Activity › Forums › Adobe Premiere Pro › PPro / Red / Decklink
-
PPro / Red / Decklink
Posted by Dean Decarlo on November 12, 2010 at 12:20 amHi, I’ve searched the forums but haven’t found a definitive answer to whether I can edit a Red (R3D) sequence and get output through my BMD decklink card. Seems the choices are to a.) edit with Premiere CS5 using Adobe’s Red setup which works fine but no BMD output or b.) edit using a BMD HD setup with monitor output but then fractional resolution doesn’t work so editing 4k files is very clunky. Is there something I’m missing? Playback resolution options are there but cannot be changed and seem to be ignored. Does BMD output only happen using the BMD setups? If so they should update to natively support the formats Premiere now supports. Right?
Thanks,
DeanTroy Murison replied 15 years, 5 months ago 5 Members · 10 Replies -
10 Replies
-
Tim Kolb
November 12, 2010 at 3:12 pmUnless you’re specifically referring to a dual link DVI port output on the BM device, it’s not nearly as easy as you envision…
RED files are RAW. They need to be decoded, and then de-bayered to create an image.
HDSDI is a fixed environment…Rec 709 color Y’CbCr, and 1920 x 1080 max res. The reason the desktop reduced resolution modes aren’t supported by I/O cards (and BM is not alone in this respect, AJA is the same) is that they really aren’t supported by the pipeline.
When you are in a BM sequence, you are in the proper color space and frame size for HD video. Bringing RED R2D files to that sequence creates a situation where you are decoding the RAW at 4K, then scaling it AND transcoding the color space. It’s a load.
My recommendation might be to shoot RED at the “HD 4K” framesize (3840×2160) which scales at half to -exactly- 1920×1080, which looks beautiful… Use the capabilities in PPro to take advantage of the RAW workflow for color correction, basic cutting, etc… Then export an HD master clip that you can use for output via your I/O card, and also do your final CC pass, etc. with your external scopes and monitor.
Unfortunately “Digital Cinema” cameras that shoot RAW are not “video” cameras…and their optimal post workflows are not based on “video” as much as we all seem to keep lumping all these cameras and production workflows together…
TimK,
Director, Consultant
Kolb Productions, -
Dean Decarlo
November 12, 2010 at 4:55 pmMany thanks for the reply. It’s pretty much as I thought. Not crazy about the idea of using one workflow to begin with and one to be able to use the scopes, monitors etc. Maybe I’ll just use HDMI output direct and bypass the Decklink altogether.
-
Dave Helmly
November 13, 2010 at 12:25 pmTim, as usual you’re right on target – thanks.
I don’t get to post as much as I’d like but saw this thread and thought I’d share what I’ve been testing in the Adobe lab. We’ve been testing an NVIDIA card with their SDI daughter card attached and it works excellent for all editing modes. The SDI port basically thinks it’s the 2nd DVI port and you use the same method of setting up the 2nd display ( Settings>Sequence Settings>Playback>1920×1080)
The great thing about this setup is that there are no other drivers but the standard NVIDIA drivers. You get all of the benefits from Mercury HW GPU mode.100% Adobe Native.
Recently, we’ve added the RED Rocket to the Mix to see if we can get FULL quality @ 2K and it works perfectly. 4K works at FULL but appears to drop a frame here and there and we are continuing to hunt that down. It may have something to do with the various SDI displays only supporting 1920×1080 and NVIDIA card combo.
Note* you can’t use the RED Rocket SDI ports as they are not designed for apps like Premiere or others to output to. They are specific to RED.
The only barrier for most will be the cost of the SDI daughter card – not cheap, but it works great.Setup is a bit weird as the control panel settings are not intuitive. Once you do it once, it’s easy.
I will say that for most people, using a DVI to HDMI or Displayport to HDMI cable attached to a $1,000.00+ flatscreen TV will give excellent results for the $$. This also uses 100% Adobe Native and GPU. I recommend this HDMI setup for most users. I recommend spending at least 1,000.00 for this type of display , the cheaper ones are quite soft and non-consistant color.
Lastley – The HP DreamColor display has a unique advantage over other Displayport monitors when using version 5.02 or higher (yes, we’re still at work on more goodness), it provides 10bit color to the DreamColor and once you see the quality , you’ll be hooked. This is a great confidence monitor setup for the $$. No expensive SDI (just an expensive but gorgeous display)
my 2 cents,
DKH
https://tv.adobe.com/show/davtechtable -
Pierre
November 14, 2010 at 2:20 pmDave.
