Digital cinema has never produced more data, and never been more fragile at the same time. As productions moved away from film, long-term preservation quietly devolved into a mix of hard drives, tapes, post-house goodwill, and the assumption that someone, somewhere, would still know where the master resides. Mnemonica Archive was built on the awareness that those inherited customs are doomed. 

TL;DR: Mnemonica Archive is a film preservation platform that treats “archiving” as an active, monitored process rather than storage. It organises titles with Versions and typed Packages (DCP, IMF, video, audio, documents, marketing, etc.), extracts and indexes technical metadata (including DCP/IMF data), creates and manages KDMs and pulls metadata via a TMDB integration. It adds controlled sharing via Screening Rooms and Deliveries, allowing rights holders or sales agents to screen, approve, and send packages without exposing masters via public links. The system runs on AWS in EU regions, uses encryption at rest and replication, and is built to keep masters findable, usable, and licensable over long timeframes – including compliance with security standards like ISO 27001, TPN Gold Shield, and Amazon Studios Tier 1 Certification.

Piero Costantini has worked in the Italian audiovisual sector since 2000 as a filmmaker and producer. His early career includes founding and operating Sputnik Media in Rome, a digital production and post-production facility that operated for more than 15 years. Between 2006 and 2008, he also served as Executive Producer at Bigberry. Since 2015, he has been based in Rome as CEO and Founder of Mnemonica. 

Stefano Diana is based in Rome and has held a range of professional roles across media, technology, and cultural institutions, collaborating on the first European telematics programs, participating in the birth of mobile internet, and consulting for many ICT companies since the end of the last century. He has taught creativity and advertising. Since 2017, he has been Chief Marketing Officer at Mnemonica. Alongside this, he works as an independent researcher and essayist in the field of socio-technical studies, and as a freelance expert and author on communication. Diana studied Computer Science at La Sapienza University of Rome.

DP: Before we dive into the Archive, what is Mnemonica itself?

Stefano Diana: Mnemonica was founded in Rome; it’s 100% Italian and proudly European. We have just celebrated our tenth anniversary here. There are 15 of us at this point. Also, Piero’s tiny little dog Nina has also recently joined us.

Most of our developers are based in Calabria at our secondary unit, where the quality of life and cost of living, along with one of the most advanced Computer Engineering faculties in the EU, make it the ideal setting to play our game in the glowing sun of southern Italy. 

As a vertical SaaS provider for the film industry, we now have a long track record and a strong base of loyal clients. All the major Italian production companies use our services, including masters like Paolo Sorrentino, major productions like Those About to Die, and amazing indies like Die My Love. Servers are specifically placed in EU territory, within the AWS infrastructure (at the moment). Data is encrypted at rest and replicated in accordance with disaster recovery best practices. 

DP: And what is Mnemonica Archive?

Stefano Diana: We actually started building it ten years ago! If you think about the company name, it’s apparent: ‘Mnemonica’ is a Greek word that refers to the art of remembering things, and we named the company that way because our mission from the outset was to do something to give a new memory to the film industry in its digital form, still a bit fragmented. 

However, it has been a long journey. First, we focused on short-term memory and production workflows to set up a well-functioning, secure, and reliable product that would be used and help us gain market recognition. Having accomplished that, now it is time for long-term memory. Of course, the coincidence with the maturation of faster internet and robust cloud infrastructures is also obvious.

DP: Mnemonica Archive is presented as a professional preservation infrastructure, not just storage. What distinction are you making there?

Stefano Diana: Digital technology has blurred many distinctions that were clear-cut in the analogue world. It has made mainstream the idea that common digital storage is synonymous with memory, given that the media are called ‘digital memory’, and that anyone can ‘save’ whatever they want in them, as they have been doing with PCs for many years now, without taking any particular precautions. 

Of course, now and then, someone loses data, but these are just stupid oversights, those unexpected events that can always happen in life. These low-level practices have given rise to a naïve, carefree concept of preservation that is most popular with producers and has swept away the meticulous methods used to preserve celluloid. But digital preservation is a politically serious and technologically complicated matter, a profession in its own right. Nothing to do with a pile of LTOs stashed away somewhere.

Piero Costantini: In the past, in the days of film, analogue data was handled by large companies such as Technicolor, by technicians in white coats, in fireproof environments with controlled temperature and humidity. Copies were regenerated on contact, first passing through a QC process performed manually by other technicians. And so the memory was safe.

With the switch to digital, it was thought that buying a hard drive in a shop would be enough to ensure its eternity. This misconception, based on immediate savings, has led to the current situation in which basements are full of hard drives and LTO tapes with perhaps something inside, far from the market and impossible to manage. A real problem whose metastases are spreading more and more with each passing day.

And the more we pretend nothing is wrong, the more the problem spreads, as we know. It is like men who, on the threshold of fifty, partly out of fear and partly out of a sense of infallibility, avoid having health screenings. Sadly, sometimes it is too late. The point is this: data consistency checks are carried out on a sample basis, only when necessary. But only by plucking up the courage can we truly understand the situation. Part of our mission is to facilitate this awareness.

DP: When we talk about digital preservation, is this still on the horizon, or are we already seeing it today?

Stefano Diana: Well, from our frequent contact with producers, it is apparent that their “gold standard of storage” is a hard disk in the basement with an (optional) post-it note on it. This is “homemade” storage, without an IT department, rigorous processes or even a somewhat tangentially proven method. 

Otherwise, there is the traditional, unagreed delegation to post-production labs, which keep complaining because they are overwhelmed by LTOs and hidden costs. In this scenario, we regularly receive reports of entire films lost, trailers that cannot be found, mutilated versions, and more. It is really happening. 

Piero Costantini: A hard drive has an average life span of seven years. The less you use it, the greater the risk of a shock while respinning it. LTOs are guaranteed to be readable for about ~25 years. IF. If you preserve them properly. If temperature and humidity-controlled environments are kept. If nothing goes wrong and no one overwrites or corrupts the metadata. If you remember where the database with the file locations is.  Additionally, no one is considering disaster recovery policies that recommend keeping two copies in locations at least 200 km apart. Is this happening? No. 

Yes, that problem is happening now, and the problem is that no one has realised it yet, except a few people who have lost unrecoverable data.

DP: In earlier conversations, you used the term “future heritage.” What do you mean by that?

