Thread: RED + Nvidia = Faster 8K Playback and Editing

Reply to Thread
Page 3 of 5 FirstFirst 12345 LastLast
Results 21 to 30 of 42
  1. #21  
    Senior Member David J. Buchanan's Avatar
    Join Date
    Oct 2016
    Location
    Minnesota
    Posts
    371
    What the heck is 15 bit color? I thought it was 16 bit?
    BLACK in Minneapolis - Streaming!
    Corporate G&E Package
    Gemini, Zeiss Otus 28, 55, 85
    davidjbuchanan.com | wecofilms.com
    Reply With Quote  
     

  2. #22  
    Senior Member andrewhake's Avatar
    Join Date
    Oct 2009
    Posts
    667
    Quote Originally Posted by Misha Engel View Post
    As long as 8k.CineformRAW plays back in real-time on an $2.000 AMD system, where I need a $15,000 Phil Holland workstation to do the same with 8k.R3D, I will keep bringing it up.
    He brought up AMD again everyone take a drink..

    Whoopdie doo. The whole point is to not have to convert from acquisition format to playback format and maintain capture quality. Here we use cameras that capture R3D, not Cineform Raw. No fucking way am I going to spend time transcoding to cineform.
    Last edited by andrewhake; 10-19-2018 at 05:47 PM.
    Reply With Quote  
     

  3. #23  
    Senior Member
    Join Date
    Jun 2017
    Posts
    1,859
    Quote Originally Posted by andrewhake View Post
    He brought up AMD again everyone take a drink..

    Whoopdie doo. The whole point is to not have to convert from acquisition format to playback format and maintain capture quality. Here we use cameras that capture R3D, not Cineform Raw. No fucking way am I going to spend time transcoding to cineform.
    It's not about transcoding, it's about why one wavelet codec (.R3D) needs so much more resources than the other(CineformRAW).

    Since you have a $15,000 workstation you don't need to transcode.
    Reply With Quote  
     

  4. #24  
    Banned
    Join Date
    Dec 2012
    Location
    Los Angeles
    Posts
    828
    Quote Originally Posted by Misha Engel View Post
    It's not about transcoding, it's about why one wavelet codec (.R3D) needs so much more resources than the other(CineformRAW).
    for starters, .r3d files are encrypted
    Reply With Quote  
     

  5. #25  
    Senior Member
    Join Date
    Dec 2014
    Location
    Sarajevo
    Posts
    967
    Quote Originally Posted by Misha Engel View Post
    Both are wavelet codecs.
    https://github.com/gopro/cineform-sdk

    Wavelet codecs are the easiest to compress and decompress while maintaining the highest possible quality.

    For decompressing 5:1 8k.R3D you need a i9-9780xe(MSRP $2,000) running at around 4.5 GHz at all 18 cores (you have to overclock it from 3.4 to 4.5 GHz all-core).
    For decompressing 10:1 8k.CineformRAW you need a i7-7700k(MSRP $350) running at around 4.5 GHz at all 4 cores (you have to overclock it from 4.4 to 4.5 GHz all-core) tested.
    For decompressing 4:1 8k.CineformRAW you need a R7-2700(Newegg $250) running at around 4.2 GHz at all 8 cores (you have to overclock it from 3.5 to 4.2 GHz all-core).



    CineformRAW needs about 20% more space to have the same quality as R3D.
    Elegantly avoided answering my question...

    My point was that RED's got it's own, proprietary, codec. It's R3D. Meaning, 3rd party parts are not needed.
    Reply With Quote  
     

  6. #26  
    Quote Originally Posted by Phil Holland View Post
    Erm.... Call me the weirdo, but most of my work for several years is being delivered in 8K and just the concept of easier editing and general QC is a good thing for me on large and smaller 8K displays. To each their own. I'm just happy people are nice and comfy with 4K now as they "see it".
    Trail blazer perhaps. I just meant that with most shutter angles and shooting motion then 8K vs 4K (in terms of delivery) i think it's very very difficult to see. Perhaps in a static shot and also those Aerials you do - the difference is there for sure.

    cheers
    Paul
    Reply With Quote  
     

  7. #27  
    Senior Member
    Join Date
    Jun 2017
    Posts
    1,859
    Quote Originally Posted by Mike Nagel View Post
    for starters, .r3d files are encrypted
    Back in the days when RED invented .R3D it made sence to encrypt it (don't make the competion wiser).
    Since the days of DSMC1 all competitors know the RED trick so it doesn't make sence to keep .R3D top heavy on computer resources.

    Quote Originally Posted by Šabović Adis View Post
    Elegantly avoided answering my question...

