The possibility of cel-look pipeline and challenges to take advantage of the hand-drawn animation

(Polygon Pictures Inc. / Studio Phones)
This document has been prepared with participants at “Film Production Pipelines and Artist Techniques 7” in mind, and has been written after the seminar, on the basis of the things discussed, in order to consolidate issues raised and to take a retrospective view of the seminar.
translated by PPI Translation Team


At this seminar, a variety of initiatives at each studio related to Cel-look depiction were introduced, and elucidation was given on strategies or trial-and-error processes that pertain to such things as improvements to the difficulty level of addressing media diversification and the establishment of robust production systems that take overseas expansion into account.
Also, this seminar brought about a renewed awareness of the necessity of things like conforming to initiatives to improve information literacy in other fields, or of establishing internal studio infrastructure operations as a way of addressing globalized clients, taking into account the demands of modern diversified clients requiring coordination between studios.
In this document, we should like to provide an introduction to and summary of ideas for infrastructure architecture for the purpose of coordinating while also making full use of the production skills that are the strengths of each studio, beginning with an introduction of the trial-and-error challenges each studio currently faces regarding matte painting as used in Cel-look CG, a topic raised during the seminar; we hope that this document will provide support to internal innovation hereafter within participating studios.

■Media Diversification and Accompanying Changes

Expansion in multi media

We found no significant differences between the present circumstances that each studio described and the information in the documents provided in advance.
However, a large number of concrete challenges were identified by studios which had made reference within a given range to the preparatory documents regarding internal infrastructure architecture at studios unaccustomed to engineering, and we decided to prepare this post-seminar document encompassing these things.
Given the fact that each studio referred to the advance documents and came prepared with strategies, at the end of this document we have consolidated these strategies into an infrastructure architecture proposal which may serve as a draft proposal for further innovation taking into account the current situation of each studio, in an item called “One Pipeline and Infrastructure Suggested by the Seminar Session”.

Cel-look CG and matte painting

Building internal infrastructure in studio not familiar with engineering

■Incorporating Matte Painting into Cel-look CG (circumstances of introduction)

Judging from the talks given by a number of studios at this seminar, it has become clear that the use of matte painting in image production has become the predominant production approach in Cel-look-style CG, as in hand-drawn animation.
However, trials in adopting 3DCG as a way of raising productivity have been repeatedly performed at various studios from a number of different approaches, starting with exploration of automated settings which incorporate a level of automation for improving productivity in pre-production phases such as layout, and explanations were given of ways in which each studio is using high-level trial and error to make even greater use of 3DCG than anticipated, such as trials pertaining to strategies for rendering techniques related to implementation of 3DCG background assets, which are limited at present to partial implementation, or using machine learning for contour extrapolation or coloration from 3DCG AOV output, as well as explorations of variation generation techniques in the pre-production phase, not to mention generation of effective instructional data called “genzu” for giving drawing guidance to matte painting staff.
What is shared in the background circumstances of all these things can be said to be improvements in productivity that nevertheless maintain the merits of a matte painting workflow that has been painstakingly built up over a long period of time, and robust infrastructure architecture and the reliable creation of administration systems expected of studios which handle large-scale overseas projects; during the seminar, each studio posed question after question about the demands made on infrastructure operation for large-scale overseas projects, which are becoming increasingly demanding.

Key Points Including Infrastructure on Large-Scale Overseas Projects
 - Storage system scale and performance
 - Render farm scale (and presence or absence of GPUs, etc.)
 - Distributed processing method on render farm
 - Data availability and back-up system
 - Number of workstations, etc., and OS used
 - Classification of software used in each phase of production
 - Network connectivity within the studio (network configuration within intranet, etc.)
 - Internet connectivity (throughput to overseas destinations, etc.) and main system tools for pipeline
 - Pipeline system design
 - In-house tools for digital image creation
 - Coordination with production management
 - Check and review flow and approval process
 - Data format and data flow in each phase of production
 - General security policy at studio
 - Limitations and auditing of internet access
 - Limitations and auditing of asset data access
 - Security agreements when partnering or outsourcing
 - Security of outside data transmissions (encryption or protocols)
Wide-ranging check items including assessments from clients

■Incorporating Matte Painting into Cel-look CG (workflow issues)