Stefano Diana: ‘Future heritage’ is a somewhat paradoxical expression we coined to summarise the idea that we must commit today to laying the foundations and establishing the conditions for what will become tomorrow’s digital cultural heritage, but only if we allow it to exist. 

As you can imagine, this concept can extend well beyond cinema. In our case, we can imagine it as a place where digitally native cinema, appropriately preserved, converges with digitised analogue cinema, which is usually the domain of a different breed of operators who worship technology as much as art and have much to teach occasional storers. We can certainly count on future generations’ nostalgia for the cinema of the past, or at least their youth; there is no doubt about that. At least on this one, we can’t let them down.

DP: Storage costs money, but archives also contain value. Have you considered monetisation strategies for older films over time?

Stefano Diana: The consumption cycle of a work has become as frenetic as any other media process, but this is an autonomous mechanism that has nothing to do with the value of the works and the feelings that can bind them to their audience. 

On the other hand, OTT platforms enable the long tail of niche distribution that Chris Anderson first discussed in 2004. Now, accessible preservation platforms like Mnemonica Archive put this possibility in the hands of the rights owners themselves. Doing it yourself means owners can always keep their entire catalogue in view and welcome any new opportunity or proposal from interested parties at any time. It could be retrospectives, festivals, schools, private screenings, or anything else. Who can say when a match will happen? Any time is a good time; you’d better be ready. My girlfriend just watched Wild Strawberries, and she loved it.

Piero Costantini: “ Data is the new oil” is a mantra we have been hearing for some time. When data is secure and shareable, monetisation is a given. It never crossed our minds to be just another place to store data, a storage facility. There are storage solutions that are much cheaper than Mnemonica on a per-terabyte basis. But we have always had a vision of data, and we bring automation to industrial processes. Mnemonica allows rights holders to extract value from their digital assets. Let me give you a small but significant example.

When Leone Film Group was listed, its capitalisation established the value of its catalogue and, therefore, its potential to exploit rights. But without the underlying files (and without the underlying film reels), the rights are of little use. So much so that Leone Film Group’s digital assets were deposited in Wall Street data centres, allowing investors to bet their capital on something tangible, albeit digital.

DP: European film production relies heavily on public funding, tax credits, and subsidies. From your perspective, do European governments have a vested interest in long-term digital preservation?

Stefano Diana: Without a doubt. Italy, for example, provides food for thought in this regard. Subsidies are viewed with suspicion by the current government because the cultural value of cinema is somewhat “less” appreciated, and, from a purely economic perspective, funding seems akin to throwing money away on non-repayable loans.

The guarantee of the preservation and continued economic exploitation of subsidised works changes the scenario, as it serves as insurance for the investment, for both the producer and the state. By preventing loss and increasing the likelihood of future cash flow, it provides concrete funding prospects. This implies a radical rethinking of conservation: it should no longer be considered a separate issue and someone else’s business, but the final stage of production. A stage that potentially never ends.  

Piero Costantini: The funny thing is that a lot has been done to preserve our heritage. Not only governments but also foundations and public bodies have created an ecosystem for preserving historical heritage. In Italy, we have La Cineteca di Bologna, which is a shining and virtuous example of this. They have embraced the concept of preservation in its entirety.

Starting with research in public and private archives, and moving through the restoration of great masterpieces of the past by bringing together their experience and network of financiers, to the restoration of cinema venues such as the Modernissimo, considered by many Hollywood stars to be the most beautiful cinema in the world. They embody preservation as a value. But obviously, not everyone can afford such an apparatus.

And those who are children of native digital cinema do not have this level of infrastructure and funding. We believe the time has come to make preservation democratic, for the benefit of all.

DP: The French model (National Centre for Cinema and the Moving Image (CNC) and the National Library of France (BnF)) is cited as the current gold standard. Why does it work better than others?

Stefano Diana: When we went to the MIFC in Lyon last October and met with local players, we discovered an amazing reality. It was also remarkable because it aligned closely with what we had implicitly envisioned when conceiving and building Mnemonica Archive. 

French law takes the preservation of cinematographic works very seriously. First, it distinguishes between two types of conservation: cultural heritage and economic exploitation. They are regulated by two laws: one general and the other specific to cinema. While the CNC and the BnF oversee heritage, the management of continuous exploitation is handled in a surprisingly forward-thinking way: public funding is firmly tied to the preservation of subsidised works. 

How? Requiring producers to enter into a contract with a technical service provider to handle storage and access ensures asset preservation for at least five years. This cannot be done in-house except for very large companies. It is quite an ingenious way of combining culture and business, public and private interests. In the background is the recent recognition by the CST (Commission Supérieure Technique de l’image et du son) of the cloud as a reference infrastructure, due to its dual capacity for resilience and accessibility, which meets both objectives simultaneously. This is definitely the way to go, and we wrote a Manifesto to state it.

DP: In your manifesto, you argue that Italy has the intention but not the implementation. What is currently missing?

Stefano Diana: Well, I’d say that the intention is missing, too. As I said earlier, this is not exactly the cinema cult moment in Italy. However, we know we are not alone. If Rome is weeping, Berlin is not laughing… 

I don’t know what the situation is like where you live, but in Italy, there is a frightening technical backwardness among politicians, as well as among producers. It takes a great deal of effort to get their attention, but we keep trying. 

For example, we organised a Cybersecurity Day a while ago to raise awareness among producers about these types of issues, which are often far from their minds. Our aim is to engage our public decision-makers with the concept of protecting cultural heritage. Nowadays this topic might rank even higher than passion for cinema. If you want, here is the whole manifesto – it is drafted for the Italian scenario but we know that those are common problems.

DP: Based on your experience, should preservation be regulated nationally, or would an EU-level framework make more sense?

Stefano Diana: The EFM and us both felt the need to organise this panel precisely to discuss these issues at a supranational level. My view, based on the research I have conducted so far and the discussions we have had, is that many EU countries face similar dilemmas, and the problems brought about by the advent of digital technology, with its often overlooked fragility, are recognised across the EU. 

Also, the fact that the Creative Europe MEDIA programme awarded a grant to Mnemonica Archive for its innovative value for European audiovisual heritage might be read as a sign of Europe’s need for a policy on this issue.

