RED’s IPP2 Pipeline In Resolve (Sort Of)

August 5, 2017

RED's new IPP2 processing pipeline has a lot to offer, in this Insight, we explore the new pipeline setup in RedCineX & DaVinci Resolve 14


Trying Out IPP2 In DaVinci Resolve

Say what you want about RED, but they continue to push the envelope with what’s possible with digital cinema cameras.

Recently, RED announced that they had revised their entire processing pipeline now dubbed IPP2.  This new pipeline brings significant improvements to RED footage image quality.

As a Resolve colorist (using version 14), I wanted to see how IPP2 worked – and if it even would.

I discovered that in Resolve 14 (currently), IPP2 is mostly implemented but has one large crutch – and in this Insight, we’ll explore what IPP2 is, its benefits and how to (mostly) implement the workflow in Resolve.

Benefits

One of the hardest things to decipher when a company rolls out a new product, technology or workflow is what’s just marketing speak, and what’s a real-world improvement.

To RED’s credit, in the past week of really pushing on the IPP2 pipeline, I don’t think there is anything about REDs approach here that screams of the marketing team.  All of the stated improvements really do seem to be visible.

Here are a few:

  • Edge Gamut/Challenging Colors Handled Much Better – I’ve always found really saturated RED material to be challenging – think concert footage, police cars, etc.  This footage would always be represented by big blocks of color, edge aberration, etc.  This has been greatly improved in the IPP2 pipeline.
  • Highlight Roll Off – Highlight performance is one of those things that people obsess about and for good reason; the detail rendered in highlights communicates a lot about a shot.  Part of the IPP2 pipeline is drastically improved highlight roll off and greater control of how highlights are handled.
  • Better ISO performance – ISO performance has been drastically improved. On Helium based sensors, the result is a full stop brighter image vs. the previous sensors given the same ISO.
  • More Detail – IPP2 is not just about color – it’s about detail too.  The new pipeline improves debayering to achieve more detail on fine patterns like clothing.
  • Better Equipped For HDR – one of the focus points of IPP2 is HDR workflows.  By maximizing gamut and dynamic range, the IPP2 pipeline suits any HDR workflow.

Again, there is a tendency in this industry to use platitudes to describe sensor and overall camera technology, but in my testing, I’ve found that RED is not pushing any BS.  All of their stated improvements are visible, which ultimately means better images to work with as a colorist.

REDWideGamutRGB & Log3G10

RED color space and gamma settings have always been sort of like composite/transfer modes for me – you kind of know what they do, but you just choose the one that looks best.

Before IPP2, RED color science has always been a little confusing (at least to me).  While I’ve mostly chosen RED Color 4 or one of the Dragon options + REDLogFilm with IPP2, RED is looking to simplify things greatly.

The concept is to standardize on what they call REDWideGamutRGB – a large gamut color space optimized for their new Helium sensors (but works for any previous sensor) and Log3G10.

Here are the official blurbs (from RED) on each:

REDWideGamutRGB – This is a new color space that’s designed to encompass every color that is recordable by RED’s current and past cameras. REDWideGamutRGB simplifies many workflows because all subsequent color operations have a common starting point regardless of camera, look, or output.

Log3G10 – This is a new gamma curve that’s designed to precisely encode the full tonal range from RED’s current and past cameras.  The “3G” represents the mapping of 18% mid gray to 1/3 of the maximum possible code value, and the 10 represents the extent of the log curve encoding a linear light value up to 10 stops above mid gray.  It serves a similar function as REDlogFilm and the Cineon specification but allows for a greater dynamic range and HDR output standards.

 

Resolve 14 supports the REDWideGamutRGB and Log3G10 – cornerstones of the new IPP2 pipeline.

 

Essentially, RED is ditching (but still supporting) their previous color science cacophony in favor of a wide gamut color space and an EOTF that is optimized for SDR and HDR.

I love this. The hardest part of RED workflows has always been the plethora of options.  It’s great to see RED make their new pipeline simple.  Other RAW controls remain per previous SDK versions.

RED Provided IPP2 Output LUTs

Here’s the thing: Resolve 14 doesn’t yet totally support a full IPP2 pipeline.  Well, it does, but kinda…. Let me explain.