In connection to partial 3DCG background asset production, mentioned in the prior section, explanations were given of various endeavors by each studio to explore amalgamation of hand-drawn matte painting with CG workflows.
It became decidedly apparent from the speeches given this time by each studio that, in continuing awareness of what makes matte painting special, exploration for new techniques of expression has emerged as one major factor in the reason for using hand-drawn 2D matte paintings in a workflow in which CG is the core component, and it seems to us that innovation in workflows was also strikingly apparent from the content of the speeches which aimed for the establishment of original techniques of expression at each studio not limited to improving productivity, including such things as continuing to explore the expressive possibilities unique to CG via its implementation, in addition to continuing to explore new hand-drawn styles as well via the implementation of visual styles unique to CG, all while endeavoring to attain a traditional hand-drawn cel anime style.

An example combining complex lighting, rain, a 3D character, and matte painting

The genzu for the above

© Gamon Sakurai, KODANSHA/AJIN Production Committee. All Rights Reserved.

Things such as unique lighting techniques for hand-drawn backgrounds, softer silhouettes, or free manipulation of perspective are expressive techniques in which CG still lags behind at present, but in recent years, a variety of creative approaches have been adopted in CG as well, and it seems as though there is an ongoing tendency toward innovation in a variety of forms, including in the pre-production phase, as each studio works to incorporate methods appropriate to its own needs.

■Innovations in Production Systems for Cel-look CG

It seems as though attempts have begun to build momentum at various studios toward implementing CG via things like general trial-and-error efforts by artists, including giving of instructions, progress management, or trials in the pre-production phase in each sort of production, on the basis of the objective to improve productivity and reduce loss in traditional production systems, which have placed heavy emphasis on manual labor.
As mentioned in a prior section, a kind of instructional data called “genzu” for purposes of giving drawing instructions to matte painting staff is necessary in all scenes making use of matte painting in production, but it appears that there is a current tendency toward reassessing most kinds of work which has heretofore been done manually, starting with strategies to provide this to matte painting staff beforehand in an effective manner. While there are strategies common to many studios, such as tools to support things like genzu creation, the high level of awareness at each studio regarding innovation including infrastructure architecture, meant to continue to raise the level of maturity of the studio hereafter, and not simply to optimize production, was perceptible during this seminar.

■Endeavors Toward New Expressive Methods in Cel-look CG

To the extent that studios already possess regularized and effective pipeline architecture for Cel-look CG and are conducting infrastructure operation to support it, and precisely because they have a grasp of the limits of their studio’s infrastructure capabilities, we felt that we saw a great number of initiatives and strategies toward achievable new production methods and expressive methods.
For example, implementation of things like new expressive methods are progressing to provide, not a solitary strategy, but rather uninterrupted production running through the overall workflow, such as strategies to support a camera move called a nodal pan (one form of 3D camera work) which were implemented in such a way as to strike a balance with a hand-drawn workflow, as it is necessary to do camera work within given limits; or things like experiments such as methods for optimizing a related operation during compositing called projection mapping as well; color management systems related to this; or furtherance of the procedural side of compositing work, etc.; and among the studios that participated, opinions were also shared to the effect that, as expected, this is a workflow that can be achieved precisely because the studio maintains its ability to construct and operate infrastructure internally.

The area in which matte painting is necessary as generated by the 3D nodal pan camera of the DCC tool

A version with instructions to matte artists painted over (genzu)

The matte painting created

Projection mapping of matte painting during compositing work

© Gamon Sakurai, KODANSHA/AJIN Production Committee. All Rights Reserved.

In actuality, it is unavoidable that, in the exploration of expressive methods in animation production, each studio tends to proceed in its own direction, given the individual differences between studios and the stylistic direction in which each studio aims to go.
Be that as it may, opinions were also raised regarding the importance of things such as partnerships taking the form of progressively integrating differences in expressive ability between studios, even as each studio pays respect to the expressive methods of other studios, and it seems as though a large number of people were of the opinion that, hereafter as well, each studio, by increasing its awareness of the issues pertaining to its own challenges and linkages, may also be able to consider aspects of further exploration of new expressive methods.