Piero Costantini: When we won the Creative Europe Innovative Tools and Business Models call, the greatest joy came from seeing that such a topic had caught the attention of the Commission to the point that, for the first time in the history of the call, an Italian company was awarded the prize. 

As a strong Europeanist, I find this subject well-suited to EU-level standardisation. The technologies are the same for everyone; films are for everyone. But we Europeans are who we are because we are the children of our cultural heritage, and for this reason we cannot afford to lose our memory. 

Europe has taken the wrong approach on some issues: it has sought to standardise matters that were too specific or too closely linked to individual countries’ identities. Preserving cinema has nothing to do with this; we just need to decide to do it and put all countries in a position to do so. Our manifesto also calls on Europe to provide funding to countries to regenerate and preserve their digital film capitals.

DP: What will be the key questions or points of friction raised in the EFM panel discussion?

Stefano Diana: At the panel, we want to discuss the risk of invisibility, first of all. We may also call it oblivion, but what’s worse for a movie than invisibility? And it comes from a kind of charm, a collective illusion that makes people believe digital cinema is somehow immortal in itself, that their movies are safe just because they saved them somewhere on some device. 

Reality is, they are just one failed hard drive, one expired key, or one forgotten contract away from disappearing quietly and forever. And that incredible patrimony becomes increasingly hypothetical. This is the heart of the discussion.

Another big problem is responsibility. Everyone agrees that digital films are cultural heritage AND economic assets. And if you ask who is responsible for keeping them alive in ten or twenty years, every person in the room suddenly gets very interested in their shoes. Rights holders assume someone technical is taking care of it, technical partners assume the producer has a plan, and institutions assume the market will somehow solve it. Spoiler alert: it usually does not. All the boats are empty, as the Taoist tale goes. 

Then there’s a tension between policy and reality. Many national and EU-level frameworks are still rooted in an analogue mindset. They are very good at describing what should be deposited, but much less helpful at keeping digital masters usable, accessible, and economically exploitable over time.

At the Berlinale EFM, we will discuss how to move from symbolic compliance to actual preservation without turning producers into accidental archivists or IT managers, which does not work. 

Funding always comes up, of course. Preservation is often treated as a “noble but optional” expense, something to consider only after the film is finished and the budget is gone. The panel will push back on that idea quite strongly. Preservation is not a luxury; it is the condition that allows future exploitation, re-releases, restorations, education, and all the things that keep films alive beyond their first release window. We should at least consider it an insurance policy on future value.

That being said, it’s important to remember that no one in particular is to blame. The panel focuses on alignment, standards, and cooperation. How do we get production workflows, technical infrastructures, and public policy to point in the same direction? 

If we manage that, Europe’s digital cinema will have a future. If we do not, we will continue producing culture at an industrial scale, while quietly losing it in basements and forgotten accounts. That would be tragic. And also very avoidable. Join us at the panel – we really want to hear your ideas! 

(We’ll also host a networking event including Apero afterwards – let’s see what we can come up with together)

DP: Let’s move into how Mnemonica Archive actually works. Where does it sit in the film lifecycle?

Stefano Diana: Mnemonica Archive is the ideal complement to Mnemonica Production, the SaaS application we have commercialised so far. Of course, you can transfer to Mnemonica Archive an existing catalogue, or create it there from scratch. 

But our basic idea is a complete life cycle. When a producer finishes a movie or a season of a TV series within Mnemonica Production, they simply move it to this new side of the application, where storage costs less. You have different tools than those you need for production. 

There, you find a hierarchical structure of projects, which are collections of Works, which are collections of Versions, which are collections of packages. Versions of a Work can be different cuts, restorations, various assemblies of assets for a co-producer, vendor or distributor, or whatever you deem useful to be archived as a separately identified set. This flexibility is necessary to accommodate the various interpretations of ‘version’ or ‘variant’ that exist in the real world. 

Material, however, is not just thrown into a Version, but must be uploaded to packages denoted by types like DCP, IMF, video, audio, image sequence, documents, marketing, and so on. There’s also a special default package called Gallery that collects the screeners for that Version for sales purposes. 

The advantage of such packages is not only that they help you keep your assets organised, but also that they can be searched at the collection level and immediately have a list of Works you have, for example, with a given director, a given actor, and DCPs inside.

Metadata is the fundamental strength of our system, as it is of any archiving system, and we exploit it to the full. First of all, we have entered into an agreement and integration with TMDB, The Movie Database, the rich crowdsourced version of IMDB that powers services like Letterboxd, so when you create a new title, you can directly import all its metadata from the database, including the title in various languages, synopsis, cast, crew, poster, etc., and then use all this information for searches. 

The same applies when you upload DCP or IMF packages: their metadata is extracted automatically and becomes easily searchable. Additionally, with full KDM management on board, users will be able to decrypt and create a “screener” of the first five minutes to review the opening sequence, language, subtitles, and aspect ratio. 

This is not happening with any other SaaS because DCPs are not streamable and require proper software and hardware to play. In the long term, this is a game-changer. 

DP: What should a producer upload at a minimum, and what is optional but strongly recommended?

Stefano Diana: What will you archive in Versions and Packages? Here is one of our key design choices, which has been applied in Mnemonica Production since the beginning. 

We give our customers tools that enable them or even nudge them to follow best practices, in this case, for preservation. But we don’t impose anything except the minimum requirements. If you want, you can create a Work and put only an H.264 screener in it that is actually your wedding film; you can do so, if that’s what you’re happy with. 

The point is that there are responsibilities that users must take on, and technology cannot and should not replace them, at least because there is no way to control everything with technology. If anything, what we do is prevent practices that are known to be unsafe within our environment, e.g. sending links and passwords that lead directly to content. Returning to your question, people can upload whatever they think is appropriate; we don’t evaluate the merits of their format choices. We offer an environment that enables them to index their assets so they can easily find and understand their functions years from now. 

Piero Costantini: What Stefano has just described is absolutely crucial. The idea of a long-term archiving service is that you upload material today, and perhaps your children will access it in 10 or 20 years. 

The challenge is to make it clear to future generations how the data has been organised and why. This is not simply a stylistic exercise, but a design principle that is much closer to human beings than to technology. That is why we have scattered spaces throughout the platform where you can save information that is relevant now but will become invaluable in the future for reconstructing how and who contributed to the creation of the master.

DP: Is there already a way to handle licensing information inside Mnemonica Archive?