Thanks for that great post!
I’ve got a pretty basic question I was hoping you could help me with.Does the Mercury Playback engine and the Red Rocket work in unison (do they compliment each other) ?
Or if one is working does the other one just sit idle ?I’ve currently got a Mac Pro with a Red Rocket installed and am seeing decent playback of R3D files in Adobe CS5 (5.0.2)… I am however using an ATI Radeon 4870 which apparently does NOT take advantage of the MPE (I need an Nvidia card for that). I’ve been thinking of installing an Nvidia GTX285 which is listed on the Adobe site a qualified CS5 card… but before I do this I want to make sure I’m going to see a significant performance boost.
Thank you!
-
Tim Kolb
November 14, 2010 at 8:31 pmMercury doesn’t change where video decode is done…it’s CPU-based.
If you have an approved CUDA card installed, you get the push on effects render.
So…with R3D files, most of us would utilize the interface for altering the R3D files as they are decoded…changing the metadata. This would be done in the decode of the frames and therefore would be done in your CPU/Red Rocket pipeline. If nothing else is done than cutting R3D frames and altering the color in the PPro R3D metadata interface (or RED CINE X for example)…it’s all handled during the RED decode.
If you added a color correction filter from PPro, say for some secondary correction (and it was an effect that is marked as being CUDA-supported…indicated by the little lego block with the arrow on it), then the CUDA card would pick that work up.
So…I guess the short answer is “it depends”. The CUDA cards have a definite impact on the speed of work on the timeline, as long as you’re not expecting it to have an impact where it isn’t designed to…
TimK,
Director, Consultant
Kolb Productions, -
Dave Helmly
November 16, 2010 at 12:13 amMichael,
I had a chance to discuss your question with one of our top engineers for the Mercury Playback Engine (CUDA HW) and here’s his response and I hope this sheds some light. I know there was just a new 10.6.5 graphics update and NVIDIA OSX drivers were updated to 3.2.17 for CUDA. This may also have an impact on your results (good lets hope).
————- From Engineering ——–
Hey Dave –
The RED Rocket and Mercury Playback Engine are designed to compliment each other with the rocket doing decodes & debayering and the Mercury Playback Engine handling rendering. Currently there are a few Mac specific issues combining the GPU with a rocket but on Windows it all works really well.We confident we’ll get the Mac issues fixed. I would caution against using the GeForce GTX 285 with RED footage, the standard 1GB configuration is not pleasant to work with RED footage. -
Pierre
November 16, 2010 at 12:45 amWow.
Dave thanks for getting that information. Straight from the horses mouth : )
I guess I’ll steer away from the GTX 285 but that begs the question, which graphics card will work best on OSX ?I guess Nvidia recently released the Quadro 4000 but is currently only listed as working with Windows machines on Adobes site. I’m assuming an update will come soon enough for OSX (hopefully).
Any thoughts on graphics cards?
kind regards,
MIchael -
Troy Murison
November 18, 2010 at 7:45 amHi Everyone,
Great info here!
I am looking into this as a solution for RED workflow and client monitoring and found all of this info really useful, I haven’t heard this kind of info out there yet.
We’re currently using BM Decklink cards for monitoring on MacPro’s and that’s really painful/unusable right now with RED footage. It works ‘okay’ with other workflows but even in a native BM sequence with native BM footage it’s just not like it should be or as it is in FCP with the same footage. To be fair, I’ve edited h.264 and other MPEG native using BM sequences which worked great and just isn’t possible in FCP, so there are some things working well with BM, it’s just slightly unpredictable feeling to me.
So I have a couple of Q’s for anyone re: the nVidia ‘way’: the Mac version of the 4000 doesn’t seem to have the SDI I/O daughter card support listed as available on nVidia’s website, so I guess we’d have to use HDMI or something as Dave suggested, right? Or do it all on a PC as that’s sounding a little more ‘stable’ with RED in particular, maybe?.
Dave (or anyone else for that matter), on the systems you’re working with, are you able to achieve any kind of remote deck control? Via Keyspan or any other way in PPro in conjunction with the nVidia cards? We are still in the HDSR world and will be for a long while I believe so that would be nice but not a show stopper.
I’d love to chat with anyone who has any experience with the nVidia digital video pipeline using their SDI card(s) especially after reading Dave’s post above! Sounds like bliss, especially with a RED Rocket along for the ride…
Thanks in advance,
Troy Murison
Seattle, WA
Reply to this Discussion! Login or Sign Up