■Cel-look CG Expressive Methods and Coordination at Each Studio

Things like settings management or awareness of asset compatibility and of assets used over the long term become necessary in order to explore new expressive methods by integrating the differing expressive capabilities particular to multiple studios and to actualize the same.
On the basis of an awareness that production costs will rise, even if the artists at each studio attempt to work in tandem, in a situation in which documentation has not been generated for seamless integration or compatibility improvement for the various assets, beginning with shaders, that are dealt with when using procedurals, there was a great deal of discussion at the seminar of things like libraries or such formats as MaterialX or USD.
Put simply, USD is a library for management of scene graphs or asset data, while MaterialX is an open format for managing materials or look development; both are said to be OSS libraries developed with the goal of handling data in 3DCG. While a large number of questions were raised regarding implementing these things smoothly between each studio over the mid to long term, a unique issue also raised at this seminar was whether it is possible, for example, to use MaterialX to also manage materials designed for use with hand-drawing tools. In other words, although opportunities for the use of these systems seem at first glance not to be forthcoming in the hand-drawn anime field, given the interchange of assets between 2D and 3D, it seems likely that the development hereafter of new tools or functionalities in 2D may come into view; and during the seminar, opinions were discussed to the effect that, by increasing the affinities between both hand-drawn and CG, it may be stimulating to continually amalgamate the techniques of each, as each learns of the advantages of the other.
As for raising compatibility between 2D and 3D tools, agenda items were presented in various forms regarding usage within existing software as well, and ideas to make full use of USD and so forth when handling 3D asset data in drawing software, especially Photoshop, were also raised as topics during discussions at the seminar, in the sense of preserving compatibility with the existing drawing workflow and overall pipeline of asset data that is used as a data exchange format with the 3DCG side.

If anything, rich discussion took place during the seminar regarding aspects of sharing challenges, operational coordination, and assets used to attempt new expressive methods, in the sense of improving coordination not only for purposes of simply improving productivity.

・USD (Pixar)

・MaterialX (ILM)

The above chart shows an example, presupposing a standard pipeline, of usage between software and between studios, and if we proceed on the basis of trends in standardization worldwide, it can be assumed that, relatively speaking, linkages between studios will advance. Based on this, while studios exist which also implement destandardization in order to foster their own individuality, it seems best to strike a balance and to design pipeline architecture from the twin vantage points of individuality and cooperativeness.

■Cel-look CG Expressive Methods and Coordination at Each Studio (difficulty level of the infrastructure architecture dimension)

During the seminar, it seemed as though all participants were surprised at how far the level of awareness of these problems at animation production studios exceeded expectations, but on the other hand, it also felt as though several studios were emphasizing the size of the challenges of raising levels of coordination at each studio, as mentioned above.
One of these, according to an opinion presented, is that, without improvement in the strictness of management systems for rights and patents, it is not possible to respond to increasingly globalized production demands at present, and, as a high level of information literacy is demanded not only of engineers but also of many artists, even for studios that are not accustomed to coordinating engineering with studios capable of an infrastructure architecture that supports large-scale production, without a more robust infrastructure architecture and the establishment of an operational system as a studio, it is feared that there will be a marked reduction in freedom of action on the production side as a result of such coordination.

Actually, even after the seminar, a number of studios have proactively brought up issues related to innovation in such infrastructure architecture or production systems, and as moderators, in order to be able to provide at least some kind of response, we decided that we ought to devise a strategy of sorts when creating post-seminar written materials hereafter. Also, we have discovered that more of the participating studios than we had anticipated are basing meaningful internal innovation on the related written materials, and are wondering whether we might investigate ways to slightly increase the scope of what is addressed in written materials hereafter; this and other things seem to have arisen as evident challenges for the moderators.

Expansion of check scope of infrastructure management

■IImproving Affinity between Real-Time Technology and the Cel-look Style

Interest in real-time processing is broad at the studios, and there were a great number of developments out of stimuli received in the game and game engine fields, starting with proactive use of game engines in the production and pre-production phases, as well as with investigations into applying fluid simulation mainly based on the GPU.
Also, at the venue, due to the participation of persons involved in the current development of game engines, advice was given on strategies for each studio based on deep technical expertise, and an explanation of algorithms related to levels of freedom for things like contour detection was also given by the designer at a studio which maintains its own in-house renderer for use in video production, which made it possible at this seminar to form a strong impression that this is a period of promulgation of real-time processing in terms of future developments in the field of video production and for traditional pipeline architectures which have mainly depended on batch processing.