Stefano Diana: Not at the moment, but it is definitely a natural evolution that is possible, and we are thinking about it.

Piero Costantini: Today, on Mnemonica, you can save whatever you want. Do you have PDFs from the legal department that you want to keep close to your data? You can do that. But that doesn’t make me drop my jaw. We will definitely create views on rights in the future. Imagine a map of the world where, country by country, you can see when a particular right was transferred to whom.

DP: Screening rooms already exist in Mnemonica Production. How did they become part of Mnemonica Archive?

Stefano Diana: Another fundamental design principle is flexibility. This stems from designing code objects that strike the best balance between generality and efficiency. The main consequence is Mnemonica’s extreme versatility, which has been used with great success in workflows we did not anticipate at the outset; casting is an excellent example. 

Mnemonica’s screening rooms are private, active environments that host stakeholders for screenings and discussions, encode watermarks and alert their audience to new content, track each viewer’s views by the second, and more. Very few changes were needed to reuse them in Mnemonica Archive, where they are the heart of the system’s commercial side.  

Piero Costantini: Actually, Mnemonica Archive was the seed of everything. We spotted the physical support accumulation problem in 2014. But we were nobodies! No producer would ever have entrusted us with their masters. 

So we decided to cut our teeth on semi-finished materials. The challenge was no less complex. We were dealing with unpublished materials of great value and of great appeal to the “piracy market”. In Mnemonica Production, we handle “simpler” files than at Archive. Still, without the experience gained from the first product, especially in cybersecurity, stability, and platform reliability, the second would not have been possible. Our Screening Rooms are unique. They are so flexible that they still surprise me today with their ability to support an incredibly wide range of use cases, from dailies management to festival programming.

DP: Once the film and all surrounding material are in the archive, what happens next?

Stefano Diana: Long live the Works on Mnemonica Archive! The difference between a traditional archive and Mnemonica Archive lies precisely in this added life. 

The Works section is dedicated to the deposit and accurate indexing of masters, but then there is the other section with screening rooms and Deliveries, which is the living part, the part directly and constantly connected to the market.

We have deliberately introduced a role that did not exist in Mnemonica Production, which we call Power Guest and which corresponds to a sales agent in the real world: an operator who cannot modify the masters, but is independent in creating screening rooms, inviting potential customers, and putting together Deliveries of the agreed materials. At that point, all it takes is a manager’s OK, and the film is delivered, ready for a new opportunity to meet the audience. A life that is renewed indefinitely.

Piero Costantini: So far, and I believe for a little while longer, our strategy has always been not to overlap with the quality work done by post-production labs. To answer your question: no, Mnemonica does not support versioning. We can indeed create H264 screeners from ProRes or MXF, but we do not do deliveries (versions configured for a specific endpoint, such as Netflix, for example). 

That work continues to be done by post-production companies, and we support them by handling archiving and logistics. As for watermarks, our rooms offer two types: overlay (we display the connected user’s email address in real time) and burn-in, where we create a new version with a text watermark customised by the user.

DP: What kinds of access can rights holders grant?

Stefano Diana: As said, Mnemonica’s policy is to be a tool that streamlines operations and a connector between stakeholders. Mnemonica is a secure environment accessible only to registered users, with two-factor authentication as needed. 

Accounts are free, personal and lifelong, like in a social network. Every operation is tracked and associated with a user. Furthermore, data transfer is integrated with both Mnemonica Production (Boxes) and Mnemonica Archive (Deliveries). Once again, Mnemonica is flexible and there are many possible configurations, but you need to shift your thinking slightly. 

For example, if the Mnemonica Archive customer is the producer, they will have oversight of the catalogue and all materials. They can invite post-production labs only to the Works under their supervision to upload the new versions they realised. In contrast, the producer or their sales agents can invite their customers to the screenings rooms and set up Deliveries, which are then approved and sent by the producer. 

But our client may be the distributor, who will keep their catalogue in Archive, including DCPs, and will be able to use Deliveries to send DCPs to theatres, after which theatres will be able to generate the relevant KDMs on the spot. 

Or else, our client may be the post-production lab that will use Archive to dematerialise the mountains of LTOs that besiege them, and continue to manage the works on behalf of producers in a more lightweight fashion: in this case, it will be the lab that invites their various producer clients to manage the respective Works. Or there may be chains of all these entities. 

Piero Costantini: Chapter 1 of the cybersecurity manual states: ‘Links are dangerous.’ We do not work with links. Mnemonica is a closed community accessible only by invitation and only to registered users. Everything else is irresponsible, including password-protected links, magic links, or anything that is unencrypted and can be sniffed by malware. I understand that clicking a link can be convenient and easy, but we are no longer in 1998, and our lives are now 51% digital. 

If we do not treat our digital lives with the same security practices as we do our physical lives, we are being naive. Cybersecurity is the enemy of simplicity, but the design effort we have put in has transformed what is usually an obstacle to adoption into a control surface for Mnemonica sessions. Anyone can log in, even on an infected computer, without fear of their credentials being compromised. Our passwordless login is not only extremely secure, but also incredibly easy to use: all you need is an opposable thumb.

DP: Are there different types of playback or presentation, including public access?

Stefano Diana: Mnemonica was created precisely to protect filmmakers from the unhealthy practices that many still rely on: using consumer platforms such as Vimeo, YouTube, or Dropbox for professional-grade work.  We have seen with our own eyes how quickly a preview of a film or series, if left online, is found and attacked. It’s really frightening. Hence, our extreme and unabashed attention to security. Code security by design, cloud security, interaction security. 

I can’t tell you how many times we’ve been asked to provide magic links, as other platforms did, to simplify screening for people who didn’t want to waste time on even minimal registration, clearly completely disconnected from the incredible value of their own content we were helping them to protect. 

We’ve always said no, even at the cost of losing clients. Instead of creating holes that would easily let strangers in, we created easy ways out. One of these is integration with Vimeo: if you want to organise “easy” screenings, you can do so outside Mnemonica and easily export clips to your external service. 

DP: Archives assume that nothing lasts forever. Is there a structured export or takeout path?

Stefano Diana: Mnemonica Archive runs on AWS because it is currently the best cloud infrastructure in the world. But the service is designed to be cloud-agnostic and to connect easily to S3-compatible on-premises storage systems. The archive can also serve as white-label middleware via its API. 

