Status update & request for feedback

[Rob Lohman]

You've all been waiting on this for a while now so here's a status update on Adobe importer.


What does it do:
-----------------------------------------
See the following thread for a video introduction: http://www.reduser.net/forum/showthread.php?t=17782

The importer allows you to load R3D files in all resolutions, framerates and camera firmware builds up to at least build 17 (unreleased as of yet).

It will decode into 8 or 32-bit depending on your settings in Premiere and After Effects.

It will be available out of the gate for both Mac & Windows


What does it not do [yet]:
-----------------------------------------
The big thing missing is audio. I know everyone wants it but it will not be in this first preview (and we are labeling it as a preview for you to look at and provide feedback on where we should go in your opinion).

It will come really soon however, it is the big thing on our list after this release for all our products, not just the Adobe importer.


Focus for the next release(s):
-----------------------------------------
Audio, performance enhancements and deeper integration with Premiere & After Effects (especially for CS4).


When is the preview coming out
-----------------------------------------
In the next couple of days. RED and our alpha testers have identified some critical bugs (Premiere hanging and images decoding wrong) that need to be fixed before we can give it to you. We are so very close but these last couple of problems are determinant to keep me up at night. We want to get it out as bad as you do!


A workflow question for you
-----------------------------------------
A RED ONE clip can consist of multiple R3D files. Normally we load the full clip (all associated R3D files) when you the user select one of the R3D files. This removes the burden of having to select everything in the right order.

Unfortunately this system causes some challenges in CS3 (we're working on solutions for CS4!) and we'd like to get your feedback on how you think the import should work.

Importing footage in Premiere can be done in two ways:

1) importing a single file. In this case you would like to import all the R3D files belonging to the clip

2) importing a folder. In this case you want to import the clip just once. Premiere, however, sees all R3D files and asks us to import each one of them. Unfortunately we cannot detect this and switch of the loading of the full clip for each R3D. This would result in the clip showing up X times (where X is the number of R3D files in the folder).

So we have two options to proceed. Keep the full clip loading in which eases the single clip loading. This creates more work if you're importing a whole bunch of footage since you might get duplicate clips (on long clips) which you will have to manually remove.

Or we can disable the full clip loading and always load a single R3D file. In this case you probably want to stick with always importing a folder to make sure you get all the R3D's belonging to a clip.

Please provide us with your thoughts on the subject.


Thank you all for your patience on this. We know it's been promised for a while now and we're working hard on squishing these last bugs and get you the best workflow possible now and in the future!

[SalaTar]

Rob,
Edited, afx'ed and titled.
Now I want to render back to r3d....

[Obin Olson]

We need to focus on getting split files into premiere as just that, split files so that we don't have weird pointers floating around that would allow for mistakes. Example: If one part of a clip is corrupt we would rather have 1/2 the take then nothing at all......

[Edgar Pitts]

I second what Obin said. IMHO the weakest link of Premiere CS3 is how it deals with subclips and multicam. There can be problems conforming from Premiere to After Effects.

The more straight forward the file and import system (even it causes more upfront work), the less headaches down the road.

Edgar

[Obin Olson]

can premiere look in sub folders for footage when importing
an entire folder? Its a real pain the way red records each new
take in a new folder - we need a way to batch import all footage
that's easy and quick.

[Obin Olson]

Rob after audio I think the highest priority are some standards
based presets for red like 720p 1080p 480p etc the weird framesizes
are just that, weird. For example I never really know what's pixel
accurate because if your off standard sizes.

What's really needed is a way for you to scale say 1024 to 1280x720 so
we can stay with standards, this will make it easer to work with hdsdi boards etc I think.

[Obin Olson]

And on a new note following audio and some standard presets
some native support for HD boards would really rock so we can get
our decklink etc working for hd edit output for monitor. Everyone in a studio
is using a 3rd screen - plasma - LCD etc for playback, we need an 'online' image playback device connection - I am a little afraid of consumer graphics cards and would rather have some standards to go by- hdsdi comes to mind or atleast 4:4:4 hdmi.

This brings me to sync, one thing that hdsdi can do is make sure your audio and picture are synced properly upon scrub and play. Very important for critical work.

One other area some type of standards based output is going to be needed is
color. We need a way to get away from the low contrast LCD screens we are all working on and have playback on a professional device - projector - plasma etc so we can judge our work on a display that's calibrated and very accurate.

Somthing hopefully higher then 8bit per channel.

[John Tissavary]

Rob after audio I think the highest priority are some standards
based presets for red like 720p 1080p 480p etc the weird framesizes
are just that, weird. For example I never really know what's pixel
accurate because if your off standard sizes.

What's really needed is a way for you to scale say 1024 to 1280x720 so
we can stay with standards, this will make it easer to work with hdsdi boards etc I think.

Presets are wonderful, but there's nothing like knowing how to 'roll your own'...


If you're confused about scaling, here's a simple formula: target width / source width = scale factor

In your quoted case the simple calculation is: 1280 / 2048 = .625
Just scale your source material by .625 and you'll be at 720p.

It's just as easy with 1080p: 1980 / 2048 = .9668
Multiply 2048 by .9668 and the keys to the 1080p universe are yours.



cheers,

John T.

[Mike Zinner]

Strong vote to treat each .r3d file as single clip.

If you cannot do the automatic multipart selection flawlessly, leave it out - it just causes confusion otherwise. Especially since it is quite logical that one has to select each .r3d file he sees in a directory.

Well, I even say treating each .r3d as single clip has advantages. I've been cutting some single-take interviews which consisted of 3-4 .r3d files (converted to Cineform .avis) and most of the time I would only import selected files and was able to delete the useless files to save space for archiving.

So, to avoid all the future complain-postings, please the multi-part import out.

Thanks!

[Mike Zinner]

Another workflow question. I wonder why it is necessary to select a fixed "scale-down" factor for each RED clip resolution.

I am sure you have analyzed the Cineform approach which is much more straight forward in this regard: It always leaves the original clip resolution in place and only adjusts the de-bayer resolution based on the CPU load in realtime. And when you stop the playback, it decodes the frame with full resolution.

This approach does not require to manually change any values and restart the application in order to perform the final output render in full resolution. And still provides 4K quality for the current frame to apply color correction & effects.

I guess this is a limitation of the Pr CS3 API? And the reason Cineform replaces the complete Pr CS3 rendering engine?

A comment on this would be nice :)

Thanks again!