Loop reel instead of copy (minimize size)

[IoannisSyrogiannis]

There is something that I was trying to do lately and I couldn't.
I think it would be a great feature though. It might be that I try to do things the wrong way or that there is no way (yet?/).
Either in creating test patterns, or ads, or repetitive video and/or audio, one might need to repeat small clips or slides.
I'll make an example with slides, to simplify things, but I can offer cases where clips are used again and again, like in audio test paterns etc.
Let's say I want to have a slide on the screen for 20 minutes, for convenience reasons, I might preferred not to have 10x2'DCPs (CPLs) one after another with the same content, but rather have a DCP of the total duration that would use the same video and audio ten times. That would make a tenth of the total size. In festival cases, where such background slides (or clips) are needed, size is of the essence and different versions occupy more space than wanted in small integrated servers. (Unless you have a Sony RAID-6 server or a closer and closer to the end of support ShowVault4 or DSS200 storage upgrade-able server.)
In another case, One might want to loop between slides A, B, C and D and do that 40" loop quite a few times.
Yes, once can make a playlist with that, repeating again and again, but that would severely increase complexity and create the need for nested SPLs on nested SPLs. (By that I mean using a playlist inside a playlist as a clip that is to be put (as a clip) into a playlist itself.) That would make inter-cinema sharing or SPL "translating" between servers a problem also.

What I would love to be able to do is making an ABCD sequence and DCP-o-matic create just the video for that, repeating it for a customizable number of times (just the way one can repeat slides on the timeline) and (let's get wild here) maybe add an ABC on top that, either by creating that ABC video separately, or -much, much more data size efficient- by using the corresponding duration of the ABCD video.
Or on another case, have an ABCD block, an EFGH block (one could read "audio channel ID test" and "pink noise channel test" here instead) and have 2xABCD, 2xEFGH, 2xABCD and so on in a desired pattern, having it to fit in a USB thumb-drive painlessly.

I did try to create a DCP and then using it repeatedly to create a multi-reel, but the content would be created (copied) again. No VF would do the job either. And I wouldn't dare to think of using the "duration" CPL feature on my experimentation, since there isn't a choice for that. (Is it?)
One can suggest doing the above manually. Yes, I know I can mess around with the CPL file, but one can appreciate the convenience of such a feature among the others of DCP-o-matic.

[carl]

Hi,

This makes sense to me. Some or all of this request may already be covered by this entry in the bug tracker. Feel free to add comments there if you'd like.

I haven't thought a lot about it lately but it seems to me logical that we should automatically re-use data for reels which are the same. So if you add images A, B, C, D, set DCP-o-matic to use a reel for each, then duplicate A it should re-used the data from the first 'A' reel (if you see what I mean).

[IoannisSyrogiannis]

It took me a few minutes to create an account, but I did.

What I wrote and I'd like to repeat here, so to put another perspective on the subject is the following:
The CPL structure can be a very powerful tool for getting more content with less data usage.
The benefit would be not only on the data volume department, but also on the time demanded for the creation of some kind of DCPs.
Slideshow DCPs, test patterns, videos used as background etc. on different occasions would not only be significantly smaller, they would also be faster rendered into DCPs., needing less rendering and less hash calculation time.
Users of the DCP-o-matic that their main purpose is to make slideshows/ads for their cinemas would be able to do so more easily, especially on computers that are not dedicated on that purpose.
While enhancements or reviews on the UI are necessary for such a use, it is not an "add a checkbox here" change, I believe the program's versatility will increase so to worth the while.

[Carsten]

The question really is how to add this functionality in an elegant way. The trouble is, so far, DCP-o-matic doesn't offer any direct handle to reels.

Wondering wether a global option 'reference content' could handle all incidents. DCP-o-matic would then have to identify individual content that is repeated in a project, create a single reel for it, and from then on, only reference this reel in the CPL. I guess the only way to enable this would be to have DCP-o-matic keep track of repeated content, and one would have to layout such content explicitly as 'repeated'. Otherwise, DCP-o-matic would have to calculate a full hash of every piece of content, and keep track of all changes to it. I guess all that is not simple.

- Carsten

[IoannisSyrogiannis]

What I was thinking is rather user defined, not automatic.
Meaning, creating a description of one or more "blocks" (where "block" is a specific part of video, audio, etc. or a sequence of different clips/imported DCPs or other files) and handling those "blocks".
If I was to try and find an analogy here, it would be the "nested" SPLs. Meaning a show playlist that is used inside another (having to do with Sony cine-servers, if you don't use a TMS, I am sure you know what I am talking about...). With the distinct difference that, while on an SPL (show playlist) one usually have every "nested" SPL only once in each show, while creating a DCP, for the reasons I already explained, the user would be able to repeat that "block", that "nested" content as many times he or she would need.
No need for extra hash in such a way. On the contrary, the files that would be hashed would be way smaller in their whole, making the procedure faster.

What you are describing, Carsten, seems more of an automatic search for repetitive parts of DCPs, in a way that (maybe) a compression procedure would work.
That seems to me also that would very easily end up creating a mess of a DCP and I don't believe anyone would like the result of it.
Again, I really think that it is more of a matter of making the user interface capable of using such a feature. Apart from the output of the CPL file, the rest of the DCP creation would be just the same for such cases. Just significantly faster and with smaller resulting packages.