    My point was that RED's got it's own, proprietary, codec. It's R3D. Meaning, 3rd party parts are not needed.
    3rd parties are needed when you don't have a mega workstation and it's good that RED implemented simultanious recording in ProRes(Apple) or DNxHR(AVID).
    You also have to supply the 3rd parties a SDK so they can implement it in their NLE's etc...
    And now the new 3rd party is NVidia.

    I'm fine with propietary codecs(as long as they supply a SDK's), just don't make them to compute intensive
    Reply With Quote  
     

  8. #28  
    Moderator Phil Holland's Avatar
    Join Date
    Apr 2007
    Location
    Los Angeles
    Posts
    11,943
    Quote Originally Posted by paulcurtis View Post
    Trail blazer perhaps. I just meant that with most shutter angles and shooting motion then 8K vs 4K (in terms of delivery) i think it's very very difficult to see. Perhaps in a static shot and also those Aerials you do - the difference is there for sure.
    That's the fun stuff as I find resolution benefits even motion blur. About the 1/3 of my work is aerials, 1/3 VFX related, and 1/3 (often with VFX related stuff) is moving things. Though I literally might be filming some rocks soon. Not joking.

    It's not difficult to see on an 8K display for me. In fact, since I've been doing more 12K and 16K array work downsampled to 8K this year, it's sort of a jaw dropping concept to downsample to that degree.

    i'm very curious what directions every camera company is going to go from here honestly. I know two companies other than RED have 8K bodies coming in 2019 and that's in turn going to put the focus back on what RED will do next.

    I have some theories, but I'll save that for off the air.


    The industry is so dynamic these days. 4K is the sort of obvious finishing target for many, 2K and below is still happening in a whole bunch of markets, especially news related. 8K is happening in very few, yet rapidly growing arenas much like where 4K is now as well as where it will be in 2-3 years.

    Outside of the industry trends I know where my personal interests land in what I'm trying to produce with my visual content. Much of that is 4K, 8K, and beyond.


    There are a lot of proprietary things found still within REDCODE RAW and inside the cameras themselves. I think everybody noticed when Helium dropped that processing was a smidge slower, DSMC2 brought some rather major image enhancements that aren't so clear, but you could see part of that story in the DSMC Dragon to DSMC2 Dragon comparisons. It's been pushed further now with the latest Monstro, Helium, and Gemini tech. There are reasons for all of this.

    I get the need for speed, RED's obviously always looking for ways to help out on that front, but the general mindset of faster isn't always necessarily better comes into play. Similar to those who didn't run the math on ProRes RAW, which is a powerful new codec for sure, but again REDCODE RAW is a bit more.

    Even stranger, with some of the newer processing stuff that has arrived and is coming it might allow for revisiting some concepts on raw encoding further down the line. But that's a battle for another day.

    Bleeding edge is often a hard path when also being tied to 5 year old + workstations and GPUs. But that's the varied landscape for sure.


    Just to echo some good advice, if you are building a new system in 2018 or 2019, don't even consider GPUs with less that 8GB if you plan on working in 8K. This is more in relationship to application memory management. You can get away with less, but this is one situation where less isn't more.
    Phil Holland - Cinematographer - Los Angeles
    ________________________________
    phfx.com IMDB
    PHFX | tools

    2X RED Monstro 8K VV Bodies, 1X RED Komodo, and a lot of things to use with them.

    Data Sheets and Notes:
    Red Weapon/DSMC2
    Red Dragon
    Reply With Quote  
     

  9. #29  
    Senior Member
    Join Date
    Dec 2014
    Location
    Sarajevo
    Posts
    967
    Quote Originally Posted by Misha Engel View Post
    (...)
    3rd parties are needed when you don't have a mega workstation and it's good that RED implemented simultanious recording in ProRes(Apple) or DNxHR(AVID).
    You also have to supply the 3rd parties a SDK so they can implement it in their NLE's etc...
    And now the new 3rd party is NVidia.

    I'm fine with propietary codecs(as long as they supply a SDK's), just don't make them to compute intensive
    Well, there's your problem...why not invest in an adequate system too instead of being only halfways there? You're not using RED Cinema cameras for YouTube vloging, are you? News Business?
    I mean, R3D, as such, is more than enough for what RED is trying to achieve with it (shortage appears on the other end, not on RED's), and there always will be RCXP, so you can transcode till the cows come home.
    Reply With Quote  
     

  10. #30  
    According to this web article, RTX Nvidia 2080 Ti 11GB provides RED 8K Full Res Premium debayer in real-time (25fps).

    https://www.pugetsystems.com/labs/ar...formance-1238/
    Reply With Quote  
     

Posting Permissions
  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts