance, data handling practices, and indemnification clauses. For ANICRA, the claim of studio-only training is strong, but you need to verify it contractually. Request a data processing agreement that specifies where your artwork will be stored, how it will be used for training, and whether it can be used to improve the model for others.
  2. Run a pilot with non-critical work. Choose a side project or a non-commercial sequence where the risk is low. Compare the time and quality of AI-assisted and manual paths. Measure not just output but also the correction time required by senior artists.
  3. Involve artists in the process, not just management. Transparency is essential. Bring animators into the decision-making early, explain how the tool will be used, and create an opt-in system. The backlash risk is real; forcing AI on a skeptical team could lead to leaks, morale collapse, and reputational damage.
  4. Build a review pipeline with clear sign-off stages. Every AI-generated frame should be tagged in your asset management system as AI-originated, include the prompt or input, the model version, and the reviewer's name. This creates an audit trail for distributors and streamers who may demand compliance evidence.
  5. Plan for skill development. If in-betweening work declines, create alternative pathways for junior artists. Consider rotating them through AI review and correction roles, or assigning them to special projects where they can develop key animation skills under mentorship. Document the impact on your talent pipeline so you can adjust hiring and training budgets.
  6. Prepare for public scrutiny. In the current environment, even legal AI use can trigger backlash. Have a communication plan ready that explains your ethical stance, the rights-clear nature of the tool, and the role of human artists in the final work. This isn't spin; it's about protecting the commercial viability of your project.

For IT teams:

  • Infrastructure: If the tool runs locally, ensure workstations have sufficient GPU memory and processing power. If it's cloud-based, use a private tenant with data residency guarantees.
  • Data segregation: Keep AI training data, prompts, outputs, and review logs in separate security groups. No one outside the project should have access.
  • Monitoring: Log all access to AI tools and flag any use of unauthorized models. Many public image generators are easily accessible; block them on studio networks to prevent accidental use that could contaminate your rights-clear pipeline.
  • Backups and versioning: When you use AI, you're introducing a non-deterministic element. Maintain strict project file versioning so you can revert if a batch of generated frames is later found to have issues.

Outlook: Wait-and-See Adoption with a Quiet Revolution Underway

Anime studios are likely to continue adopting AI tools quietly, behind closed doors, until the legal framework firms up and public sentiment softens. The Japanese government, which views anime as a strategic cultural export, may eventually provide guidelines or certifications for copyright-safe AI tools. That could accelerate adoption.

The real test will come when a major commercial production openly credits AI assistance. If audiences accept it and the work is commercially successful, the floodgates may open. If it sparks controversy and boycott campaigns, studios will retreat further into stealth mode.

For now, CrestLab's ANICRA represents a pragmatic middle path: a tool that respects copyright boundaries and positions itself as an assistant, not a replacement. Windows-based studios and IT shops that want to stay competitive should start evaluating such tools now, but with the disciplined governance described above. The technology is ready; the industry's culture and legal infrastructure are not. How they converge will shape the next decade of anime production.