■Improving Affinity between Real-Time Technology and the Cel-look Style (regarding use cases)

Although the cel-look style has heretofore been developed mainly in the context of video production, recently, as a result of diversification in media expansion, there has been a widening expansion in VR content beginning with console games and mobile games, and iconically represented by VTubers. Although measures are being developed at many studios to be able to apply expressive technologies heretofore cultivated in hand-drawn anime within video production in the form of Cel-look CG, in recent years the fields of gaming and VR have also seen an increase in activity in the application of these expressive technologies.
For things like game or VR content, while it is the case that content is developed using real-time CG technology best known through game engines, etc., there are increasingly more cases of technological development in which art direction or expressive techniques earlier developed in video production are being applied to real time, and during the seminar, opinions were shared that cases are actually increasing in which video production studios engage in real time content production such as games or VR.
While there are many differences in the techniques and algorithms used in video production and things like games, due to the differences between each in the machines used for calculation or in the calculation resources that can be handled, and the general practice at present is to produce each as a separate form of content, there are nevertheless great advantages to being able to share asset data or to share expressive techniques themselves, so that the need to increase affinities between these things going forward will be felt more and more strongly.

Example of background production using UE4

© BANDAI NAMCO Entertainment Inc.

In the presentations by each studio, opinions were promulgated from the various vantage points of the conditions at each studio due to these circumstances, but, in the opinion of the moderators, one thing that can be said in general is that this is due to the fact that this is an implementation period for continuing development on the server side of a better fusion of real-time processing and batch processing.

■Improving Affinity between Real-Time Technology and the Cel-look Style (viewed from the pre-production stage onward)

One attraction of real-time processing is interactivity and response speed when previewing. We felt that one side of things which came to the fore was the fact that the greatest significance of making use of these things in video production is in aiming for improvement in iteration number during video production, thus building an environment whose purpose is to allow artists to explore their own ideas more smoothly, or to achieve high levels of exploration in overall pre-production.
While it is true that, on the business side as well, it is anticipated that content production systems geared even more toward multiple forms of media expansion will be in greater demand hereafter, on the video production side, there are trends toward exploring ways of developing video production techniques, which have been cultivated mainly with traditional batch processing, in order to make use instead of real time processing more effectively, and it seems as though this is an important awareness to have in terms of the individuality of studios working to expand their business. In other words, along with demand for production techniques which allow similar resultant forms of expression between video production (which expends rendering calculation costs in order to generate imagery) and game development (where real-time rendering is conducted within fixed limits on calculation resources) it seems possible to say that, for both sides, the exploration of new expressive techniques has become more important.

However, the present situation is that it is no simple matter to incorporate production techniques used within traditional video production into content production that relies mainly on real-time technology. To take styles, depictions or forms of direction that have been fostered within video production and actualize them using real-time technology presents numerous technological challenges, important points in asset production differ, and methods of asset management and concepts behind them also differ at present.
The seminar this time was fortunate to have participants from both video production and games, and certain challenges in planning the next seminar were also discussed after the seminar by person planning to attend the next one, including speculation about areas in which affinities might be increased hereafter, in such things as asset management in the event that assets are made to be held in common between video and games, a workflow which allows both sides to proceed with production concurrently, the necessities of the pipeline, or optimization within pre-production work.
As each side furthers technological development, necessities such as development cost increases or training a workforce capable of addressing this work are likely to emerge, and further developments are anticipated as a result of advances in the hardware which forms the bedrock of real-time technology or the appearance of new expressive techniques, but since we can be said to have entered an era in which visuals with the affectations of traditional hand-drawn animation are being developed as multimedia, the moderators, believing that collaboration between the differing technological fields of video production and game development will likely become indispensable in the future, intend to plan the next seminar accordingly.

■Challenges in Approaches to Designing Pipelines for a Hand-Drawn Animation Style

Certain significant challenges became clear during this seminar as problems common to all studios, mainly thought to be capable of being implemented in the infrastructure architecture on the server side, including such things as system architecture to enable speedy internal and external information sharing, the establishment of progress management systems which include asset management, a highly flexible infrastructure reconfiguration, and improvement in the speed of reconfiguring the pipeline itself, all against the background of production circumstances within video production, in which what is sought is production that is intent upon delivering within a limited period.
As participating artists or the working environments of those artists are diversified, and reflecting the reality that directorial responsibility for outsource studios is getting more problematic year by year, it seems as though there is a trend, against the background of the increase in clients who are diversifying things like video broadcast, toward the emergence of extraordinarily high-level challenges in things like fortification, flexible innovation, and achievement of coordination for such things as speedy and accurate schedule management, establishment of progress management systems, flexible changes in production plans corresponding to changes in phases of production themselves, and creation of systems for recalculating budgets, all while also incorporating expressive methods from hand-drawn animation.

