Let’s say you wanted to grade some footage coming out of a DSLR camera in Natron, what tools would you use? What would be your workflow? Colorspace? Graph setup? I know Natron isn’t specifically designed for grading but the power of node compositing and separation of channels offer an endless number of combinations that I believe can allow sophisticated grading . Natron is a relatively new software and I’m assuming in the future we may see the addition of tools such as an RGB parade, luma waveform scope, 3 way color correction, false color node, vectorscope, embedded blending mode in nodes, etc…that will make grading even easier. So much can already be done with a simple colorlookup node well used. Let’s share our thoughts on grading in Natron? Now and in the future?
You can start by using a grade node that is very powerful to make black/white balance and main color adjustment.
After that it depends on how far you want to go with grading, you can use mask or keyer to isolate parts and do secondary color adjustments. In fact as you say there are many possible combinations.
For grading IMO tools are there. Compared to a dedicated application what is missing is being able to go from shot to shot to make them match. And also some scopes. I’ve made a plugin that do a luma waveform that you can get here :
I don’t know how exactly this is supposed to work so it may not be perfect, Also I’ve tried to make a vectorscope but it was too computer intensive with my poor coding skills.
it would be really fine, if we could realize something like LUE for nuke as a natron addon. this would be really helpful, not least because LUE doesn’t work faultlessly in nuke non-commercial, because of some python restrictions of the free edition.
but i’m in doubt, if it really makes sense to utilize natron for color correction?
i also think, it’s very important, to have a lot of context information and usefull opportunities to visible compare sets of shots while grading. that’ very well implemented in davinci resolves gallery view IMHO. for a strict single shot centered application, which doesn’t have any more powerful timeline concept up to now, this looks quite unfeasible. but on the other hand, i also would really like to see this kind of high quality color related processing in the free software world. maybe natron could be used in the meanwhile just as kind of plugin for other existing free NLEs for this particular purpose…
yes – i also did some experiments, to archive this goal, but i never find time to polish the code and refine it for useful sharing. but at least as a stimulating advice: take a look at the really fast and hardware neutral acceleration capabilities for histogramm calculations, color space conversions, and reorder trickery in ArrayFire for this purpose.
I also start with a grade node for color correction then adjust specific parts of the image using masks then use a colorlookup for color grading. If the shot requires denoising, I denoise before color correction and color grading.
Yes, 3 way color correction is so much more intuitive.
Resolve? It’s node based and designed for grading
Since you are coming out of the DLSR camera, I would suggest that transcode the file(s) to an intermediate format first. Something like an image sequence format like PNG, DPX, TIFF or TGA. That would be contigent upon your hard drives if they can perform realtime playback/write simutaneusly without bottlenecks.
You can then pipe a Grade Node, then Color Correction node and then ColorLookup node. The attached image gives a very basic workflow but extremely time comsuming in the long run.
I agree with ELDirector about going to Resolve and if you can transcode the DLSR video clips to an intermediate image sequence file format.
Thanks. I know Resolve is great for grading but as a pro open source person, I always give it a try on open source softwares first. If none fit my needs, I buy a licence of the appropriate software for the job. Natron has the potential of being a great grading software, it isn’t missing that much.
Then in future maybe this could be very interesting for you.
If you use a Canon DSLR you are in luck, because they work together with A1ex from Magic Lantern. So good compatibility with ML RAW footage should be granted.
It is not finished by now but there is a Github to test.
Thanks. I shoot 4K on a 5D mark 4 and it will take a while before Magic Lantern enables 4K raw on the 5D4, if they ever manage to do so. A1ex isn’t seriously working on the 5D4 yet. I bought the 5D4 knowing of it’s lack of log and some of it’s restrictions but I actually don’t regret it (incredible autofocus, 500/Mbps bitrate, etc…).
Yeah I bet its a great camera.
I see now my post was a little mixed with topics. The main thing I wanted to say was the Opencine part
With the hope they are getting somewhere useable in the near future.
If someone where able to port this simple Nuke script into Natron then it would be possible to import an EDL (edit list) from ex. ShotCut, Flowblade or any professional NLE. That way we could import each clip as individual nodes in a string of nodes and it would be possible to grade a video sequence in Natron.
(The script is pretty barebones, so a few things have to be added for more succesfull imports)
Importing edit lists would be great. Unfortunately, I wouldn’t know how to port the script but someone here might.
Is it possible to import LUTs (.cube) in Natron?
Natron can do a quite of few things that may seem hidden, but if you look in the Color category, you will see the OCIO sub folder with other OCIO nodes. The node you want is called OCIOFileTransform Node. See attached image for all the LUT extentions that you can import.
Many thanks Omar. I did a simple grade in Resolve, exported a 3D LUT and imported it in Natron. It woks but colors are very different. Probably a color space setting? As well as gamma?
Can you post a screenshot of your Natron settings?
@Christopher there are many things to consider.
First and most important one when it comes to color grading is to have properly calibrated and profiled devices and color management on your system in place use a color probe (like eye1, spyder, colormunky or others) and take the time to understand how it all works. Color management cannot be done just by eye and tweaking your monitor to the point where you think it looks OK… How else can you tell if the colors you see are the colors someone else will see in her/his monitor?
Then a few considerations: the user interface Resolve is not color managed by default, Resolve is designed to use external (and properly calibrated and profiled monitors). The image in the user interface can be displayed via custom LUTs to be able to work on a defined color space, usually BT 1886 (Look into https://hub.displaycal.net/wiki/3d-lut-creation-workflow-for-resolve/ ). To match the results on the screen with Natron (or any other app for that matter) you need to ensure that you have all apps set to a common set of parameters, and make sure you use the correct display transforms in the same color spaces.
i think, using ACES on both sides could work in more compatible manner.
I graded a short music video in Natron back in the early days. It didn’t go great, but it worked.