This is a very short post to cover a few points about using the new Red Pipeline footage (IPP2) in Premiere Pro CC. The approach here is one of many and i’m not sure if it’s even the best way to go about this, but right now it works and i’ll try to explain why.
(i also have tonnes of other blog posts about all sorts of things sort of waiting, just incredibly busy in post - once there’s a breathing space i hope this blog will be a lot more active.)
IPP2 is the new colour pipeline from Red. It was designed with the Helium sensor in mind but applies to other cameras too. The end result here is basically a really nice mapping from the wide gamut space that Red can capture into various destination colour gamuts of our various displays. The most common, and basic, target is still Rec709 or BT1886. When shooting with IPP2 the camera will add a LUT into the folder of the recorded R3D, or this LUT can be downloaded from various other sources. It’s effectively a display LUT.
Why is this conversion normally so difficult? Mapping from wide spaces into narrow spaces is just like squeezing a gallon into a pint. You’re going to loose stuff. But the way you map the various colours makes all the difference. It’s not a mathematical mapping but a creative one. Often bad mapping results in unbalanced colours, especially saturated ones and also issues with bright highlights and out of focus areas - LED car lights, street lights and so on. These are usually out of gamut for the very limited 709 pallette. This is the beauty of IPP2 and all of Graeme Nattress's work and research.
Now in Premiere this is how i approach this.
Firstly all R3D clips imported via the media browser so that segmented R3D files are joined together behind the scenes.
Then all those files are set to a colourspace of RWG (RedWideGamut) and a gamma of LogG10. This pretty much ensures that you have all that the camera saw in a desaturated looking flat image. The Red settings in Premiere are always applied to the Master Clip. This is important. Then still on the master clip i would apply a Lumetri colour effect and the only thing this has is the IPP2 LUT (as an Input LUT)
Being on the master clip, all the subsequent edits taken from that master will then share these settings and they will all just look right, this is the default mapping.
Then on each individual clip there can be an additional Lumetri effect, where creative work can happen on a per clip basis.
Ah you say but this is wrong. The IPP2 LUT is designed for adding at the end of the chain and technically you’re right. But the Lumetri controls in Premiere don’t really work with native log or flat footage, they are really designed to work with display referred footage - i.e. stuff that looks right going in. Pretty much all of Adobes products are based around manipulating display referred imagery, i hope one day this changes but right that that's what we're working with.
An example of this is that the exposure control in Lumetri is not a technically accurate exposure at all - it's neither a gain (for linear data) or addition (for log), it’s an S shape adjustment - most likely this way because Adobe recognise that the vast majority of footage people are working with has a 709 style gamma curve to it. So when you move the exposure down, they keep the brights higher because they know that being display referred there isn't a huge range of superbrights. It looks good basically. Now i have a whole host of issues with this approach but that's what we've got.
Here's an example of comparing adjusting exposure in Red vs just in Lumetri to see some of the differences. It's not possible to match exactly because as i mentioned Lumetri isn't just taking exposure down but doing a bit more. The values on the sliders mean nothing either, -1 doesn't mean one stop or anything.
It’s my understanding that internally PPro is all 709 anyway. And i’m not 100% sure what the bit depth is between the layers - on import it is 10 bit maximum so what happens with Red i cannot tell.
So my reasoning for the above approach is
In the Red settings i would adjust exposure overall for the whole clip. Generally a whole clip is shot at a certain exposure, different clips may need adjusting relative to each other and the best place to mess with exposure is in the R3D settings - that way you will get the absolute best base image out of the source. Don’t forget the Red footage is high bit depth, way more subtle than the possible import of PPro. When you make your exposure changes here you are using the full camera data and then passing on the subset range to PPro. If you make exposure adjustments in the Lumetri panel at the end of the chain you may miss or clip data relative to what you shot, as seen in the images above.
So the IPP2 Lumetri effect on the master clip handles the conversion into what PPro normally wants.
And you’re free to adjust each clips Lumetri settings to taste. All the creative looks expect 709 looking footage anyway, so those film LUTs and so on. Making small adjustments to exposure is fine. I'm not saying there are any huge world ending issues with changing exposure in this way but you will get better quality if you do it at source.
You can of course just add the IPP2 as an input LUT to each clip, but that’s a bit of a hassle, it’s easier to do it to the master clips once IMHO.
When i get a moment i will see if this is scriptable but you can always apply a batch operation in RedCine to set all the R3Ds to RWG/LogG10 before import.
This is how i see it right now at the beginning of 2017, subsequent versions of Premiere will no doubt change as Speedgrade slowly dissappears then hopefully some of the control that offered will continue in Premiere. But i'm always happen to hear alternative approaches.