However, many participating studios having grasped the present situation, the current circumstances in which they have begun to create systems aimed toward innovation are understandable, and it seems as though a significant inclination has appeared to continue to actualize pipeline innovation, including forthcoming large-scale infrastructure innovation, together with high-level information literacy, on the basis of the establishment of production systems which push forward globalization.

■The Need for Diverse Seminar Forms and Discussion Spaces

Apart from this “Film Production Pipelines and Artist Techniques” seminar, the necessity for diverse studios to plan and hold seminars under their own aegis seems to have been thrown into stark relief during this seminar.
The context of this is that, in the present situation, there are challenges so large that they can by no means be encompassed within the scope of knowledge which the moderators ourselves are capable of covering, and in addition to this, we fear that possibilities for development may be inhibited if discussions of the varied cultures of each of the studios are carried out in a single seminar like this one, so that, on the basis of the idea that, in the course of each studio exploring seminar forms best suited to itself, it might be possible to gain opportunities to foster diverse cultures and manifest the individuality of studios, we as moderators also began, at this seminar, to feel that a limit had finally been reached in seminars planned on the basis of limited knowledge.

■Hand-Drawn Animation Style and Production in CG (from a production perspective)

Both during and after the seminar, various discussions related to optimizing hand-drawn animation styles have continued to develop, both within the seminar and in the form of exchanges of opinions with the moderators. Beginning with a reexamination of phases of production which have heretofore not seemed to require much digitization, but rather could remain as they had traditionally been, the present situation is that we have received a large number of proposals regarding how, treating the advantages of those phases of production as strengths from a “hand-drawn” point of view, these can be incorporated into cel-look CG.
During the seminar, there were also explanations of the strategies used at a number of studios for creation of genzu, and it seems to us that there is a strong tendency toward thinking rationally about the reasons why, compared to normal CG, cel-look CG can involve an increase in the phases of production and why the pipeline system becomes more complex, and that they are strengthening their tendency toward exploring future innovation, making optimization through the workflow their key concept, rather than working from the point of view of optimizing each separate phase of production.
For example, during color design, there is a phase of production in which the fill and line colors of the characters in a given scene are determined in a color sample format called a color palette, and to make it easier for color design staff to change colors using simple painting, it can be said that it becomes necessary to institute a workflow in which, for each asset, images are actually prepared without antialiasing applied, and the fill areas and palette numbers for each are linked, so that once color changes for the palette are complete, the palette is applied to each scene and rendered; and it can also be said that it becomes necessary to proceed with operations while managing large amounts of color palette data.
In addition to this, for 2D drawings, animators normally require the time sheet, which is the direction information indicating the timing of each cel frame (the key frames that indicate the timing of the animation), but in the CG production workflow, in which the time sheet does not exist, one challenge is how to convey the animation timing to the 2D-drawing animators.
Of course it is crucial to consider how to optimize each of these challenges as well, but one thing that we realized this time was that the challenge that is of greatest importance at present may be the exploration, in a form suitable for animation production, of a server-side progress management system with high interactivity that allows the workflow to proceed uninterruptedly; and through this seminar, the moderators began to think that solutions to issues in this connection should be explored first.

■Hand-Drawn Animation Style and Production in CG (from a systems perspective)

It seems as though the present situation is that, compared to CG-related studios, where flexible, large-scale asset management and system design architecture and operation with high information literacy are gradually becoming indispensable, the feel of things in system architecture or the operations side at studios specializing in hand-drawn anime or cel-look CG is quite different. However, at this seminar, we recognized that the present situation is that studios specializing in hand-drawn animation or cel-look CG have begun to steer sharply toward system architecture and operation with high information literacy, and judging from the fact that, after the seminar as well, there were a great number of proactive requests for reference materials to gain such knowhow, we have become convinced that there are likely to be great changes in future.
In the infrastructure that supports production centered around hand-drawn anime, it is still possible to perceive production systems that traverse between studios phase by phase, the amount of data not being terribly large; but at this seminar it was seen that, from the point of view of infrastructure architecture that can be made use of even with large-scale overseas projects in the mid- to long-term, a large number of studios have begun to show strongly reconciliatory attitudes, and a number of opinions were expressed regarding innovation within the sort of industry in which it is possible to aim to establish both coordination and individuality.
What is important here, in the opinions of the moderators, is decidedly not to firmly establish a direction in which to go, but rather simply to hold such a mindset, and on the basis of an awareness at each studio of aiming to establish a highly individualized workflow and skills, that there may be a need for a seminar plan which also concurrently takes into account coordination with a large number of studios both domestic and international, and that what was important this time was that we were able to share this point of view among the participating studios.

Continue to build in-house system

■One Pipeline and Infrastructure Suggested by the Seminar Session

Beginning in this section, as a way for the moderators to bring together what was covered in the seminar discussions, we should like to imagine and examine a workflow, suggested by what was seen during the seminar session, which might be thought of as a “shared” framework, as well as a corresponding pipeline design and a structure to support this, as a springboard for discussion at future seminars. During the seminar, there were times when opinions were strongly expressed that this challenge is seen as a shared challenge between the participating studios, so we should like to write this up as a way of bringing together expertise acquired during the seminar.
However, regarding elements of infrastructure architecture whose implementation would have a high level of difficulty within current technological limits as we understand them, on the basis of imagining that such things would become possible if certain systems were achieved, some of the following must remain within the bounds of conjecture; but, with the intention of creating something useful for discussion during seminars hereafter, we should like to bring these ideas together with a focus on technological challenges, so please bear this in mind as you read the following.

■One pipeline and infrastructure that was visible from seminar (Basic policy of infrastructure building)

Achievement of smooth progress management and speedy interactivity, as well as consideration of manipulability of flexible calculational resources, might be comprehensively considered to be the challenges raised at this seminar.
In other words, depending on the calculational cluster structure on the server side, we feel that it may be indispensable to build, not only a render farm suitable for the Cel-look, but also a system that can drive various procedures in the pre-production phase, as well as Web system architecture that can centrally manage all production-related information and can flexibly share information (with rights management included) internally or with related studios.

At this point, things like highly compatible asset data and scene graph management between studios are in demand, and we think that what may take on central importance is the creation of a system to do things like asset data or scene graph management relying mainly on MaterialX or USD, asset management which is based on an open format for purposes of look development or materials management, or continual archiving of assets. In addition, what may also be important, while actualizing the easy transfer of assets between 2D and 3D, which was discussed during the seminar, is, for example, being able to use MaterialX for materials management suitable for hand-drawing tools, and the fact that the building of operational techniques and management techniques capable of continuing to support the ongoing crossing-over of a workflow that, in the past, developed in a branched-out form, will become more important.


■One Pipeline and Infrastructure Suggested by the Seminar Session (regarding hand-drawn workflow)

First, in the following overview of “Present Production Flow for Background Matte Paintings in Hand-Drawn Anime” and then “Production Flow for Background Matte Paintings in Cel-look CG”, we should also like to provide an explanation regarding a pipeline and workflow, as well as the infrastructure to support these things, from the point of view of how the hand-drawn workflow can be effectively incorporated, as a way of drawing together opinions presented during the seminar.

The above is a typical example of the production flow for background matte paintings in hand-drawn animation, but the production workflow for background matte paintings in cel-look CG is not dissimilar, and one example of this would be to add phases in which rough 3D models for use in layout are created, these are positioned, and camera work using a 3D camera is added, from which a genzu is created; following this, the background matte painting staff proceeds with drawing. In other words, even as this fundamentally emulates the production flow for background matte painting in hand-drawn animation, in areas in which it is possible at present to adopt CG, we believe we have reached a stage at which CG phases have begun to be added.