[Carsten]

Again - the question is how to integrate such a feature in the current GUI in an elegant way. I think this is not an easy task, to offer access to the individual reel. It's a completely different approach, which so far is source based. I consider the current GUI already clogged for the average user, due to the added features in recent DCP-o-matic versions. The trouble is to hide the advanced options from the essential ones. While I absolutely see the advantages of being able to reuse assets in this way, it is still a very limited number of users who will actually need it. I we find a way to do it automatically AFTER a switch has been set deliberately, it may probably be the better option. In that case, we can still use the current content-based approach for arranging, but tell DCP-o-matic that individual content items MAY be reused as reels where ever possible. That should be strictly on the content level, e.g. a piece of video, or a still of defined length, etc..

For general work, as overlookmotel has put it on Mantis already, I always prefer the most basic reel arrangement, which is, one reel per asset type.

But, maybe it would be better to create a separate application for this. The problem is, very few people will actually use it. It would probably only deal with MXF files of the three common asset types, and allow to arrange them in a timeline. That type of dealing with MXF files as references only is completely different from what is available in DCP-o-matic main now.

- Carsten

[IoannisSyrogiannis]

Dear Carsten, by starting with "Again", you gave me the impression that either of us or both try to convince the other about what our personal opinions are and what we want to communicate. That isn't and wasn't my intention.
My intention is to share the idea of a feature that, indeed is not available in DCP-o-matic right now and according my believes, would greatly add value to the application.
The question, at least for now - to begin with, is not how to integrate such a feature in the current GUI in an elegant way, as you suggest. It is if Carl would be interested to see that feature integrated in DCP-o-matic. The rest, derives if only that first question has a positive answer.

I wouldn't dare to suggest that the implementation of such a feature is easy or otherwise. As far as the GUI goes, I wouldn't be able to categorize it and nor I would the average user. What I get from this forum's users, I without problems can find among them both users that would not touch such a feature and users that would, you find yourself longer and more in this forum, you have a wider idea on that. I understand that what I see can not be put into numbers, so I won't argue. I already explained the possible uses of that and they vary from making a slideshow, that is a basic use of the application to making a looping video or a test pattern, that gradually advance in rarity of use.

I didn't want to argue with overlookmodel on the Mantis, since I find the forum a better place to have a discussion, but the fact is that manipulation on one or more reels in the CPL is there since day one of Interop and series I projectors and servers would deal with it as they always do. The structure was built that way and its flexibility has afterwards been inherited to other audiovisual packages.

That is my view on the subject. I am ready to share my idea and answer questions about it, but I don't want to change your mind or anyone else's for that matter.

[Carsten]

Ioannis - I don't mind any form of 'reasonable' discussion - be it pure statements, or trying to convince someone

As a tech guy, I certainly would love to have an application that allows me to create any type of DCP that is supported by the standards. But we also have to think in favour of the common user. I have a need to create small DCPs as well, in fact, so far I resorted to other ways. E.g. I created a long intermission DCP which uses timed-text over a very low datarate black background. One could probably do a similar thing with a detailed background using a PNG.


Wondering wether there would be a way to create 'another' timeline view accessible from the DCP tab. Just as the current timeline view is source based, there could be another timeline view under DCP that is DCP/reel/asset based. Users wouldn't need to touch it, and maybe it would only be there if we'd enable it with a another selection under 'reels'.

I wonder how other DCP mastering solutions enable access to reel referencing. I think most of them separate the J2K/reel conversion and the final DCP creation. While the J2K conversion is source based, the DCP creation then is reel based. So, if you created a reel/MXF from source content, you can then reuse it as many times as you wish in the DCP. That makes sure that reels are fixed structures with a single UUID when the DCP is created. DCP-o-matic combines compression/reel creation and DCP creation, so, that is a very different approach. Also, if we separate reel handling from source handling, how do me make sure that audio and timed text stays in sync? I guess DCP-o-matic would need to keep an in and out point for reels that are referenced together.

Best wishes for the new year!

- Carsten

[IoannisSyrogiannis]

What comes to mind when I think about such a feature is actually a combination of different projects.
Like the Add DCP feature.
Actually, being able to add DCPs that can be used as reels (or as multi-reels) to create a combination of them without duplicating the assets, but using them repeatedly. As a further evolution of that, I would imagine a combination of projects instead of ready-made DCPs. Meaning creating and saving different DCP-o-matic projects that would be able to be used as “blocks” (as I refer to them), to… make a building. And the ability to have fractions, like 4,5xDCP (or project) would be great for using looping videos or shortening slideshows without using another reel or DCP for the half part. Yes, that would need in and out points for reels.
That, I consider the simplest way to be used for construction.
Of course, a unified depiction of different reels just from the beginning, as you describe other DCP mastering solutions, would both make things easier editable and complicated. As one wouldn’t have to go back to a DCP or project and forward to the “construction site” to make last minute edits, but would call for a more complicated structure of the interface (another tab, probably).

In that sense, of using “blocks”, it seems to me that the matter of synchronisation of audio and timed text would be taken care in the same fashion that it is already. I am unsuccessfully trying to remember from heart how I used making a VF out of a DCP to shorten it without making it anew, but I have the impression that most of such concerns are already taken care individually. Thus, they need not much more care, unless a more elaborate structural representation, like the one you mentioned, is to be implemented.

All that are guesses more than facts. Yet, I have the impression that the distance between them is not great.

A great new year.
(I haven't yet read the note on mantis.)