Again, we want our clients to have peace of mind and be able to make the infrastructure choices they prefer. All this is aimed at ensuring reversibility, one of the four basic requirements of the CST (French Technical Commission) for preservation contracts (See above). This means the ability to easily regain access to one’s data upon termination of the contract for any reason. However, the digital preservation landscape is taking shape, and many aspects will need to be tested and evaluated in real-world business scenarios as they arise.

Piero Costantini: It can also be much easier. If you need to retrieve your data, log in to Mnemonica and download! Everything will be indexed as it is in the cloud. Our transfer manager Mnemonica Gate is a tiny tool capable of transferring huge archives with zero losses, even if your system crashes. Is the line back? It will resume from where it stopped.

DP: Let’s get technical. How is Mnemonica Archive built under the hood?

Stefano Diana: Over the years, it has become a very complex application, but there is nothing particularly exotic inside.  Python, Angular Material UI for the web app, and a lot of open-source libraries. The mobile app for iOS and Android was made with Flutter. By the way, I must say that our mobile app is excellent. Far from being a side gadget, it is rather a pillar of the platform. 

First, because you can do with it pretty much everything you do with the web app (have you ever made a KDM from the beach?), or finish what you were doing there, while you’re on the go. 

Second, because it’s a master key that combines security and simplicity: it lets you access the web app without credentials via passwordless login, authenticates you for multi-factor authentication, and controls all your sessions.

Piero Costantini: For the sake of professionalism, I don’t want to go into too much detail about the cloud infrastructure. However, we use S3 for storage and CloudFront for distribution. 

DP: Mnemonica relies on European-based cloud infrastructure. Why was that a strict requirement?

Stefano Diana: We use AWS for the reasons mentioned. Our provider is the Irish subsidiary, and we are limited to European redundancy regions, which ensures compliance with the GDPR. That being said, we would be more than happy to use a European cloud provider; we are passionately European, and we always proclaim ourselves “proudly based in the EU”. We had high hopes for the Gaia-X project, but it seems to have stalled indefinitely due to disagreements between France and Germany, and the usual American suspects are taking advantage of this. 

This is unfortunate because data sovereignty is a critical issue for digital cultural heritage and for deciding which infrastructure to entrust our precious audiovisual material to. Not to mention how susceptible it is at this moment, when Mr Trump’s total unpredictability seems to be leaving transatlantic trade relations to chance or sudden power grabs. We will bring the issue to the attention of the panel; it is important that Europe wakes up on this front, too. 

DP: What does Mnemonica integrate with in the production and post pipeline today?

Piero Costantini: Integration with production and post-production processes does not occur solely through systems’ ability to communicate with one another. Fortunately, humans and their craftsmanship are still involved in these processes. Every production is different, and modelling everything to ensure 100% machine repeatability is unrealistic.

When Mnemonica Production is chosen for a film or series, it is not only because it integrates with other systems, but also because it enables the people involved to perform their tasks with minimal effort while maintaining control over the processes. Mnemonica is not just a software solution, but a true partner for those who manage workflows.

In fact, every time we are involved, we work side by side with post supervisors and the DIT department to ensure they get the most out of the platform by simplifying their practices without disrupting them or requiring omniscient expertise.

But we are well aware that simplification also comes from automation. Given that the editorial departments we serve are primarily Avid-based, we are developing a Media Composer plug-in to enable our cloud and its contents to synchronise with editing timelines with as few clicks as possible. It is already possible to export the comment feed as talking markers in Media Composer, but this step will soon be integrated seamlessly. Then we’ll port the same plug-in to Resolve and finally to Premiere.

However, this all sounds a bit old and dusty; connecting systems via API is the way to go. But for a small company like ours, it would be too much effort to follow up with every single client on this. We are considering jumping ahead by a few decades. One idea we are considering is to create an MCP server that would allow AI systems to communicate directly with Mnemonica. Instead of developing our own bot, let other bots learn and manoeuvre our ecosystem. 

I believe the era of SaaS management dashboards is coming to an end, and we are already moving toward UI-less services that display data only through chatbot conversations. With a well-documented MCP server and an n8n project, you can accomplish migrations that were unthinkable just a few months ago. 

DP: When uploading material, which codecs, formats, and resolutions do you recommend?

Piero Costantini: Here, we must make a distinction between Mnemonica Production and Mnemonica Archive. The semi-finished products we manage with Production are mostly proxies and intermediates. ProRes, DNxHD, and H.264 are our screening rooms’ daily bread. Our internal file transfer unit, the Box, is format-agnostic and used for transferring OCFs or nested file sets of any kind. Mnemonica Archive, on the other hand, also supports complex formats such as DCPs and IMFs. 

We have developed an understanding of what a master is and which files to include. A future proof and market ready archive can consist of a DCP (encrypted or unencrypted), an IMF (but this is sadly still a little-used format, at least in our experience, with a great potential in the storage efficiency perspective), a Prores 4K HDR, a Prores HD SDR, all PCM audio stems, subtitles, H.264 screeners and related materials such as trailers, teasers, posters that can serve marketing purposes.

Going forward, you may consider archiving both audio and video editing sessions, along with all metadata. Finally, by adding a DCDM to the package, you can make your work even more future-proof. This can become as complex as you want it to be. That’s why it should be taken seriously.

OCFs could also be stored, but the cost would be high in proportion to their use, which is very rare. For this type of file, physical storage on LTO remains the most efficient, provided it follows disaster recovery protocols and includes the necessary regeneration for obsolescent media.

DP: How does Mnemonica handle large-scale ingestion for broadcasters or studios with extensive catalogues?

Piero Costantini: We have encountered both small and medium-sized companies with highly capable staff who can manage these processes, and others that require more guidance. Every archive is different, and so is migration. Usually, large archives have valuable knowledge and data, which is a strong starting point.

DP: You differentiate between hot and cold storage. How does that work in practice?

Piero Costantini: Currently, all Mnemonica data is stored in hot storage, meaning it is immediately available and backed up in cold storage in different data centres.

With version 3.1 coming mid-year, Archive users will be able to choose which packages to freeze. In practice, this means removing them from hot storage and transferring them to LTOs in the same data centre. Doing so will result in significant savings, but users will have to pay the price of waiting and the cost of retrieving. The waiting time depends on how much the customer is willing to pay to retrieve it; it can range from a few minutes to a few hours.