Resolve 14 supports REDWideGamutRGB (RWGRGB) and Log3G10, but it doesn’t support output mapping from RWGRGB/Log3G10 to the color space you’re working in i.e. 709/P3/2020.

Red Cine X Beta (Build 50 and Higher) supports an end to end IPP2 workflow and part of that is an output transform that allows you control contrast tone mapping and highlight roll off based on the IPP2 monitoring preferences you’ve setup including HDR.

 

RED CineX (Beta 50) includes support for the IPP2 pipeline including output transforms for tone mapping and highlight roll-off. Resolve currently only supports LUTS for this purpose. RED provides a bundle of 52 different ones with different combinations of settings, targeting different gammas like BT.1886, 2.6, etc.

 

IPP2 Monitoring in RED CineX (Beta 50 or later) controls how the output transform is handled.

 

As of the publishing of this Insight,  Resolve 14 doesn’t support REDs output transform in YRGB, RCM or ACES projects. While RED IPP2 is part of REDs 3rd party SDK, Blackmagic has not implemented the version with IPP2 procesing…yet.

Luckily, RED has provided 52 different 33x33x33 LUTs that can effectively be used for the same purpose – although keep in mind that the LUT approach is different than a mathematical transform approach (think Bitmap vs. Vector).  UPDATE: On 9/8/17 RED released an updated LUT download of 32 3D LUTs.  In addition to supporting Rec 709 as the original package did, the new package supports Rec. 2020. However, the LUTs are all using a BT. 1886 EOTF, unlike the power functions that were also available in the original distribution of the LUTs.  

 

The RED IPP2 REC 709 LUTs are 33 x 33 x 33 and meant to emulate the true output transforms of the IPP2 pipeline.

 

Practically speaking, the RED provided LUTs work well but they’re only mapped to REC 709 output.  I’m sure someone with some free time might try to extrapolate to wider gamuts like P3, but it’ll be interesting to see how Blackmagic handles the output transforms of IPP2 in future versions of Resolve.

Additional Resources

Over the past few years, I feel like RED has become really good at explaining their tech, making it approachable and understandable.  That’s no different with IPP2..

Here are a bunch of additional resources to help you understand what the IPP2 pipeline really means:

Video

In the video below, I’ll put the IPP2 workflow into action in Resolve.  Huge thanks to my buddy John Aldrich & Ben Masters/Logan Lewis from Fin & Fur Films for the Red Weapon (Helium) 8k Clip.  When the IPP2 pipeline is fully implemented in Resolve (if it is), I’ll do a follow up in another Insight.

As always if you have any questions or something to add to the conversation please use the comments below.

-Robbie

Member Content

Sorry... the rest of this content is for members only. You'll need to login or Join Now to continue (we hope you do!).

Need more information about our memberships? Click to learn more.

Membership options
Member Login

Are you using our app? For the best experience, please login using the app's launch screen


Comments

Homepage Forums RED’s IPP2 Pipeline In Resolve (Sort Of)


  • Willian Aleman
    Guest

    Hi Robbie,

    Thanks for the insight in RED IPP2.

    Since IPP2 is so new, it would be interesting to have a follow up to this insight regarding the workflow using IPP2 within ACES and Davinci Resolve.


  • Robbie Carman
    Guest

    Hey Willian –

    In an ACES (or RCM) workflow the RAW data is automatically decoded in to ACES space. so the IPP2 choices I show in the movie become irrelevant (from a user control point of view). Now, what I’m unsure of is if part of the overall IPP2 initiative at some point that the way that transform to ACES will be updated. The white paper I link to above shows Matrix Values from REDWideGamutRGB to 2065 APO. So I would assume that at some point that handling of RED RAW to ACES will be updated, but I’m not sure.


  • Willian Aleman
    Guest

    Hey Robbie,

    Thanks for the response. Your response aligns with a similar thread going at ACESCentral:
    http://acescentral.com/t/red-ipp2-color-workflows/1055


  • Margus Voll
    Member

    I have been working with Helium material in ACES and overall it works but sometimes it seems like we get some noise introduced to the material. Not that super much but it is there. Client says if they look the material in IPP2 they do not see it and after Resolve and ACES there is some and i have to NR it most of the times. Maybe you could take a look and evaluate what you see about that?

