Epic module request...epic idea

[Zakaree Sandberg]

I know there was talks of a prores module at one point..id like to add onto this discussion a better idea.
Id love to see a module that can make any of the popular file formats or whatever you upload.

So the camera shoots 5k r3d files to the ssd media, and the module records, say...avid DNxHD 36 8bit to cf cards.

This of course, only if the camera cant dual write to the ssd card eliminating the need for an extra module.

[Jacek Zakowicz]

Knowing Red they have been on it already (just speculation- no inside info)

[Jeffrey T. Morgan]

Check this out:

http://www.atomos.com/samurai/

[Jacek Zakowicz]

If it only worked as advertised....
Anyway I think Zak meant R3D 5K files to various formats- not video assist signal- there is a major difference.....

[Curran Giddens]

http://www.sounddevices.com/products/pix.htm

[Stephen Pruitt]

Does the Epic (and RED MX for that matter), do a full deBayer before the HDMI signal output???

Stephen

[Zakaree Sandberg]

I think the hdmi out is a separate signal from the sensor...its not taking the raw image and turning it into 1080.
Theres different paths, one moves through the system to capture via r3d and tge other spits out a 1080 image

[Rich Schaefer]

You can kind of do that now with the Convergent Design "NanoFlash". It will record the HDsdi out of the camera to FCP, of Avid files.

http://www.convergent-design.com/Products/nanoFlash/tabid/1666/Default.aspx

I have to say, a module would be very cool to eliminate cable issues...

Cheers,
Rich

[Tom.Wong]

no module is really needed, i guess the convenience of being able to slap it on as a actual piece of a camera rather than something dangling off. but I think a better recommendation would be to have RED or lobby to the companies that make external recorders to follow the way of the ki mini and retain all RED metadata including file name on the record via sdi, so you can have instantaneous offline files for editorial as you are shooting. lower debayer quality is of no consequence if it's just for editorial. If people want a "debayer" module to create instant fully debayered quicktimes, than that's a different story. and kind of defeats the purpose of shooting 5k raw... when you can just do that in post, and gain RAW manipulation and 5k re cropping if you need it. and it would probably incur a lot of people to not even shoot the r3ds and just go for the 1080p which RED has always been against...

[Bob Minervini]

yep, a module that does what the kimini would be great, less cables would be very practical. Red should get the module connectors diagram to other companies so that they can design custom modules.

[Jacek Zakowicz]

For TV work the monitoring signal is inadequate especially compared to Alexa and such. Like it or not 1080p is the delivery format for TV and it will remain it in short term.
Such fully debayered properly scaled files would do wonders for Red in TV industry....

[Blair S. Paulsen]

No doubt, a module that wrote ProRes or DNxHD 1080 files would be popular - particularly for TV work. That said, there are more reasons for Alexa's popularity in TV than just being able to write ProRes files on the fly. No matter how much I may love working with the Epic, for a lot of veteran DPs the familiarity of the Arri built Alexa is compelling. It would be easy to argue that the Epic does so much more, but for some DPs that is not the cheese. No offense to RED, they can't help having done so much so fast ;-), but 4 years of making cameras is just not the same as decades of excellence from Arri. While I share the consensus around here that Arri talks out of both sides of their mouth about resolution, they are still Arri. Never forget that our business is like most others, people want to cover their butts and Arri is considered a safe choice.

What's my point? I would prefer to see RED and its users embrace the power of a high resolution RAW workflow. As Mike Most and others have noted posting R3Ds is not actually more difficult that other post paths, it is merely seen that way. Is it harder to change minds than to fall in line? Sure, but IMO RED has never been about dumbing down to suit entrenched stakeholders but rather about finding a better way.

Cheers - #19

[Jacek Zakowicz]

What's wrong with HW encoder using " TV LUTs" for TV work? Or a custom ones created by the DP? How is it different than RR and editing suite? It would be faster for sure....

[Blair S. Paulsen]

What's wrong with HW encoder using " TV LUTs" for TV work? Or a custom ones created by the DP? How is it different than RR and editing suite? It would be faster for sure....

IF the debayer quality made possible by the ASIC in the Epic is first rate and care is taken to create quality LUTs that can be loaded into the camera itself then OK. I still think you are leaving some of the IQ on the table (there is an "informed algorithm" issue here) but I can see the argument that a 1080 rec 709 master for TV probably isn't capable of showing every nuance anyway.

It is entirely possible that my affinity for the "virtual darkroom" aspect of creating evocative images colors (sic) my take on the subject. The Gestalt approach I prefer encompasses that color grading step as an inextricable part of the process of authoring the image. Yes, I know you can grade RGB material, but grading RAW (particularly the 16bit Epic material) is so much more powerful.

Cheers - #!9

[Jacek Zakowicz]

Isn't grading taken care of by the LUT? I've seen "Gestalt" grading so screwed up so many times that I'm sure it was not the DP's plan and he/she would much prefer to retain the "LUT control" for final product....

[Zakaree Sandberg]

As Mike Most and others have noted posting R3Ds is not actually more difficult that other post paths, it is merely seen that way.

Cheers - #19

does avid edit r3d files?

Almost every commercial I do is now AVID. and the new common sneak attack by post facilities is to get the DIT to do the transcoding for them, so rather than being on set helping the dp out with exposure decisions or camera settings, running video lines, and helping out the crew, im stuck in a hot ass tent or truck shuffling multi camera cards around from my back up machine to my rocket machine, and my day is spent.

Id rather have the camera auto pump out the offline edit files (DNxHD) and all i have to worry about is making sure that data makes it onto multiple drives.



PS.. biggest pet peeve = being asked to transcode alexa files.. GFYS post

[Zakaree Sandberg]

Isn't grading taken care of by the LUT? I've seen "Gestalt" grading so screwed up so many times that I'm sure it was not the DP's plan and he/she would much prefer to retain the "LUT control" for final product....

yes basically pop off a shot in pre light or 1st up in the day, do a grade, plop that lut into the camera, BAM!! done deal.

ESPECIALLY for TV.. stage = lut through out the series, location home interior tungsten = lut throughout the series, location home exterior day so on so on..

commercials = a little harder... 1 - 3 day shoots.. multiple locations.. not much time to really get in make a final look and set it, but FOR SURE a quick reference for dailies/offline edits

[Dan Kanes]

With all due respect.

F*XK The KiPro.

Red should make a module that does exactly what Zak is asking for, and records on SxS cards in ProRes or Avid.

I'd call it the Instant Dailies module.

All that external shit is garbage.

One key to red Workflow on set is the fact that it doesn't need an off board recorder.

[Zakaree Sandberg]

With all due respect.

F*XK The KiPro.

Red should make a module that does exactly what Zak is asking for, and records on SxS cards in ProRes or Avid.

I'd call it the Instant Dailies module.

All that external shit is garbage.

One key to red Workflow on set is the fact that it doesn't need an off board recorder.


exactly.. any external recorder is just BS.. AC's dont want that shit dangling, nore do OPS with a camera on their shoulder.

the main thing here is getting offline files made on the fly

[Steve Sherrick]

does avid edit r3d files?

Almost every commercial I do is now AVID. and the new common sneak attack by post facilities is to get the DIT to do the transcoding for them, so rather than being on set helping the dp out with exposure decisions or camera settings, running video lines, and helping out the crew, im stuck in a hot ass tent or truck shuffling multi camera cards around from my back up machine to my rocket machine, and my day is spent.

Id rather have the camera auto pump out the offline edit files (DNxHD) and all i have to worry about is making sure that data makes it onto multiple drives.



PS.. biggest pet peeve = being asked to transcode alexa files.. GFYS post

Yes, Avid can edit R3D natively but there's still more work to be done there. :-)

There should be no sneak attack Zak. Is this part of the process not being communicated until you get to set? It shouldn't be the case. That's a part of the process that needs to be discussed ahead of time so you can plan for it, and also charge accordingly. You mentioned commercials, which I admit can sometimes be a bit tricky, especially when you're brought on late in the game and then you're only on for a few days at most. I don't feel transcoding is a given. As a DIT your priorities are making sure the DP is getting what they want out of the images, making sure the camera is setup properly in terms of image settings, making sure the monitoring chain is getting the images properly, and if you are taking on the role of data manager as well, then that too becomes a top priority. Transcoding is something that should fall below all the rest in priority list, but if your system is setup for it and it can be done while getting the other things accomplished as well, then that's fine but you should make sure you are compensated for it because at that point you are essentially a dailies house as well. I know each situation is different and perhaps that's not the way everyone feels, but that's how I see it.

As for the Alexa transcoding, I had a client hat chose Alexa over RED because they didn't want to transcode files. Then at last second post said they didn't want to deal with ProRes 444 for editorial so they ended up transcoding on set down to ProRes 422. Too funny!

[Zakaree Sandberg]

im set up for it.. but only 1 computer has the rocket, and im not about to get another when productions dont even pay what i should be getting on rental as it is..

1 camera is no problem, when you start getting into 2 or 3 cameras with real long takes... it starts to suckkkkk

[David Battistella]

I think the real issue here is licensing.

You can't just build a box and have it record prores without paying a fee to Appke to license the codec. Depending on the deal ARRI has with Apple (which could be exclusive) it might mean RED can not or does not want to pay a huge fee to Apple. They would have to sell a lot of units to cover those fees.

The same goes for AVID codecs.

The kipro mini seems to be the best solution for offline prores as it already does everything you want a prores module to do. No fees for RED to deal with.

H.264 recording presents all if the exact same problems you get with DSLR material.

The best solution might be a wireless or wifi modual that distributes a signal that can be viewed on mobile devices and recorded to a codec of choice.

This too might be tricky because getting a high res stream might be hard.


I throw down this idea.

Why not make have that module send out a high quality, high compression, low bandwidth 4k stream of REDRay.
Then all the devices and monitors on set could watch a very high rez stream which could be recorded to an editing codec on the fly at the DIT station.

That would be high quality, smart and cool.

David

[M Most]

Why not make have that module send out a high quality, high compression, low bandwidth 4k stream of REDRay.
Then all the devices and monitors on set could watch a very high rez stream which could be recorded to an editing codec on the fly at the DIT station.



Because according to just about everything we've been told about RedRay, it cannot be encoded in real time or likely anything close to real time. At least not at this point.

[Steve Sherrick]

im set up for it.. but only 1 computer has the rocket, and im not about to get another when productions dont even pay what i should be getting on rental as it is..

1 camera is no problem, when you start getting into 2 or 3 cameras with real long takes... it starts to suckkkkk
You would have hated the movie I just wrapped on. 2-3 cameras running all the time, long takes at 5K 5:1. impossible to manage and transcode all that material with a single system. We were setup with a DIT and a near set dailies station, so no problem. But couldn't imagine doing it any other way. Too much footage!

In terms of rentals, I agree with you Zak. You can only invest so much money if you're not seeing ROI on it. It actually makes no business sense to buy lots of expensive gear if people won't pay for it. Even if it means turning down jobs, no one should get things for free if you've had to invest the money into it. Sure, the next person may take the job and give it to them for free, but history tells us, they probably won't be around long because they eventually have to pay off their investment too. :-)

[Alexander Ibrahim]

H.264 recording presents all if the exact same problems you get with DSLR material.

Uh .... only if you choose to do it the way DSLR's do it. Cheap and fast. The real problem is the data source: a line skipped sensor feed with the wrong OLPF. Even if you were recording 10 bit 444 uncompressed in camera you'd have a compromised image.

h.264 can handle up to 4K resolution at 14 bits per channel 444 with a bandwidth of ~940Mbits/s, or 117Mbytes/s. It can even do predictive lossless coding.

That's just not how its commonly used ... but that isn't h.264's fault. Frankly h.264 kicks ass.

The best "popular" use of h.264 is AVC Intra 100, which is 10 bit 422 at up to 1080p. I think that its every bit as good as recording ProRes HQ off the HD-SDI. (This is because the camera has access to the uncompressed 14 bit sensor data to do colorspace translations and compression, where an external module must use 10 bit 422.)

Intel's IPP can do up to 12 bit per channel at up to 1080p, but it lacks a ton of other features.

The only downside to a codec like h.264 is the processing required. Of course, OpenCL addresses that adequately, and you can always transcode to a more edit (i.e. processor) friendly format for post if OpenCL or CUDA are not available. Also the updated "Sandy Bridge" Core i series processors (2011 and newer Mac models) include full raster h.264 decoding in hardware that is just as fast as CUDA or any OpenCL implementation.

[David Battistella]

Because according to just about everything we've been told about RedRay, it cannot be encoded in real time or likely anything close to real time. At least not at this point.

I get that part Mike.

I'm just thinking that REDray has always been a "transport stream". "If" there were a module that could broadcast that stream (for high rez monitoring purposes) and something at the receiving end that transcodes that to an editing codec or HDSDI signal that can be recorded, then I am exploring that conversation as a possibility.

There has been little talk of the H.264 module (which would require processing to an editing codec as well).

So this is what leads me to REDray (low bandwidth stream) that could be broadcast via wifi from the camera. This is also a large budget production solution as smaller productions might just use a KiPro Mini.

David

[David Battistella]

The only downside to a codec like h.264 is the processing required. Of course, OpenCL addresses that adequately, and you can always transcode to a more edit (i.e. processor) friendly format for post if OpenCL or CUDA are not available. Also the updated "Sandy Bridge" Core i series processors (2011 and newer Mac models) include full raster h.264 decoding in hardware that is just as fast as CUDA or any OpenCL implementation.

This was my point. It's not an editing codec, it's just not optimized for editing. This is why Prores exists.

The other side of this argument is this.

EPIC shoots RAW.
RAW image capture is the cornerstone of RED's program.
What happens when RAW is the backup and you have a Prores file from the module that everyone edits with? (rhetorical)
You see how it is kind of against the whole point of building a camera that shoots to RAW files.

The Alexa can shoot a RAW image but nearly no production opts for it, going 1920x1080 prores sxs cards instead.
SHould EPIC try to mimic that or be what it is: A RAW DIGITAL CINEMA CAPTURE TOOL that replaces film acquisition?

I know these discussions always turn to turning a RED camera into an EFP style (HDCAM type) production camera, but the whole philosophy of RED is to make a cinema camera that replaces film.
This discussion was toiled over at length with the RED ONE too.

It might not be a practical position but sticking with RAW workflows is what RED should always push for, from a craft and technical image quality perspective.

Cue the big lebowski...

"But that just like, my opinion, man"

David

[Blair S. Paulsen]

At the risk of hijacking the thread, Zak and Steve S. both hit some critical points about on set data management, transcoding, LUT management, time with the camera team vs pushing data around.

I have had some very interesting clashes of expectations on some projects due largely to the wide range of duties/skill/gear expected of a DIT. Exacerbating the issue is that a proper setup for transcoding dailies requires more than the $150/day kit (box) rental listed in the budget the UPM (Unit Production Manager) is looking at.

I have been working on creating relationships with post facilities where my DIT work, including dailies generation, is actually included in the post budget rather than production. The idea is that as soon as the media comes off the camera it comes to me and has for all intents and purposes entered post. Other than accounting for meals, power, parking and other logistics issues this should actually make the UPM's life easier.

This probably deserves its own thread...

Cheers - #19

[David Battistella]

Blair.

Start a new thread....

I have a lot of respect for this position. It requires many skills, trustworthy people, camera knowledge, post knowledge, audit trails, correct procedures for copying and transcoding data.

A very, very important and essential bridge between camera and post.

More now that ever on a film set, this person is in charge of the safe keeping of the negative.

What it is NOT:
Digital loader
Data Chimp
back up guy

[Paul Leeming]

$800.

That's all it takes to negate this entire thread.

Buy a copy of Adobe Production Premium CS5.5 (using the Final Cut 50% off switcher discount). Install. Download the Epic 5K plugin for it (free).

Drag and drop your R3D clips directly onto the timeline. Set playback resolution to the capabilities of your computer. Edit using the online, 5K RAW files. No offline transcoding needed. No doubling of hard drive space needed.

Manipulate. Colour correct. Render once, at the end, to your desired delivery format.

Done.

Why is this so hard for the Apple faithful to grasp? Is it Final Cut myopia preventing them from seeing the woods for the trees?

I really don't get it....

Cheers from Tokyo,

Paul :)

[Zakaree Sandberg]

Problem is I work in high end production, and have no authority telling production what post house to use or what gear that post hpuse should use..most are AVID, most have been asking for dnxhd

Pain in the ass when it sucks up my time away from set.


So rather than trying to get mulyiple software companies on board with native editing, i think the internal offline file creation would be the real winner. raw is great, but if you have to take it into a computer and make new files, it just adds cost and time (i dont do it for free)

[Dan Kanes]

Here's a question for the Adobe people - how fast is it to batch process and output dnxhd files or prores 422 for that matter using a decent gpu (or two) - is it as fast or faster than RCX with a rocket?

Would be cool to have a second generation rocket that is cheaper/faster without sdi breakouts just for transcoding using either a later generation DSP or Red custom ASIC. Some kind of portable, expandable, affordable super-DSP that you could gang several of them together to speed render times.

It's at the point where an on-set transcode job really needs a 1:1 ratio of computers to cameras. I completely agree with Zak.

Throughput even on an esata/esata transfer is just not enough for some of these jobs.

[Matt Uhry]

100% with Zakaree. I don't get to dictate to the agency or editor ANYTHING. You get a spec sheet of what THEY want.

Guys who do their own post are in a different situation. Awesome that Premiere 5.5 is doing the trick. Will look into it

Module would be very useful.

Matt Uhry
www.mattuhry.com

[Blair S. Paulsen]

I wonder if RED would have any interest in partnering with Sound Devices to create a version of the PIX 240 with an Epic module mount/interface?

I am imagining something that would be similar in some ways to the Pro I/O module that was on the roadmap a year ago but with a bit different feature set. I suggest Sound Devices because their audio quality is legendary, their build quality is solid, they are already making a physically small device (PIX 240) designed to live on camera, they are already writing both ProRes and DNxHD (Avid), the PIX 240 is a full blown timecode generator and reader, etc.

Yes, I can just hang a PIX 240 on my Epic and cable it up but it seems a shame to waste the elegance of the modular design. Power, metadata, start/stop, hero audio pass through back to the R3Ds, etc all cable-less and solidly mounted with no extra gack.

A man has to have a dream, right... ;-)

Cheers - #19

[Bob Minervini]

I wonder if RED would have any interest in partnering with Sound Devices to create a version of the PIX 240 with an Epic module mount/interface?

I am imagining something that would be similar in some ways to the Pro I/O module that was on the roadmap a year ago but with a bit different feature set. I suggest Sound Devices because their audio quality is legendary, their build quality is solid, they are already making a physically small device (PIX 240) designed to live on camera, they are already writing both ProRes and DNxHD (Avid), the PIX 240 is a full blown timecode generator and reader, etc.

Yes, I can just hang a PIX 240 on my Epic and cable it up but it seems a shame to waste the elegance of the modular design. Power, metadata, start/stop, hero audio pass through back to the R3Ds, etc all cable-less and solidly mounted with no extra gack.