As we learned from examples presented during the seminar, in recent years genzu production within hand-drawn animation has not only involved hand-drawn genzu production; a flow is also on the rise in which models for layout work are prepared and genzu are created, in much the same way as in cel-look CG. In other words, depending on the circumstance, adoption of techniques such as drawing a texture to be mapped to a model and, in the 3DCG phase, mapping this texture to the model and rendering it in that form, is also increasing year by year.
What these things mean is that it can be understood that, in the present situation, as cel-look CG is introduced to animation production, this is causing the workflow to increasingly resemble that of a typical CG studio.
Among the participating studios, in order to make instructions easier to understand for background matte painting staff in both hand-drawn animation and cel-look CG, 3D model data prepared for purposes of genzu creation will have a certain level of detail, and during the seminar, opinions were also expressed on the CG production side such as the wish to give greater freedom for camera work later on by drawing directly onto the 3D model data, so that it seems to us that the workflow and feel of things is drawing nearer to that of a CG studio.
Among the proactive suggestions shared during exchanges of opinion at the seminar, another developmental idea was raised: by building a workflow that will allow background matte painting staff to draw directly onto 3D model data, thus allowing hand-drawn anime-style background matte paintings to be created as 3D assets, it may become possible both to eliminate the genzu phase of production, and to concurrently grant a greater degree of freedom to camera work; and opinions were also strongly expressed that it is important to be aware of things like reducing idle time for artists during existing workflows, as well as to create working environments that grant productions a higher degree of simultaneity (in other words, innovation toward a concurrent pipeline), and that this sort of systems creation and infrastructure architecture to support it may also be important, in the additional sense of moving projects forward smoothly and on schedule. Of course, in response to this, the CG-related studios pointed out that things like USD or MaterialX are important for the building of such systems, and we believe that a lively exchange of opinions occurred during the seminar within the respective domains of both kinds of studio.

Looking even further, there were proposals for the adoption of 3D background matte painting beginning in the design stage, which is to say, during pre-production. Possibilities were implied such as the idea that work might proceed at the storyboard or story-reel stage making proactive use of 3D background assets and checking the camera work in 3D, making it easier to conduct the iteration needed for trial-and-error work and, by incrementally brushing up 3D background matte paintings, thereby implementing a more efficient workflow. However, at the same time it was also pointed out that support will concurrently be necessary for background matte painting staff to transition away from a workflow and tools to which they have become accustomed. The opinion was also expressed that a renewed exchange of opinions between both hand-drawn animation and mainly cel-look CG studios on the one hand, and CG-style studios on the other, will be crucial, recognizing anew the fact that things like pipeline architecture making it easier for drawing staff to work with 3D assets or handling on the infrastructure side for a marked increase in data size can be difficult for staff who are unfamiliar with engineering.

■One Pipeline and Infrastructure Suggested by the Seminar Session (regarding procedurals on the server side)

The point of view was reinforced during the seminar that server-side production infrastructure architecture, although its operation demands high information literacy, is nevertheless important, so that, as a way of incorporating opinions from past seminars, a lively debate arose organically around procedural-related matters on the server side.

For the fundamentals of server-side infrastructure architecture, please refer principally to the most recent reference documents on the related reference materials page.

Related reference materials page:

The above is a visualization of server-side infrastructure architecture presented as a conceptual chart used in an explanation at a prior seminar.
In an ideal user case, if these forms of workflow, including drawing, were executed on the server side, and then administered in a form similar to a Web application furnished with a UI and tools, then asset management on the server side, or an infrastructure that may potentially be developed even further as a new workflow, would be something that we began to think necessary based on discussions during the seminar.

Things such as Web services, which are also needed elsewhere, would be as detailed up through the prior section, but in phases such as pre-production, as well as production, due in part to the fact that, as a result of its characteristics, a system in which the staff involved proceed with work at disparate locations is often seen, a system with a browser interface that is based on Web applications could also possibly provide a location-free environment, which could conceivably be an extraordinarily effective system for pre-production.
It can be said that, in pre-production work until now, there has been a tendency to shy away from work with a high computational load, due in part to the fact that such work has typically been done within a working environment located on a stand-alone machine; but an approach in which, by doing such processing on e.g. the server side via Web applications, the artist’s machine can expend computational resources only on highly creative work, while processing which is simply heavy can be processed on a computational cluster running in the background, is also likely to come into view. While the moderators do not by any means consider the utilization of server-side infrastructure to be a simple task, in consideration of both its high flexibility and speed, we still consider this an important approach.


Even more than anticipated, we learned of the present situation at various studios, dealing mainly with hand-drawn animation production or cel-look CG, intent upon ambitious innovation regarding technical matters; and in the course of such topics being raised as high-level information literacy and the outlook on the server side, the moderators came to feel that this was an extraordinarily meaningful seminar.
The items in the “One Pipeline and Infrastructure Suggested by the Seminar Session”, with which we have concluded this report, have only been given here as a springboard for innovation hereafter, as a way of bringing together things raised during the seminar; but we believe various strategies, not only limited to these ideas, will continue to be developed.