    I have not talked to Graeme about that but i was planning to take that up after i make some test’s around here. Also could be the thing that you point out pipeline is still not implemented in full detail. In theory it should not be an issue after IPP2 to ACES transform but still curious about it.


  • David Jahns
    Member

    is there a link to these RED provided LUTs somewhere? Searching Red.com for IPP2 LUT yields nothing.


  • Robbie Carman
    Guest

    Hey David –

    I think you just missed the link above (about half way through sentence starting with Luckily RED has provided 52…..) Here’s the link just as a URL https://www.red.com/downloads/58de98eff77f300ebe00053f

    Also worth noting they’ve updated the LUT package to also support Rec 2020 Transforms (updated 9/8)


  • Robbie Carman
    Guest

    Hey Max – Sorry I missed this!

    Indeed I have been seeing the same behavior. I’m thinking there is something off with the transform to ACES. I’ve written the BMD team but not heard back yet. Will keep you posted.


  • David Jahns
    Member

    Thanks, Robbie – not sure how I missed that!


  • Tyler Hawes
    Member

    14.1 added IPP2 support it seems? I’d love to see a new/updated insight about how it’s working now….


  • Robbie Carman
    Guest

    yep. It’s on my list for the next week or two to do an updated Insight with 14.1

  • Robbie,

    Basic question. What is the relationship of the OUTPUT LUT setting in the preference page of Davinci and where I place a 3D lut in the node order for output? So for example I have a raw clip imported and set up an an ipp2 profile in the settings of the 3D ouput in the Davinci presets. If I then set up the settings in raw camera to RedwidegammaRGB and LOG310, why then would I need to setup in preferences the OUTPUT LUT settings? Sorry for the confusing post but because of this I hesitate to rely on Davinci to grade. Simply put should I ignore the 3D lut output settings in preferences and just drop the ipp2 settings in camera raw and go from there?

    Thanks
    J

  • bump…I’m not seeing any movement on this front in Resolve 15 beta.

    I’m also struggling with implementing creative 3D LUTs that are used for on-set monitoring. My director likes to use cinematic LUTs to get a base grade on-set…so the DP adds it prior to the IPP2 output transform. If I want to use this as my starting point, or at least reference what the director was liking, I’m forced to hunt for the LUTs used since they do not appear anywhere in the camera metadata. Sometimes I can find them buried within R3D folders if there were on-set proxies churned out, but I’m still adding pre-IPP2 manually and dialing in the strength until it approximates what the DP saw on set. Embarassing.

    All in all, attempts to implement a reliable Resolve “starting point” is a very sloppy science…which sadly means a lot of professional colorists are shrugging and just grading with the on-set prores proxies to avoid the headache 🙁

  • Hello Robbie,
    will there be an update on the topic? Meanwhile in DVR 15.0. the people of BMD have integrated “Output Tone Map” and “Highlight Roll Off” in the Camera RAW tab. Other Resources on the Web say ColorSpace is to be set to Rec709 and BT. 1886 which yields the same results then applying the 33x33x33 LUT’s from RED after using RedWideGamut and Log3G10 in the Camera RawTab. Can you confirm that?


  • Mel M
    Guest

    I know this post is old, but I was just wondering about working with flattened IPP2 RedWideGamut/Log3G10 ProRes files (exported from Premiere) in Resolve, using REDs 709 output transform LUTs.

    I understand that the “proper” IPP2 workflow is to grade before the output transform, but I’m finding that it makes working with Qualifiers less than ideal, due to the fact that Resolve’s qualifiers don’t appear to be designed to work with log encoded footage, and makes selections very difficult.

    How do you deal with this? For the most part, I have little need to worry about having to transform a grade into different colorspaces (the majority of my work is for broadcast TVCs), so should I just treat the Output Transform LUT as a simple 709 conversion LUT, and not feel guilty for grading after the OT LUT?


  • Raphael W
    Guest

    I’m also struggling with the creative luts and following this thread(http://www.reduser.net/forum/showthread.php?174706-IPP2-Creative-LUT-don-t-load-Davinci-Resolve-Premiere-Profa) didn’t work out… Did you find a solution for this problem yet?

Log in to reply.

1,000+ Tutorials to Explore

Get full access to our entire library of over 1,100+ color tutorials for an entire week!


Start Your Test Drive!
Loading...