DP: Let’s talk pricing realities. Beyond storage, what costs should users expect when egressing or distributing archived content?

Piero Costantini: Our business model is to purchase cloud resources such as storage, computing power, and traffic, transform them with our software, and resell them as a service. Among the many types of resources we offer, we have identified the ones most understandable to our customers (storage, traffic, and encoding) and set prices that are always affordable, under any circumstances, for both small independents and major companies. 

Mnemonica Production uses a production budget-based bundle model that covers needs without incurring additional expenses. The bundles can be customised according to your needs, and we also offer flat enterprise packages. The model allows unlimited users, a unique feature of Mnemonica that gives customers flexibility and provides them with certainty regarding their expenses.

Mnemonica Archive, on the other hand, has a very low fixed monthly cost and, since size matters a lot in this case, a variable prepaid or pay-as-you-go storage component. Outbound traffic is sized by package, and economies of scale and pre-purchases are possible.

To give some figures – or not! – Regarding this last component, let’s say that we cost about a third of Aspera and half of Signiant. This is also because with us, you pay only for outbound traffic, not inbound.

DP: Pricing depends on scale; can you provide a rough example? For example, I have 10 feature films in my catalogue, each with 2 terabytes of data, including metadata and materials.

Piero Costantini: If we are speaking about Mnemonica Archive, we are around 6000€/year. Hot storage, unlimited users, double redundancy and 5 TB of deliveries included. 

But imagine this scenario. Let’s start with the premise that cloud data is most useful when shared, so there must be a strategic vision behind the cloud migration. That said, today, if you need to take a DCP from an LTO and deliver it to any endpoint, it typically takes 2-3 days and costs €500-800. Normally, if you have to do this, it’s because you’ve closed a deal, a sale. When a sale is involved, the buyer usually covers the logistics costs.

The great thing is that with Mnemonica, delivery is immediate and costs around €40. This means that if you want to offer your client a better price, say €400, with Mnemonica, you would have a margin of €360 that you could use to pay for storage, and you don’t need a complex process to retrieve your media.  The more deliveries you make, the less you pay Mnemonica. To the point that you could even go into credit. Isn’t that evil?

DP: Looking ahead, what are the next development steps for Mnemonica Archive?

Piero Costantini: As noted above, the development plan for Mnemonica Archive will increasingly focus on expanding tools for this particular market. We have had clear plans for many years regarding where to go. However, I do not want to give too many details about this yet. On a technological level, the world of AI has certainly been appealing to us for quite some time, and perhaps the first step will be to work on an MCP server.

AI will change Mnemonica mainly behind the scenes. Instead of a classic “marketing company” , we built an internal media unit and created extensive documentation over the years. We are now organising this knowledge as a shared, open resource for the whole team. That enables faster audits, quicker draft deliverables, and a more structured approach to documentation and support, with internal answers found more quickly and easily. 

This is NOT about AI generating our content. It is about making knowledge collective and usable. AI agents support processes, but core development remains firmly with our engineers. Even technical work, such as the documentation for Python development, can now be turned into clear narratives with no extra effort for developers.

DP: Is there an end-of-life strategy for Mnemonica Archive itself?

Piero Costantini: The good thing about operating in the data centres of a large cloud operator is that you don’t have to worry about the obsolescence of the media or the technology itself. They take care of that. As long as you keep paying, your data will remain there forever, guaranteed to 99.999999999998%.

DP: Far future question. Could you imagine a “Mnemonica marketplace” where cinemas or institutions rent access to a film for a single showing? Or can film aficionados access it if the specific show is not available anywhere else?

Piero Costantini: (Laughs) Did you break into our secret plans?! :O)

DP: And finally, looking even further ahead, how do you think about abandoned media and works entering the public domain?

Piero Costantini: Dreaming of a Mnemonica Foundation is no cost. That would be the place “where pies go when they die”.

The post Mnemonica sets out to fix digital cinema memory first appeared on DIGITAL PRODUCTION and was written by Bela Beier.

What is HDR, Really?

DP: Hi Michael! The box on my monitor has stickers for HDR10, HDR+ and Dolby Vision. Isn’t that actually all the same thing?
Michael Hackl: You’re addressing a topic that is often very confusing for users and consumers in the HDR ecosystems. Even if it seems like a manufacturer is trying to fill the box with marketing buzzwords, these various acronyms describe HDR formats that are indeed different, each with its own technical specifications and workflows in content creation. Let’s take a quick look:

HDR10 is a basic HDR format. It uses the ST.2084 (Perceptual Quantizer) EOTF (Electro-Optical Transfer Function) and predominantly a maximum brightness of 1000 nits. It offers a color depth of 10 bits and offers two static metadata values to adapt the image to the capabilities of the respective display. It’s an open standard and each manufacturer tends to apply their own adaptation.

HDR10+ is effectively an extension of HDR10 that uses dynamic metadata (created at encoding). This is used to enable automatic adjustments from scene to scene or frame to frame to adapt the content to the screen. It also uses 10-bit color depth and supports peak brightness of up to 4000 nits.

Dolby Vision is the most advanced of these formats. Well, everyone will expect me to say this, but Dolby Vision is so much more than “just” another HDR format. The fact that we offer creatives a true “unified-master” workflow and give them real control over how their content will look on devices is the crucial point to me.  It uses dynamic metadata that gets created by the colorist in the post-production process, supports a color depth of up to 12 bits and can handle brightness levels of up to 10,000 nits (although current professional displays usually reach their limits between 1000 to 4000 nits). Dolby Vision’s dynamic metadata allows for precise control over how content is displayed on a plethora of devices, providing a more accurate representation of the creator’s intent. There’s also a side benefit of getting a better SDR version too.

Dolby Vision in Practice

DP: Okay, so what exactly is Dolby Vision?
Michael Hackl: Dolby Vision is designed to meet the needs of content creators, studios and content owners while providing an exceptional viewing experience, whether at home, at the cinema or on the go. It is crucial for filmmakers that their movie is reproduced as faithfully as possible on different devices and in different environments. This is completely understandable given the immense time and effort that went into its creative design. Dolby Vision integrates seamlessly into the post-production workflow and therefore requires only minimal adjustments to the workflow.

In practice, the Dolby Vision process for “Movies and TV” today starts with color grading in HDR, which is usually completed in a color grading suite on a professional reference display with peak luminance levels of 1,000 nits (or even up to 4,000 nits for some productions). Once the HDR grading is finished and approved, the colorist then runs an automatic Dolby Vision analysis that generates dynamic shot- or frame-based metadata, analyzing minimum, maximum, and average luminance levels along with the color space used to describe the color volume of each shot. In essence, this metadata captures the exact “light and color recipe” of the content.

With this core metadata in hand, the colorist can then map the HDR grade to various lower luminance HDR targets, as well as all the way down to 100 nits. This 100 nits target serves as our HDR anchor for mapping and will also allow users to derive their SDR version. All these mappings can be previewed and fine-tuned using the Dolby Vision trim controls. This single color grading process allows for accurate translation to a wide range of output formats, from TV and mobile devices to theatrical releases in DCI and Dolby Cinema.

Dolby Vision offers enhanced creative control and the ability to ensure content is displayed as intended on a wide range of devices. This is critical because today there are so many different types of displays (with differing technologies and capabilities OLED, QD-OLED, LED/LCD, etc.). It is a comprehensive solution that extends from content creation to delivery to playback and is designed to provide a viewing experience that is as close as possible to the creator’s intentions. Dolby Vision enables the creation of a unified master that describes all necessary versions via metadata.

Why it’s Complicated

DP: You would think that these are solved problems in 2025 ….
Michael Hackl: When we were all watching movies on our old cathode ray tube TV in the darkened living room in the evening, we had less variance in this respect. Such a television set came relatively close to the picture characteristics of the CRT TV used for color grading back then. And a darkened living room comes closer to the ambient conditions in color grading than when we watch content outdoors on our smartphone today.

Virtually every display of every device today has different image characteristics in terms of contrast and color, and viewers use these devices in completely different environments. And these environments got much brighter. Therefore, content really has to be adapted to the respective environment and capability of the display.

In the context of post-production, this means providing tools and workflows that enable precise control over the image, efficient creation of multiple deliverables, and confidence that the final product will render accurately across a wide range of display devices and environments. Dolby Vision seamlessly bridges the gap between the creative process and the viewing experience, ensuring that every nuance of the image is preserved and presented as intended, regardless of the playback device or viewing environment.

DP: And you have to build that into each screen individually?
Michael Hackl: Dolby works with manufacturers and implements Dolby Vision through a combination of components in compatible devices – Displays require specific hardware capabilities including a panel that meets defined characteristics and a processing unit capable of decoding Dolby Vision content, and the integration of licensed software into the device’s firmware or operating system. We also work to characterize their display and tune the results – that’s key to understand. These devices must pass Dolby’s approval process to ensure they meet the required specifications for Dolby Vision playback, a manufacturer can’t just say “Yeah, that’ll do!”

DP: What do you do to ‘certify’ Vision in terms of image quality – in other words, what do you test screens and end devices for, what measurements / minimum requirements do they have to fulfil?
Michael Hackl: While we don’t review or approve each piece of content, as the choice of metadata is a creative decision, the approval process begins during the creation and analysis of the metadata by working closely with our partners. We work closely with them to ensure that the Dolby Vision metadata can be created according to our recommendations and delivers consistent results on all approved systems. This exact specification and verification is of fundamental importance in order to be able to think about a consistent playback at all.

To ensure this, we work with consumer device manufacturers to implement Dolby Vision in their devices. Here, too, we ensure that the metadata created in the creation process is processed correctly and thus leads to a correct visual result. This is a key differentiating factor for Dolby Vision. We ensure that both metadata creation and the respective application in the devices are implemented correctly and consistently for tone mapping. With other formats, there are significant deviations in the calculation of metadata using different tools, and implementation in the devices also varies across manufacturers.

DP: Oh, that’s why it’s considered expensive – it’s not even built into office screens ?
Michael Hackl: The fact that Dolby Vision is perceived as expensive may reflect the high-quality expectations associated with our brand. However, Dolby Vision is not a feature priced for the end customer. Our OEM customers license Dolby Vision for their systems and offer them at the prices they set. A quick look at popular online marketplaces shows that a TV with Dolby Vision does not have to be more expensive than a TV without Dolby Vision. The price of today’s TV sets, smartphones, etc. is determined by other components, not by Dolby Vision.

Dolby Vision is already included in post-production systems such as Resolve Studio and Baselight. Metadata creation, tone mapping, and rendering are therefore possible at no additional cost. If users wish to customize the metadata, i.e., use the Dolby Vision trim controls, they can obtain a license with no time limit and independent of the system via our online store.

Tools for Colorists and Post Facilities

DP: So in finishing and mastering I can already set up exactly everything for that?
Michael Hackl: : Dolby Vision functionality, in particular the creation of dynamic metadata and its adjustment (trimming), is available in common color grading systems, such as Resolve and Baselight, as well as in mastering applications such as Transkoder or Clipster. For others, there are simpler tools that create the metadata automatically like in CapCut, Luma Fusion, Final Cut Pro and others.

For the creative process, it is of course incredibly important that the industry provides a proper end-to-end HDR production pipeline. Many well-known manufacturers have recognized the importance of flexible HDR/SDR editing and have already implemented it. Let’s hope that others will follow.

DP: From an editorial or mastering standpoint: what new control enhancements can colorists, editors, or mastering engineers make use of in Dolby Vision 2 compared to earlier versions?
Michael Hackl: Over the last 10 years we’ve listened carefully to the feedback from creatives, colorists, and color scientists at the leading post facilities around the world on the tools and workflows they’d like to see added or changed in the existing Dolby Vision mastering process. As a result, we’ve continued to innovate with new improvements like moving from Content Mapping 2.9 (CM 2.9) which was launched in 2015 and featured a subset of the metadata we use today, to CM 4.0 in 2019. 

CM4.0 features enhanced trim controls and was further enhanced in 2023 with new Analysis tuning options that make the starting point even better, meaning the colorist can get to the trim metadata that delivers their vision even quicker than before. The goal for the next generation Dolby Vision content creation process is to build further on that progress, offering even easier, more flexible and compelling tools and workflow options that answer creatives’ need to best deliver their artistic intent in both HDR and SDR.

Studios will get the Dolby Vision unified master deliverable they desire – while not adding time over the current Dolby Vision workflow. While we can’t share specific details at this time our plan, working with our partners, is to deliver a comprehensive package of improvements.

DP: And, from a pipeline view, this should be predestined for using an IMF master..
Michael Hackl: IMF is certainly a perfect format for today’s multi-format deliveries. The Dolby Vision metadata can be transferred in an IMF package as part of the image track in the MXF container. In this case, the metadata is interleaved with the image content. Transfer as a separate ISXD track is also possible.

For the creation of theatrical deliveries, mastering systems can use the Dolby Vision metadata created and optionally modified in the color grading process to create corresponding DCPs, either for DCI (SDR) or Dolby Cinema (HDR) playback.

Multiple Screens, One Master

DP: But then in regards to color grading, it does get stuck – creating different versions for the screens – mobile, TV, PC, VR goggles, and what else is there?
Michael Hackl: Nowadays, we consume content on a wide variety of devices in different viewing environments. This circumstance may initially seem uncontrollable, since a corresponding grade would have to be created for each of these different use cases. However, this would not be efficient in many respects. Therefore, we use different technologies, both in creation and in playback control, which adapt the content created by the colorist to the end devices, taking into account the colorist’s decisions. Studios and content owners receive a singular master with Dolby Vision metadata, which then manages playback on different displays.

Dolby Vision is already implemented in literally all device categories today. These range from TVs, PCs, HMDs (head-mounted displays), such as Apple Vision Pro or Meta Quest, to tablets and mobile devices. And when it comes to theatrical distribution, we can use Dolby Vision metadata to create separate stand-alone DCPs for DCI (SDR) or Dolby Cinema (HDR). Colorists appreciate the consistency and efficiency they get mapping the HDR home grade with our tools to create theatrical deliverables.

DP: And what are these differnet technologies?
Michael Hackl: The basis for this is provided by our Dolby Vision display mapping, which can be found in TVs, laptops, tablets, and mobile devices. Simply put, this technology knows the image reproduction capabilities of each display device. As soon as a Dolby Vision bitstream is played on these devices, they switch to Dolby Vision mode and perceptually adapt the content to the respective display. For cinema deliveries, we can use metadata created in post-production to directly create discrete DCPs, either for DCI or Dolby Cinema.

VFX and Editing

DP: But at least in VFX / pure CGI it makes no difference?
Michael Hackl: We must distinguish here between HDR and Dolby Vision as an HDR format. In VFX, as in HDR color grading, we work completely without metadata at HDR reference level, because we work in reference environments across standardized displays.

At least in color grading, it is not necessary to adapt to different playback devices as the colorist or even multiple colorists work on reference displays that are identical in spec. In the field of VFX, artists today might already work with HDR-capable displays or at least conduct their VFX reviews on HDR reference displays. When processing, it is important not to lose any details so that the colorist, who integrates the final VFX shots into the color grading, has full control over contrasts, brightness and colors. There is still room for improvement in VFX when it comes to a user-friendly end-to-end preview workflow for HDR and SDR.

DP: And it’s not relevant in editing either?
Michael Hackl: Editing is certainly a crucial process in film production, and some manufacturers already offer basic support for editing in HDR. However, that is not everything. Many editors prefer to edit using the UI viewer rather than primarily using an external (HDR) display. To do this, the viewer must run in HDR while the other UI elements are displayed in SDR. It is also desirable to be able to switch the actual video output flexibly between HDR and SDR. This again requires dailies in HDR and SDR. But here, too, what we have already discussed for mastering applies. It is simply not efficient to create dailies in HDR and SDR and then manage the two formats in parallel during editing.

For the creative process, it is of course incredibly important that the industry provides a proper end-to-end HDR production pipeline. Many well-known manufacturers have recognized the importance of flexible HDR/SDR dailies and editing and have already implemented it. Let’s hope that others will follow.

DP: But if you’ve edited something and want to output it to a Dolby Vision screen, how does that work?
Michael Hackl: In the area of post-produced content, conforming, a so-called online edit, takes place after editing in a color grading system like Resolve, whereby the editing information is applied to the original camera data. Depending on the size of the project, the original camera footage may also have been edited directly so that the timeline and data can be used directly for color grading, which varies in scope depending on the project.

In the area of post-produced content, conforming, a so-called online, takes place after editing, whereby the editing information is applied to the original camera data. This conformed material is then transferred to colour grading.

DP: And what happens in the background?
Michael Hackl: The content is then colorgraded in HDR and then the Dolby Vision process described above takes place. As soon as this is complete, the image and metadata information is exported from the color correction system and handed over to the mastering team. The image data is exported in ST.2084 (PQ) in P3D65 or Rec.2020, e.g. as QuickTime ProRes or 16-bit Tiff.

The metadata information is exported in the form of an XML file. In the mastering process, further elements such as start and end panels are added, along with the audio, and all elements are combined into the respective contribution format or mezzanine format, such as a Dolby Vision IMF package.

DP: And if we look at what Dolby Vision could be in 2040: What will be on the to-do list there – in your personal opinion?
Michael Hackl: Ideally, neither the creator nor the viewer needs to worry about where and on which device the content will be displayed. Dolby Vision ensures that the content is displayed accurately and that the display meets the expectations of the creator and viewer. Users get the best visual experience everywhere.

Want to know what others are saying about Dolby? Here is a Playlist of directors and filmmakers about working with it, including some of our favourites, like Denis Villeneuve (Part One and Part Two), Tim Burton, James Cameron, Jennifer Kaytin Robinson, Pete Docter and many More!

Two very recent ones are “Das Kanu des Manitu”, the wildly popular satire and long awaited sequel to “Schuh des Manitu” by Bully Herbig.

And “Welcome Home Baby“, the Austrian Horror / Mystery Movie

Learn More

DP: That seems reasonably straightforward – but what if I have additional questions?
Michael Hackl:: If you want to learn more about Dolby Vision or learn the workflow from scratch, you can sign up for our free Dolby Vision training courses at learning.dolby.com. These will guide you step by step through the workflow. You can find a lot of additional background information about Dolby Vision on professionalsupport.dolby.com and you can chat with other users and the Dolby team on the Dolby Vision User Forum.

We also offer regular Dolby Vision Technical Webinars, which interested users can register for via the following link.

The post Dolby Vision, Dolby Vision 2, and Beyond! first appeared on DIGITAL PRODUCTION and was written by Bela Beier.