A man has to have a dream, right... ;-)

Cheers - #19

+ from me... great idea! it would be soooo coool if we'd start seeing modules that make possible less clutter around the camera

[Kevin Olsen]

I wonder if RED would have any interest in partnering with Sound Devices to create a version of the PIX 240 with an Epic module mount/interface?



Has RED said anything about opening up their system to outside development? I know that they are very approachable and willing to work with many vendors that add value to their system, but I imagine that RED has little interest in sustaining a codec their rivals utilize while attempting to push out their own.

Could we see modules developed by others that attach to RED and interface with their system? That's the million dollar question.

KO

[Alexander Ibrahim]

This was my point. It's not an editing codec, it's just not optimized for editing. This is why Prores exists.

Well we disagree there.

The only reason these higher end codecs aren't "optimized for editing" is the processing power they require.

They are in every other regard editing ready just like ProRes and DNxHD.

In fact ... you could make the same argument regarding DV. It wasn't "edit friendly" when it debuted in 1998. For years computers couldn't process it in real time without special hardware.

Same for DVCPRO HD, and RedCode ... which still isn't "edit friendly"

In five years standard computers will be able to handle these codecs at 5K 3D 48fps in realtime without special hardware.

My point being that an editing codec has technical features other than "it runs fast on the computer with my NLE software installed."

Intraframe coding and resisting generational loss in post are two important features ... h.264 offers them.

Now, I realize I am using a technical definition of an editing codec ... some here are using a business definition, as in "what the post house requires us to deliver in."

[Alexander Ibrahim]

Why is this so hard for the Apple faithful to grasp? Is it Final Cut myopia preventing them from seeing the woods for the trees?

As a user of both Final Cut Pro 6/7 and Premiere CS5.5 can I just say Premiere Pro sucks.

I actually learned NLE with Premiere ... so its easy to get acclimated to how Premiere does things again. Mostly this reminds me of why I left in the first place.

That said, there are a lot of features that seem so much like they are just check boxes without real thought given to their usability. For example the scopes in Premiere are awful. They blast you with color, update slowly, and never show the full data. Compare that to the scopes in Color, FCP X, or FCP 6. I can set those to monochrome and they are fast. You lose resolution on FCP 6/7, but FCP X shows full data all the time.
(Well, Color is slow but accurate.)

Also who on earth thought that including i/o hardware preferences in SEQUENCES was a smart idea? I can have a perfectly good edit in a Premiere timeline and be unable to use it on another Premiere system because they have different i/o hardware. That should definitely be a system configuration setting, not chosen on a per timeline basis! You know, like FCP, Avid, Resolve, Color etc.

Premiere Pro kicks some ass. It has many technological features that lead many to say Premiere Pro CS 5.5 is the FCP 8 we never got. There are some stupid decisions though that make it unpleasant to use, and sometimes make it unusable.

In my current projects these issues are a nuisance, not a showstopper. I don't even concern myself with scopes in Premiere, I just wait until I am in Resolve. Eventually I'll add a BMD Ultrascope, but for now I'll just ignore it.

If FCP X had XML import and export, I'd rather just use it, even with all its other issues.

Why did I get Premiere Pro? I didn't. I got Photoshop and Illustrator. Premiere just happens to have been in the box and useful.


I really don't get it....

And you won't until you see that there are other ways of working.

I am editing a music video for a friend ... and its a project that will live entirely in Premiere until its time to export to Resolve for grading. (Probably via EDL given the XML problems Premiere is having with Resolve right now.) I'll probably do some limited effects ... but I'll send those out to Shake/Motion from Resolve

Its not complete crap - its become an important tool for my in my studio.

Its also a tool I long to replace.

[Max M.]

+1 for the pix240 as a epic module, that would be a dream module :)

[D Fuller]

Well we disagree there.

The only reason these higher end codecs aren't "optimized for editing" is the processing power they require.

...

In five years standard computers will be able to handle these codecs at 5K 3D 48fps in realtime without special hardware.



Not meaning to question your wisdom, but isn't that an awfully long time to wait to get your editing done? ;-)